APEX/AltiumLive DFM PROF DEV: Design WITH Manufacturing
Carl Schattke and Julie Ellis present how designing with manufacturing make the entire team from concept to fabrication a winning one, live from the IPC APEX Expo in San Diego.
Keynote Highlights:
- Interplay between the sides of design and manufacturing
- Insight into different teams involved in parts of the design process
- Overview of different leadership roles within teams
- Importance of panel drawings and when they are necessary
- Laser drilling as a way to reduce costs
Additional Resources:
Transcript:
Carl Schattke:
Welcome to IPC APEX EXPO 2022. I'm Carl Schattke, this is my co-presenter Julie Ellis. We're thankful that you're here. To those of you online, we're thankful that you're watching. We hope our presentation today will be of high value to you. The concept is design with manufacturing, and we're going to talk about building a winning team.
The concept today, we're going to present it like it's a game, and we're going to go play a game today. We have an offense, it is the designs, and then we have a defense that manufactures, and we also have special teams that perform other tasks. And our mission is to win for our fans and for our owners and for ourselves. So if you choose to participate with us today, you will learn about these teams, and we'll look at the players and the playbooks and the support staff that make winning possible, and goals are scored when each product makes it to a customer.
Your coaches and coverage team today are Julie and I, and we've helped get millions of products into the hands of fans. And our mission today, should you choose to accept it, is to learn from our playbook how this game works and the best plays we have found for each of the teams. And as I said before, we have this offense that designs, we have a defense that manufactures, and special teams performing other tasks. And our greatest wins come from outstanding cooperation. Our mission is to put great products in the hands of our fans, and we have many players and a large pool of talent that supports us in doing that, and every person on our team plays a pivotal, valuable role in this success that we're looking for. But we are going to look today at the roles and responsibilities of the players and the support staff on our team, so that you can learn how to be a great team player with us.
We're going to talk about game day. So winning at the design and manufacturing game is not a matter of chance, but one of preparation and execution. Your coverage team, Julie and I, will explain the game and keep you posted on what must be done to win. Not every play wins, but no team wins without play. When you get back to your home team, we hope you can steal the good plays from our playbook and score goals with them. We will look at the roles and responsibilities of the players and support staff on our team, so that you can learn how to be a great team player with us as well.
Now, the offensive team, the purpose is to conceive and explore and design products. So they get extra points when the products do something better or in a more useful way than what the competition's doing. And this team supports wins with the defensive team by staying on the field and minimizing the time the defense is on the field. They must produce elegant designs, outstanding documentation, and do it quickly. The rules for this team come from many places; IPC standards, FCC, communications standards, UL standards, CSA, signal integrity standards, design for manufacturing, design for tests, design for assembly, design for reuse, design for rework. We will examine how the game is played and the rules are met. Now, the defensive team. Would you like to present this one?
Julie Ellis:
No.
Carl Schattke:
Okay.
Julie Ellis:
You're doing great. But how many people here would consider yourselves on the offensive team? Can you raise your hands? So we've got one, two, three, four, one, two, three. Okay.
Carl Schattke:
Okay. And how many would be on the defensive team? Okay. About 50-50. How many would be on special teams? Okay. We've got people represented in our audience from all of the different teams that we're talking about. So the defensive team, the purpose is to create and build reliable products at the lowest cost. Extra points are given when the products do last longer and perform better and continue to improve the production metrics.
This team wins by supporting the offensive team with great field support. This team wins by helping the offense make good plays, and giving them the best feedback on what works and what does not work well. This team is an invaluable partner to keeping the offense sharp. The rules for this team come from various places; ITAR, MILSPEC, ISO9001, NIST, RoHS, REACH, UL, OSHA, EPA, EHS; all of these different areas have rules for this team. What this team must do is produce products on time, that are reliable, cost controlled, whether it's for prototypes or mass production.
Julie Ellis:
So can you tell me, how many people-
Carl Schattke:
You're going to have to speak into the...
Julie Ellis:
Oh, sorry. So how many of you are working more on prototype development as opposed to volume manufacturing?
Carl Schattke:
Prototype? Okay, a lot.
Julie Ellis:
Yes, a lot. And then-
Carl Schattke:
Volume?
Julie Ellis:
Volume?
Carl Schattke:
A few. Okay.
Julie Ellis:
And so for those of you who are doing the prototype development, are you trying to design for any sort of production, or are you just designing to prototype standards? Volume?
Carl Schattke:
It's prototype with the design-
Julie Ellis:
Okay, with the intent of migrating to something that could be manufactured repeatably.
Carl Schattke:
Yeah.
Julie Ellis:
Okay, good, because that's what-
Carl Schattke:
We're going to definitely spend time talking about that. And while you design for small volume, it's a different kind of flow and process than for high volume, but if you design for high volume it'll definitely work for a low volume production. But we'll go over some of the distinctions to that as we get into the plays that we run.
And then we have our special teams squad. The purpose is to excel in areas that are not specific to the offense and defense. Extra points are given when design and manufacturing problems are mitigated. This team supports wins by testing, coordinating, interfacing, sourcing, supplying, and doing the business things that the offense and defense need. This team is not on the field as often, but it can impact the outcome of the game in a great way. This team is an invaluable partner to keeping the offense and defense ready, and this teams looks to exploit and perform accurately and quickly. Mistakes by all the teams are costly, but special teams' mistakes can change the outcome of the game quickly.
Then we have our management teams. The purpose is to provide the resources and capital to bring the best technology, players and coaches, and support staff and organizations, to make the best experience for the fans. This team sets the culture for the other teams. This team hires and fires all the other teams. This team helps the other teams get what they need to do a great job. This team coordinates the commercial, legal, human resources, and the facility activity. This team makes all of the other teams better, or could be worse. Extra points are given when star players and top tier suppliers are recruited. When healthy, this team supports all the other teams in being effective. This team is measured in the marketplace; profit and loss, cashflow, stock price, P&L, brand, news. Those are all the measurements for management teams. Management, while not on the field, makes the hard choices that cause wins and losses.
And we have our coaches. So coaches are in charge of the personnel decisions for each play. They're responsible for play calling and clock management. They call timeouts when needed, and they align everyone on the goals. The duty of the coach is to communicate the mission and the goals to team and fans. They're responsible for team preparation. They take the blame for losses and they give the credit to the players for the wins. The head coach has a staff that also coaches. And great players coach and inspire other players. So players can be coaches too.
The offensive player introductions. All right, so now we're going to go into the different teams and who's on them and what they do. So we're going to go through the different roles on the offensive team, we're going to start with system engineers. What do they do? Well, they design requirements to meet the target functionality. What do they need? Well, they need to get information about what would make a successful product and production. Who do they get the information from? Well, they would get that from management or customer feedback, surveys, or user experience engineers. Who gets the information from them? The electrical and mechanical engineers. Then what are their deliverables? Well, they're going to deliver text documents of functional requirements and concept drawings.
So mechanical engineers, what do mechanical engineers do? They figure out how big or small something is going to be, they're going to develop the packaging, they're going to develop the way that things come together in a mechanical sense. They need 3D CAD tools to work on what they're doing. They need information from management and engineering on what the functionality is that they need to support, and then they need to be able to create effective packaging for that product and also effective assembly sequences.
So where do they get their information from? It's going to be from electrical engineers, it might be from a program manager, a manager; they're going to get information from a lot of different sources that needs to be put into the product in a lot of different needs. So they take information from a lot of different sources and then make sure that it can be built, and that you can build it reliably and cost effectively. They have a large impact on the results, because they specify the parts that are going to be chosen in building that.
So who gets the information from them? Well, they would be giving a part outline drawing to a PCB designer; they would be giving drawings for fabrication to the fabricators; there would be a bill of materials that they would put which would be going to the purchasing department at a company so they know what to buy to build the thing. So they're going to specify build materials that would be coming out of that. So those would be the deliverables mechanical engineers.
Julie Ellis:
So do we have any mechanical engineers here today?
Carl Schattke:
Pretty sure we do, I was talking to one earlier.
Julie Ellis:
Nobody's raising their hand.
Carl Schattke:
Okay.
Julie Ellis:
Because... Carl, can I interrupt?
Carl Schattke:
Go ahead, please do.
Julie Ellis:
Okay, so I always come from the assembly fabrication side, and one thing that makes it really difficult in fabrication to build a cost effective product is when the designs of the printed circuit board were designed way upstream and it's an oddball size, shaped board that doesn't fit well on a standard fabrication panel, like an 18 by 24 inch panel. So your typical maximum length that you can get on a printed circuit board is approximately 22 inches. So just keep that in your mind, that a board can't be bigger than 16 by 22 inches.
And then if you want to divide that, that's the size that we have on a standard fabrication panel. And any time you violate that, so you can't get like two 11 inch sides on your 22 inch side, or you can't get two eight inch sides on your 16 inch side, you're really going to kill your material utilization, and in manufacturing fabrication that's going to cost a lot of money in your circuit board. So as far as circuit boards, once we have the technology set, the stack up in technology, lights and spaces, fitting it well on a fabrication panel at your circuit board manufacturer is your lowest hanging fruit for optimizing cost.
Carl Schattke:
Thank you, Julie. So the mechanical engineers are going to be saying the cost of a product by the size of the circuit board that they decide to design. And by effectively using the most area on that panel... If two boards fit and that panel is $100, each board would be $50. If you can change a few different variables and get six boards on that panel, well, now those boards are $16 and they're roughly 33% of the cost of the previous product, just by making a small change. And if the designers can work with that smaller space, it can save a lot of money. So that's definitely a way to create a winning team product.
So the electrical engineers, what do they do? Well, they select the components and the logic to meet functionality requirements. What they need is solid requirements, from either a system engineer or management or the end user or component suppliers. They gather resources and information from a wide variety of parts and places, and then they use that to come up with a concept for the solution that this product is going to need. So who gets the information from them? Well, the mechanical engineers are going to get information about how much room these different parts that they chose take; the PCB designers are going to get information from them about the components that were chosen and how they're connected; and the program managers are going to get information from them regarding schedule and how much work is involved, the scope of the project the cost of the project and things like that; and then manufacturing's going to get information from them as far as the build material and other aspects of the design, might be firmware or something like that, and then they might do a test plan, things like that.
So their deliverables are schematics to the PCB designer; build material to the build designer and also the end manufacturer; signal integrity requirements, they're going to put the rules together for all the signal integrity aspects across how close things can be together, how far apart they need to be, whether it's based on voltage or current. And then they're going to have to consider all of those things, timing requirements, how much delay, how much of a budget is there for timing delay, things like that, depending on whether it's a high speed board or not. And then also they would typically be designing a firmware that would make the initial board boot up and work and be functional. And then they would also be responsible for the test plans to bring that board up. So electrical engineers have a large responsibility, it's one of the hardest positions there is on a team. They would be like the quarterback of the offense, is the electrical engineering.
PCB designers, what do they do? Well, they take the parts, they figure out how much room there is, they put them in there in the right place, they make sure everything's where it's supposed to be physically. Electrically, they make sure that everything will work, that all the design rules are met electrically, that it's going to be fully functional as designed. And then they make sure that it can be manufactured in a profitable way. They have to respect all three of these different areas and make sure that it fits and it can be connected and solved from a placement and routing standpoint. They have to make sure that it meets the mechanical requirements and everything is placed in a way that it's not going to inhibit operation or assembly or manufacturing. And then they need to make sure that it's going to be effective from a manufacturing standpoint, you want to make money with that board.
So what do they need? They need a schematic, a build material. They need a good, hard outline drawing. They need to know the environment that that product's going to be. They need to know the volume of production, so they would design differently if it's a prototype versus high volume, perhaps. They need to know whether it's going to have ICT pads on it. There's a lot of information that they need to know; how much current is involved on the board, what's the voltage? Are we concerned about high current, are we concerned about high voltage, or are we concerned about high speed? Or are we concerned about any other factors that could come into play based on that?
And then who do they get that information from? It's typically going to be from the electrical engineer and the mechanical engineer, possibly management. And then for the details about manufacturing that product, they're going to need to go to the manufacturer to make sure that they can actually build what it is that they're thinking about building. So they're also working with all of the other teams, and PCB teams are typically at the same level, as far as interface with the number of people, probably with the electrical engineers. They're working with a lot of various vendors and there's a lot of information that they need to learn in order to be able to do the job properly.
So what are their deliverables? Output files that are going to include schematic drawings and assembly package; there's going to be a fabrication package, so the assembly vendor needs one set of data, the fabrication vendor. When I speak fabrication, it's going to be the board vendor. And then assembly would be who's going to put the components on it. So they're putting these two packages together, all the documents that would be with that. So for assembly, they're going to have an assembly drawing, a schematic, pick and place files, those kind of things that are going to allow the automation of that product to be built. ICT test point locations and any special requirements for impedance, or TDR measurement for timing. All the things that are going to be needed to make that board effectively. That's on the assembly side.
On the fabrication side, they're going to output the physical board. That could be in a Gerber, an ODB++, or an IPC type of format. And then they're going to give out drill information, there'll be files for that, there'll be a drill drawing that has measurements on it that need to be checked on incoming inspection. There'll be detailed notes as far as plating and finishing, cleanliness, and all of the various aspects of that board design. Typically a board might have anywhere from 20 to 30 different specifications for all of the things that would be required in order to build that board in a way that's going to produce an effective product, that it's not going to fail, something that's going to meet the needs. So those are their deliverables.
And to the audience, if you have questions, please feel free to raise your hand and either you can step up to the microphone to ask your question, or you can stay in your seat and I'll repeat the question. So if we say something and you want to add or get some further clarity, please feel free to raise your hand. So we have a question, go ahead.
Julie:
Apologize, I came in late, I can't find the slides. Is there slides available so I can-
Carl Schattke:
The slides aren't posted yet, but they will be later after the presentation. So the PCB librarians, what do they do? Well, when a board is designed, each component has a certain area that it takes up and it has certain information that goes with that component. So they're going to choose the land pattern in order to create a reliable solder joint, whether that's as small as possible or nominal or a large footprint for a high vibration environment. So they're also going to figure out how the solder mask works on that, whether or not it has a thermal pad on it or something like that, they're going to figure out the exact geometry that's going to be giving the PCB designer a reliable footprint that will then be transferred to the manufacturers. So it's very critical that they have a high degree of accuracy in the work that they do, and they typically are very busy because the teams need dozens or hundreds of parts on a new design. They are responsible for keeping all of that information.
So they will design a footprint that'll have a courtyard, that's how much room that part takes. They're going to design a part that's going to have an assembly layer that'll show exactly how much room that part takes. They'll import a 3D model, so that when the board is designed that 3D model can then be exported to the mechanical engineer, and then he'll be able to check that that fits in the box and there's no collisions or anything like that. And then they'll also put any special information on it, like if it's a connector that has to be near the edge they'll put a line where the edge is in the part. So building that footprint accurately and to company standards is going to be a huge, huge part of building a reliable product, because if you start with bad blocks, your building will fail. If they make a mistake, it could literally ruin a board, or it could-
Carl Schattke:
... it could literally ruin a board or it could ruin tens of thousands or hundreds of thousands of dollars worth of product. So it's super important that the librarian's accurate and checks their work carefully before they release the part. They need to get accurate data sheets and accurate parts from the industry. Unfortunately for them, data sheets are often not right and manufacturers have been known to make errors on their data sheets, so they have to be very careful about getting the right information. And then they also need to read that information in a way that is accurate and not with a mistake. So they need to ferret out ambiguity like a pitbull on a meal, like they have to really get good information. That's really critical for them to get the right information on the parts.
And they have to make sure that the part number that they're building is exactly the right part. So a lot of times data sheets will have hundreds of parts on one data sheet, and you have to decipher what the right part is out of that. They need to be kind of magical at interpretation of drawings. All right, so where do they get their information from? Well, data sheets. Customer ... like supplier vendors will have data sheets for them. And if it's a custom made part for the company then they need to get that from the mechanical engineer that would be designing that part. So often times parts that would be custom would be like magnetics, because those are going to be a custom thing for an application, and they're going to need to get the exact size of that from the magnetic suppliers.
It's going to be a lot ... sometimes parts are one of, and they would need to get a part like that from that custom part builder. What are their deliverables? A PCB library that can be used by the CAD team. They would also be responsible for the database that would serve those parts to the design community that they serve. So that's the role of the PCB librarians. Very important function, you need people that are really, really accurate and attuned to detail to do that kind of work, and it really helps if they understand the electrical and mechanical aspects of that.
Julie:
So Carl,
Carl Schattke:
Go ahead. Speak into the mic, please.
Julie:
Well, as we're going through these positions, I keep thinking of things to go wrong because I work a lot in the quality problems. As Carl is talking to the positions, do you want me to give you heads up on things that I see that come back in the field, or do you want me to discuss those things as we go, or finish through the game plan?
Phil Nichol:
I think you guys are probably more familiar with the presentation. You probably just plugged in where it makes sense.
Carl Schattke:
All right. So it sounds like they want us to just ... if you see something that makes sense to fill in, fill in.
Julie:
Okay, yeah.
Carl Schattke:
If you want to add something, then feel free to just interrupt me, okay?
Julie:
You can stop problems from happening, So one of the things that I see, like I just ... a contract manufacturer and one of our customers, we delivered a lot of boards, and the library wasn't right, so they went through assembly and found out that the footprint on the device wasn't correct.
Carl Schattke:
Pretty common problem.
Julie:
So that's a real common problem. And then another common problem in assembly is if it's not in the library, that it's some sort of water sensitive or not hermetically sealed package that needs to be assembled after a normal process or after a wash cycle or something. So those kind of things, like crystal and oscillators, different kinds of parts-
Carl Schattke:
Some parts should not be washed. If they get an aqueous solution in them, they're going to fail later. Other parts can be washed. So it's important to know the assembly and the mechanical engineering. The PCB designer need to make sure that's accurately communicated, otherwise you could have manufacturing problems stemming from that. The other thing with regards to footprints, so how do you make sure that these footprints are right? Well, you design the board and you have all your parts placed, and then you take that database and you ship it to a third party that will load it into a tool that would check against its database for that same part, check those footprints to make sure that they're accurate.
So that's one way to verify that the librarians did a good job.
Phil Nichol:
Can I make one? You're-
Carl Schattke:
Go ahead.
Phil Nichol:
Yeah, your point about the non-hermetic parts and not being able to be cleaned is excellent. A lot of people don't think of that. So what we did as a provision in our library ... I'm sorry, go ahead.
Carl Schattke:
I don't know if-
Julie:
Oh yes.
Carl Schattke:
... it will help if you step up to the mic, get pretty close to it, that would be great.
Phil Nichol:
I apologize, yeah.
Carl Schattke:
Then the people online can hear well.
Phil Nichol:
All right. I'm sorry, I-
Carl Schattke:
So please repeat the question.
Phil Nichol:
Yeah. I'm Phil [Nicol 00:36:00] with L3Harris Technologies. Your point about non-hermetic parts not being able to go through cleaners is excellent, and it's one of those things that isn't always thought about. One of the things we do with our library to catch that once we identified a part that had those unique requirements, we removed the pace layer from that particular footprint so that it can't physically be put on a SMT. Then one of the other advantages of having a footprint library is you can put notes in the library that the PCB designers can see, so we would add a note that says, "Can't be cleaned." And that way they can identify that early.
Carl Schattke:
Right. So one thing that the librarian would be responsible for is setting up the properties of each component. And you could put in there a field for aqueous wash and you could have, "Okay," or, "No." Like it could be a yes, no, aqueous wash, or there's other types of washes as well. Or you would specify, "No clean," like if the cleaning process is not allowed, then you might have on property, "No clean," and if you have that on your board then it would mean it couldn't go through that kind of a cleaning after it was in the assembly process. So very valid point. All right, thank you for sharing that.
So signal integrity engineers are engineers that are specifically looking at the electrical characteristics of a printed circuit board, and they'll do modeling on that printed circuit board to make sure that it's meeting the timing requirements, to make sure that it's meeting requirements with respect to signal integrity, make sure that there aren't going to be problems with clocks interfering with reset lines. Like, there's a lot of signal integrity problems that come up. They're going to be modeling that with software typically and with their intuition and experience to look at the design and see what potential pitfalls there would be with the way that it's currently designed. They need to get accurate information from the electrical engineer as far as the parts and timing that's required, and then they'll also work with the printed circuit board engineer to see how that has been physically labeled, physically placed and routed, and make sure that the timing patterns work, that the delays are all in place.
And they'll also work sometimes with the electrical engineer to set up the electrical properties to have a successful route of that board that's going to solve the electrical characteristics of the printed circuit board. That's their primary concern. They get their information from the electrical engineer, they also get their information from the PCB designer, and they'll also get information from the software that they're using. They might also use third parties to do deeper analysis if there's something like a radio or something like that on the board, like Bluetooth antennas or wifi antennas. There might be different things that they would look at from a signal integrity standpoint. It depends on what the product is and what they're trying to design.
And then also, they give information back to the designer, "Please change this, please change that," and they make recommendations. Their deliverables might be a PowerPoint or a document or an email, or something that's going to explain to the PCB designer what needs to change to make that routing successful.
Julie:
Carl -
Carl Schattke:
They're in partnership with the PCB designer.
Julie:
And as we get into high speed designs right now, do you know customers are requiring to choose lower loss materials, so lower DF number on the data sheets. And one of the pitfalls I see sometimes is a signal integrity person might be talking to a material vendor about a low loss material, they design it in for the prototype, you go to buy the prototype and nobody in the United States makes it because it's an uncommon material. So the high specialty materials are like ... everything is a whole science in itself. So if you've got quick turn requirements and you're trying to pick a special high speed, low loss material or low decay material to get your impedance lines wide enough, make sure that you're checking with your NPI suppliers to make ... that they are qualified for that material and that it's readily in stock.
Carl Schattke:
Yeah, that's a good add. So to further clarify that, so what do they need? Well, the stack up ... they need a good stack up because that's going to set the signal integrity for that board, is the stack up. So every signal that is transmitted comes back on its return path, so the stack up is a large part of the signal integrity of the design, so they need a good stack up. Typically they'll get that information from the suppliers. That was a great add, Julie. Thanks. Now, power integrity engineers are a lot like signal integrity engineers, but they're looking at the power draw of a particular circuit. And sometimes the electrical engineer will play this part too, but there are companies that have power integrity engineers that specifically look at DIT ... DI over DT, like how fast does the current get used? And what's the ramp time for that current to come to a certain levels that's going to allow the changes logically to take place?
So power integrity engineers make sure that power gets to devices in a reliable way with a low enough noise to be useful for the particular IC that's using. Certain ICs are way more susceptible to power bounces, or ground bounces as the case may be, because they might be referencing that for what they do. Like if you have a radar antenna or something like that, it's going to send information out and get it back, and it's going to compare it to that reference. So if the reference change between the time it went out and bounced back, well, now your values are wrong, right? So it's really important to have a good power supply for a device like that.
Another situation might be a processor that can turn on a whole buss all at once, and it draws a lot of energy right at that time. Well, if you don't have an adequate power supply with enough capacitance, it won't be able to deliver the energy to reliably keep those ones and zeros in the right place. So power integrity is a really big important part of the effectiveness of a product. They need to get information from the component suppliers about how much current it will draw, they're going to get their information from the electrical engineers. They are going to give information to the PCB designer as far as trace [WIS 00:42:56] and layer usage in order to make sure that board is designed in a way that's going to be effective from a power delivery standpoint.
On high power boards, they're going to be way more involved in boards like that where a high degree of current and voltage is being transmitted. So they work out the equations to make sure that the power supplies are going to be functional for the product. Their deliverables are typically going to be design guides and information for the PCB designer. RF engineers, what do they do? They look at radio frequency type products. So this would be like the antennas I spoke about just a few moments ago. They're going to be creating models and CAD data that will make sure that antenna transmits or receives at the proper frequency, so they're concerned about the length of the antenna, they're concerned about the width of the antenna, they control the amplitude of that signal, and they tune that radio based on the harmonics.
So there's a wave, and they're designing what's going to capture that wave in the most accurate way possible. So when that waveform goes ... they're going to be designing something that's going to capture that wave, right? So that's what radio frequency engineers are working on, typically the radio part of a circuit. And they need really good software to help them with that, they need to be able to operate that software really well, they need accurate detailed information from the PCB designer. They'll take that information and run models to accurately tell if that's going to be doing what they expect, and then they also put it into the real world. When the board comes back, they'll put it into a chamber and actually take measurements off of it to make sure that it's doing what it's supposed to do.
And having a good RF engineer will be really key to developing a radio product, because as the effectiveness of their work gets better, there's less loss. So if you do a bad job on an antenna, maybe it gets like halfway through this room. If I do a good job on the antenna, it might get all the way through this room. So they're going to impact the range of a product based on the work that they do, they're going to impact its susceptibility to other problems. So radio frequency engineers play a big part in the goal being scored.
Carl Schattke:
Go ahead and share something.
Julie:
And so these days we're seeing a lot of radars for automotive products, and one of the ways to keep the radars lower cost from a printed circuit board point of view is to use your high frequency material on layers one and two, and possibly the lower layers, and do what we call a core stack up. Well, whenever obviously when you've got an antenna or a high speed circuit and you're drilling a hole all the way through your stack up, but your signal only needs to go to an internal layer, that creates a little antenna itself, right? So people are either doing back drill if they've got enough room to ... once you've got your via hole and then we drill a slightly larger hole on the bottom up, and it's a controlled depth drill to help give a shorter signal instead of having that via go all the way through.
Bu on the true RF boards, if there's a high speed material, say layer one to layer two, the lowest cost way to have no stub and no antenna hanging off those is to do a laser micro via from layer one to layer two. And one of the common mistakes I see here is on those holes, they're blind holes, so they just go from layer one to layer two. They're shallow. And for us to effectively plate those, we have to have an aspect ratio of less than .8 to 1, so that means the diameter of our drill actually has to be bigger around than the depth of copper plus the dielectric that we're drilling. So a lot of times I'll see a radar is set at a certain frequency, they'll have all these mode suppression vias and all their other vias, but they'll have ... I recently received a design with a 10 mil core, which is a standard Rogers core, but the vias were only supposed to be 6 mil.
And we can't do it, we can't plate those. So this whole radar design was based on that and it could not be produced in volume production. So when you're doing specialty stack ups like this, make sure you understand the process capabilities of the different drill structures and what drills you have available for these.
Carl Schattke:
Can you also tell them what makes a material RF? Like you said, "RF material," what's the property of RF versus a regular material?
Julie:
Yeah. So the RF materials are lower loss materials, and a lot of times they are not supported with the glass, like a standard FR4 high TG material that we use. The glass has a certain decay, the resin has a DF loss, and the resin has a certain loss, and those combined give the total loss. And the decay on the glass, I can't remember what it is, but it's a little bit on the high side and so is the loss. So what happens is in the RF materials, they make a slurry of a low loss material and then cure it, and then make it. So a lot of times the RF materials don't always necessarily have that FR4 to support the material through the fabrication process.
So we're working with materials now, standard FR4, high TG, all the way up to ... there's a whole pyramid of materials. And as we get into them, they get obviously more expensive, and quite a few of them are more difficult to process. But in trying to keep up with 5G automotive radar and everything, all the material suppliers are working frantically to come up with supported laminates that also have that low loss to support the high frequencies. So one of the things they'll do is they will ... there are two types of materials used for high frequencies, one is really standard decay, that's like about decay=3, and one vendor does it by a slurry of material that it just has silicon slivers or silicon balls to help support it.
But another supplier has figured out to just make it really consistent glass resin and super consistent in the process so that it turns out to be the same decay. So there are multiple ways of going. So when you're trying to figure out your RF material, sometimes the best lowest loss materials aren't the easiest ones to fabricate. And fabricators who specialize in RF, not just high speed, but true RF, like antennas and everything. There's a lot that goes into that control of the sides of the traces so they're not widely trapezoidal, because the signal sees what's against the laminate if your antenna's on layer one, right? Or wherever your etch is.
So the fabricators who really have experience in RF circuits are probably your better choice rather than somebody who says, "Yes, I can do it," because they don't know the nuances of the RF. And that's another good point, is one of the things that the simulations for RF, to accurately simulate using like HFSS or one of those programs, the programs are great, but if we don't have the correct approximation for the traces, the trapezoidal shape of the traces where they're narrower on top, bigger on the base, if we don't have all those representations built accurately into the model, including possibly all the way going down to what surface finish it is-
Carl Schattke:
Yeah, absolutely.
Julie:
... because ENIG nickel tends to be ferromagnetic, so at about maybe 5 gigahertz or higher, especially on long traces, it introduces its own little bit of loss. So even the ENIG ... and we have an RF engineer on our team, his name is Tom Buck and he's one of the smartest engineers I've ever met in my life, he complains if he has to go and actually build the models to put like ENIG, the nickel and the gold, because the models take a long time just to draw them correctly and get them right in 3D. So RF is just ... high frequency is always been like black magic for everybody, right?
Carl Schattke:
It's definitely one of the more advanced engineering ... like it's a very narrow discipline, and they impact the cost a lot. Like we talked about materials, typically the low loss materials are a lot more expensive than the normal materials because they're not built in the same volume and there's more work to developing those materials. Yeah, RF ... so who gets the information from them? It might be like a exported shape that would go into the printed circuit board. Their deliverables are typically going to be CAD data or some information on loss to the electrical engineer and things like that, like what they modeled, like the poser models to the team to be able to look at.
So EMI and EMF engineers, what do they do? Well, they make sure that products are going to be compliant in the range of radio frequencies that the FCC other bodies are concerned about. They make sure that those products are going to be safe for people to be around. If we have a lot of electrical energy and we have people next to it, you don't really want to park your body next to electrical energy that's doing bad things for you. They need to know what the products are and what they're doing, so they take the design from the board designer and from the electrical engineer, and they're going to model that. Oftentimes they'll work with a finished board prototype to make sure that it works in the situation that it's going to be employed, and whether it's a cell phone tower or automotive vehicle or something. All parts have to pass FCC in order to be sold commercially in the United States.
Around the world there's different bodies that also deal with that. They need to get accurate information from the EE and PCB designer, and then they will give information back to the electrical engineer and the PCB designer in order to make the changes that are going to bring about the signal integrity that's going to help that product pass. So they're going to be giving you examples of changing the location of an antenna on a design, they'll do things like changing the size of the antenna, they might do things like changing the power to the antenna, they might do things like changing the shielding around certain parts, they might be doing things with the mechanical engineer to change the material that's around a printed circuit board so it behaves in a way that's going to be better electrically.
I have heard of radios being designed that went into a metal case and they didn't work. Imagine that. It's happened. If you don't understand the parameters of all the teams around you, you're not going to have a winning team. That's why we're having this talk today, so that we can learn what the different players do so that the whole team plays better together, and you know what you do, and you know what the other people do. And if you can cover for them and they can cover for you, that's how you make a really high performance, high caliber team. So it's really important to know what other people around you are doing and how they do it so that you know what they need and what they give to people, and what they get from people.
So that's why we're covering this in detail. I know it takes a long time to go through it, but that's what our talk's about, is building that better team through an understanding of this.
Julie:
And there's also a level of trust here, because as engineers, we tend to be a little bit egocentric and confident. We want our product to be the best. But a lot of times we can't know everything, so we can't get upset if somebody comes in and says, "Hey, I've had a problem on a circuit that's like this. Here's what happened." Take that input and have trust that they are trying to give you advice of something that you just haven't seen before, and they're trying to help you prevent it. So with my customers, I always ask them, "IF I feel anything in my gut that something could go south on this, can I just bring it up and we can discuss it, and take it off the table?" And I'll bet you probably 99% of my customers are happy about that.
And there are times when I bring something up and I'm just kind of sick to my stomach about it, and we review it for an hour or half an hour, or 10 minutes, and we find out that it wasn't a concern after all, but we were glad that we put it to bed. So this is really a team collaborative effort to bring all the knowledge, because the longer we're in this business the more stuff we've seen and the more experience we have. And the more people from different groups and different expertise, the better off we are.
Carl Schattke:
Yeah. I'd like to add one thing to that. The stupid question is the one that doesn't get asked, okay? If you could have asked a question that would have saved a product, the stupid question is the one that you didn't ask, right? So you really need to be inquisitive about the product. And if you don't understand it yourself, then please feel free to ask that person that you think understands it, how they understand it, and ask them to explain it to you, and then that way you'll continually grow your knowledge in the field.
I always like to ask about, "Why are you doing it this way? What's the concept behind the design, not just the design?" It really helps you learn a lot more about what their thinking is. So compliance engineers are going to be working with different standards bodies, it might be the UL. So they basically are working to make sure that a standard is met, so that could be like a UL, like Underwriters Laboratories, make sure that a product is safe for a consumer. Or it might be the FDA is going to make sure that a medical product is safe for a patient, or it might be a FCC compliance, where they're making sure that a radio meets the frequency requirements. So that's what compliance engineers do.
They need information from the board designer, they need to be aware of the standards that need to be met, and then they get their information from those standards bodies, and also from the designer and the electrical engineer for that product, and the mechanical engineers. And then the information they would give back would be to the EE and the printed circuit board designer, and that would be information about what kind of changes could be made to bring about compliance. So they would make recommendations to improve the product, those would be the deliverables that they'd be giving.
Reliability engineers are going to be engineers that are tasked specifically with improving the reliability of a product or a process. They need to get information about the design, the build materials, electrical schematics in the PCB design, and then they're going to take that, look at how it's being manufactured, what materials that have been chosen for the manufacture and the assembly. They're going to look at things like solder type, solder quality, reflow types, rows, make sure that materials are going to be reliable. And then they will set up experiments often to determine reliability, might put it through vibration testing, they might put it through environmental chambers, they might put it through chambers for heat and cold.
There's a variety of different tests that they'll do for reliability. They might put it under high voltage to make sure that it passes CAF testing, that's conductive anoid filaments that get created when there's a high voltage. The ions will migrate through any open space, so if you have a printed circuit board that's been drilled, that drill is going to fracture that material as it drills. We've all drilled holes, you see it kind of gets gnarly around the hole, right? Well, that is fracturing the substrate that's being drilled, well, that creates a gap, and the ions when there's high energy, can actually flow through that and make shorts in a circuit and ruin it, and it won't be reliable.
So they put together information about this. And then the board manufacturers have specific standards that they have set because they know and have tested that. So they might say, "These two kinds of holes have to be this far apart," and as designers, we want to make sure that we put our design rule in, that if we can only have those holes 10 mils apart, well, they better be 10 mils or farther. And we need to put a design rule in to make sure that we're meeting the needs of the board manufacturing for reliability. And any time we can put holes farther apart, it's better. Any time we can put traces on our board farther apart, it's better, unless of course they're coupled, then that doesn't apply.
Julie:
And more copper. Bigger is better, copper is king in printed circuit board design as far as high reliability.
Carl Schattke:
Okay. Yeah, so the deliverables for reliability engineers are going to be typically standards for the PCB designer to use. They'll be making suggestions on how a design could be improved, it might be a component change, it might be a circuit change, it might be a material choice change, maybe we change the finish, maybe we change the sequence of operations, maybe we change the stack up in a certain way? So they're going to be looking at all of those things. They're impacting cost and they're impacting quality, but they're mainly concerned about the quality of that product and whether or not it's going to last, and whether or not it's going to meet its mission wherever it's put into.
There's all different environments. I talked to somebody in the audience today, they design deep sea equipment, talked to another person ...
Carl Schattke:
Design deep sea equipment. Talked to another person that designs stuff for outer space. So we go all the way from the bottom of the sea, high pressure, all the way out to space, vacuum-type environment. All right.
Thermal engineers are going to be looking at the heat dissipation characteristics of a circuit. So they're going to be looking at how much wattage is on a circuit board and how do we dissipate that wattage in a safe and effective manner. They need to know the schematic. They need to know the printed circuit board. They need to know the components that are on there. They need to know the physical and electrical properties of that. How much heat is going to be dissipated, for how long, at what frequency. Is it short versus is it on all the time? That kind of thing. So they're going to be looking at that.
Then they're going to come up with calculations of how much area on the printed circuit board is needed as a heat sync, or they're going to come up with... That would be considered passive cooling. Or they might look at active cooling where you'd have an actual heat sync that would be touching the board and touching hot parts on that board and then using liquid or some other, Freeon, or something like that, to cool that circuit down. So they're going to be putting that kind of information together, and they're going to be drawing off of the CAD data, and they're going to be drawing off of electrical thermal models that do that. So they'll have software that's going to help them with that. They'll also have... They'll be able to calculate the wattage that needs to be dissipated. Things like that.
They're going to get their information from the board designer and the component manufacturers. They give information back to the PCB designer and to the mechanical engineer to make the components. For the mechanical engineer, they would typically make a custom component for that circuit board to cool it down if it's going to need a custom heat sync. Or they might buy an off the shelf heat sync. They could recommend that.
So they're working primarily with the mechanical engineer and the board designer, and then of course interfacing with the electrical engineer to get the information they need on that. Then they're deliverables are going to be information on the components and things for that designer.
Structural engineers are going to be looking at... They're going to get information from the product designer, as far as how much room things are going to take and how much... Structural engineers are going to need information on what the environments like and then they make sure that the product's going to meet that environment's needs. They're going to work very closely with the mechanical engineers and electrical engineers to make sure that... They design cabinets and things like that. Our infrastructure is going to support that. It could be the body of... It's an embedded system, they're going to do the enclosure. Things like that.
The information that the structural engineer would be needing would be... They're going to need the CAD data and they're going to give CAD data out. That would be what the structural engineers would be doing. They're going to make sure that things aren't going to break, that are mechanically...
Firmware engineers are going to get information from the electrical engineer. They're going to come up with the software programming that's going to boot a device, or boot a computer on a printed circuit board. Then they'll also work on what is going to make that... They're going to be developing the software that's going to make the computer meet its function.
Software engineers are like firmware engineers, but they're focused more on the use of that computer than making sure the computer works. So the firmware engineer wants to make sure the system works, the software engineer is going to be working with that functioning computer. So, that's the difference between those two engineers.
Product technicians are going to be setting up labs for... They're going to be basically working on that product, making sure that it can be built. Making sure that there's safe environments to test it, making sure that the electrical engineer has the tools at his disposal to debug the board, make sure that he can bring up the board so that... Product technicians are helping with that kind of thing. They might set up experiments and set up a testing apparatus. They're going to be working in the lab primarily with the schematics from the electrical engineer, working closely with the mechanical engineer. They're working with everybody on the team to make sure that that board can be built, put together, and work. Their output is basically, make sure that product types can be brought up and work. They finish their task when they've enabled the rest of the team to have a decent product that works.
Validation engineering, their input is going to be the CAD data from the PCB designer, or the electrical engineer. They're going to model that and look at it to make sure that it's a valid solution for manufacturing. So they'll load the board into software, they'll look at the errors that that software thinks are important, then they'll share that back with the board designer to make sure that it's a valid design. So they'll be using software to do that. Then they'll be giving that data out back to the PCB designer so he can make the changes and they'll recheck it, make sure that it's ready to go to manufacturing.
Component engineers are going to look at reducing the cost of the product. Typically, that's their main value to a company. The component engineers are also going to be making sure the data sheets have been read correctly by the librarian. They're going to be making sure that as it goes in to a parks database for the company, that all of that information is correct and where it needs to be. The component engineers are going to be researching new components to bring to the electrical engineering team, that might be more effective or more cost effective or more functional than the ones that are already being used. Component engineers are going to be reducing the cost of the product, typically by coming up with newer, better solutions, or more cost effective ways to do things that a company's already doing. Their output is going to be information to the electrical engineer so that he can make changes the circuit that are going to help the company produce parts that are more functional or at a lower cost.
We have a lot of different offensive players and I think we've reviewed the offense. Any questions on the designing process and whose doing what? Did we miss any or did we... Is there any that we missed? Do you have somebody on your team that you think is doing something that we didn't cover? All right, we must have did a pretty decent job. I don't see any hands going up.
So thoughts on the design profession. These are really, really challenging positions and we as engineering staff are challenged with coming up with new solutions to previous problems or better solutions to previous problems. The work that we do on the design side enables manufacturing to have products to build and the value of a good design team is a good design. It's more attractive to the marketplace. It meets the needs of consumers, and it's simply going to be a more effective, most cost effective product for that company. It's really the life blood of a growing and developing company is to have good engineering. So the offense is really charged with making sure that the new stuff is better than the old stuff and that it's going to be more attractive in the marketplace.
Now we're going to cover the defensive players. We've segregated our offense and defense into the design and then the defense is building a defendable device. It's going to be solid in manufacturing. It's going to work. It's going to be a device that's going to last and be reliable. It's going to have high quality. It's not going to be junk. It's going to be a long time before it's junk. It's going to have a long, long life. It's going to meet the needs of the consumer for what it is. It doesn't necessarily have to have a long life, it just needs to be effective for the life its designed for.
Julie:
The reason we're going into so much detail about how much access you can have to information on a new design, is because you can always build efficient cost into a design. When you start a project and you plan accordingly, it's a lot easier to take cost out of a design at the inception of a product than it is after its already been gone through the MPI production. A lot of times we'll get circuit boards and it's already a complicated board and they say, "How do we save on cost on this?" And that, "How do we save on cost," should be at the beginning of a project and really design the stackup, design the components, design the spacing, design the technology, to be as cost effective as possible. So that's why we're, at least for me, is why we're going through so much detail on the resources that you have, so that you can choose your best resources available when you're designing in the first place.
Carl Schattke:
With respect to costs, what are some of the ways that you can drop cost out of a product? First for cost on printed circuit boards is make is smaller. A smaller board is going to be automatically lower costs than a larger board, for obvious reasons. You can make it fit into the panel better. We talked about that earlier, that's another way to make it cost effective. We can reduce the number of layers in the boards. That's going to lower the cost. Those are your big drivers on costs.
The other thing that we can do is we can reduce the number of layers that are used for component assembly. Now you have one less assembly step if all your parts are on one side. We can reduce the technology of components that are on that board for cost. If we have surface mount parts on both sides, that's two surface mount assembly steps and different type solders for each side, so one is going to have a lower melting point.
The other thing that we can do is put through hole parts on one side, surface mount parts on the other. Now we've reduced a step. So the number we can reduce the sequential lamination and make a through hole board, a through hole board is typically going to be way less money than a sequential lamination board. There is an exception to that. If you have a sequential lamination board with no through hole vias, no parts requiring through holes, well now the whole board can be laser drilled, top to bottom on any layer. Laser drilling is a couple of seconds per panel. Mechanical drilling could take hours on a very complex panel. Usually not, but it's going to take a lot longer, like orders of magnitude longer, to mechanically drill and do laser drilling. So, that's another way to reduce cost on your board.
Those are your big drivers for lowering costs. We can also change plating finishes. We can also change exotic materials out. If we can design a board with lower cost materials, obviously it's going to save money. We can also save money by doing things that would speed up the assembly of that board. Maybe create a board that mounts as a snap-in instead of a screw-in. There's a lot of different ways the mechanical engineer and the printed circuit board designer can reduce costs in a product and you have to be really creative about your packaging and your printed circuit board design to get the lowest cost out of a product. There's usually a lot of opportunity there that isn't looked at if we're not really thinking about it, or if we don't know what the options are. So if you want to drop cost out of your product, you really want to involve as many people as possible and have them take a look at the design and look at ways to optimize that or design out some costs.
Not all boards can be cost reduced. That is the sad truth of it, but a lot of them can be, and often times the total cost is not the cost of the board. It's the cost of the board, plus the time to market. A board that comes to market faster than one that doesn't, and gets to market first, is going to capture market share more than a board that takes a long, long time to develop. It might have a lower cost, but by that time the consumer doesn't care, he already bought the high cost part because it was available. So time to market impact is a large share of cost as well. Definitely the cost equation comes in to factor on anything that we build and most cases.
Julie:
One of the ways to determine whether you do need laser micro vias with sequential lamination, because the laser micro vias are blind holes and they only drill from an outer layer of a stack, down to the next layer, and then we plate, and then we put more material on the top and the bottom, press it, which is lamination, drill the out layers micro vias. So every time we go through one of those lamination cycles
Carl Schattke:
We can also do skip vias though right? Those are laser drilled?
Julie:
Yes, but normally only on the outer layers. We can't do those buried inside. But the rule of thumb is, once you hit a .5 millimeter pitch BGA, center-to-center of the pads, if there are more than... Row by column, if there are more than two rows by column, more than six pins or eight pins, then any time you have a .5 millimeter pitch BGA or less, you have to use micro vias. And .5 millimeter pitch can - To fan it out. Because if you have all those rows and columns... You can fan out your outside row. Can you guys picture this? If you've got a BGA with rows and columns, rows, columns... The outside is what we call the picket fence and you can fan those out, route those out on the outer layer.
Then the problem is, if the next row in, if the distance between the two BGA pads is too small that you can't run a trace, on a half millimeter pitch party you can one three mil line trace between those two balls. So you can technically get two rows fanned out per layer. But if you've got a lot of rows, then you have to use laser micro vias because to get inside that field under the device, the pads are only 10 mil, 10 or 12 mil, that does not support a mechanical drill, which requires the drill bit diameter plus eight mil for the annular ring.
Carl Schattke:
So going back to the component engineer, and the electrical engineer, by choosing a smaller pitch part they might save money on that one part, but the total cost might go up for the reason Julie just stated. The electrical engineer and the component engineer need the design for total lowest cost. Sometimes one part that may be lower could actually raise your costs because it moves you to a different technology. So something to be aware of as you're designing your products. Any other questions on cost? All right, we're going to move on.
We're going to consider that are defensive players are going to be our PCB manufacturing team, could be field application engineers or liability engineers, quality engineers. These are players on our defense. Let's look at...
Julie:
Did we miss the...
Carl Schattke:
Did we not put-
Julie:
I've got the wrong... Oh.
Carl Schattke:
Let me go back here. Hang on. Okay, so let's look at our team for PCB manufacturing.
On the PCB manufacturing team, the PCB designer and electrical engineer are going to be interfacing with the PCB manufacturer. It could be a FAE like Julie, or it could be something that sets standards at the company for design. They could also be just interacting with the website that has the details they need for PCB manufacturing. The PCB manufacturer is going to work with their customer to make sure that all the information is there on the fabrication drawing. That's going to be the input that they get. Then the information that...
What they're going to do with that is do a pre-job check. They're going to take that information into CAD, they're going to load it in to there, their going to run it with software to check to make sure that that board can be built. If it hasn't been designed right, they're going to feed that information back to the PCB designer and the electrical engineer that submitted the work to them and they're going to say, "We've got problems with this. What do you want us to do with it?" And they'll typically put that in a spreadsheet for the electrical engineer and the PCB designer to resolve.
With as complex as boards are getting today, it's pretty common for... It's probably way more common for you to be giving feedback than not giving feedback, right?
Julie:
Yes.
Carl Schattke:
Probably 100%. What percentage of boards are getting feedback? What percentage do you get in that you actually have to give feedback back on?
Julie:
100%.
Carl Schattke:
100% okay.
Julie:
Because there's always going to be something like... Like we want to put our own internal marking somewhere and we need approval to do that in copper so that we can manage our lots. So I never see a circuit board come into a site and not have TQ's. Unless it's a re-spend and the customer actually took the previous technical queries and designed them out or corrected them in the next revision.
Carl Schattke:
Yeah, very important. Also, be checking for things like matching part numbers. They'll be matching things like PO's with part numbers. You would be surprised how often that gets messed up. So making sure that all of the ducks are in a row is really, really important when you start the production. Once they start, if there's been a mistake, a lot of people are going to be unhappy. They're going to be unhappy because they built a board that the customer's not going to want to pay for. The customer's going to be unhappy because they didn't get it on time. Customer's going to be unhappy because, "Why didn't you tell me about that?" The manufacturer's going to be like, "Well I did tell you but you approved it." So this communication link between the manufacturer and the designer is extremely critical, to be open and honest and forthright. So the better that they communicate the problem, the better that the engineering team can look at the problem.
Some of the problems that I see with that link from manufacturers is they'll put a picture on there, but there won't be a reference designator, or an XY location, I'll have no idea what they're talking about and I have to go back to them, "What are you talking about here? I can't find that on the board," because they show a small, tiny part of the board and it's a big board and I have no idea what part of the circuit they're talking about. So manufacturers, make sure that they give you details like XY coordinates, or some reference designator that it's near so you know exactly what they're talking about. Or they might list like, "We see 20 instances of a net antenna." That's great, where are they? Each of those is important and needs to be cleaned up individually.
Sometimes I'll see information come in in a general format where it would be much more helpful if it was specific. Or, "We see net ties on your board but we don't understand how..." So we try to send a chart with net ties information. Like, "This net is purposely shorted with this net, in order for us to be able to have sense lines with different rules and our power rules." So there's a lot of interesting ways that electrical and PCB designers try to give the information to the vendors, and they're at the mercy of the data that they get. They're not privy to all of the information that went into the design. They only see the XY data, how big it is, what it's got to do, and they're trying to build it in a way that's going to be accurately built. So that communicate is extremely important to building a good board that'll be defensible once it's built.
Julie:
Yeah, I'd like to add to that.
Carl Schattke:
Yeah, please do.
Julie:
That's a really good point. So I have worked in both contract manufacturing and printed circuit boards, and one of the... When you go through the technical query, or DFM process with either your contract manufacturer and your printed circuit workshop, one of the biggest holdups is that if a customer approves something on a DFM form or a technical query without re-spinning their Gerber files... Because a lot of times on release documentation that's really difficult to do. I recommend setting up it's own file for stack-ups for a raise that the board is going to be purchased, format that it's going to be delivered for manufacturing.
Also, technical queries. So that if you're not doing your own assembly in-house, and you send this package to the contract manufacturer, the contract manufacturer doesn't get hung up on that stuff and have to go back and forth between your buyers. Usually what happens is, a tester process engineer at the CM says, "We can't build this because we don't have enough space on the edge to grab the parts. So we need to put this circuit board, or groups of small circuit boards, in an array with rails on the long side so we can convey it down the assembly line." Well, that goes back to the program manager who takes it back to the buyer, who takes it back to the circuit board fabricator, and you've got 10 people in there.
So make sure that you're pre-planning your whole design, especially the printed circuit board assembly, for the assembly process, not just for the fabrication. And that you've got the good documentation for all those requirements when you take it to the end company or end division that's going to do the assembly.
Carl Schattke:
Yeah. You bring up a really good point, and I see a lot of back and forth with suppliers in this area. So one thing I'd like to talk about is panel drawings and when they're needed and so forth. And also, you mentioned working panel strips, so when those are needed.
If you have a part that overlaps the edge of your board, but you've asked somebody to route that panel apart when the components are... If they need to separate that board and you have a component overlapping the edge, those two processes don't mix because that router would tear that component apart. So the board designer needs to put a slot and design a panel so there's a slot underneath that component when its installed. Then when they come and separate the panel, they can do the routing but they don't have to go underneath parts to do it. So the board designer is responsible for making a good panel. If the PCB connectors, or something like that, are going to overlap the outline of the board. So, that's one point where you're going to need a working panel.
The other is if they need to have tooling holes and there's no room on the board for assembly for tooling it. They need to register this in the machine that's going to do the pick in place. If there's no room on the printed circuit board for that, they need to have a working panel outside of that to put some of those tooling holes where it can be accurately, repeatably put into the tools that are going to be doing the assembly. So that would be another area where you might need a working panel for that printed circuit board. Now that's going to increase the size, it might increase the cost. So something to be aware of as you're designing your product. And this is another aspect where the communication between design and manufacturing could be the difference between winning and losing. Knowing these things or not knowing these things could raise costs, could knock your product into a higher price point, could bring it into a position where it's less desirable with the consumer, or less profitable for the shareholders.
Julie:
And that should be done... There should be a pre-design plan on, how is this board going to go down an assembly line. Make sure that if you need support, like rails for connectors and everything. So that means, what's that team? That's your electrical, your mechanical, your components engineer, your circuit board fabricator who will be able to optimize the material utilization on their fab panel sizes. For instance, even though 18 inches by 24 is the most common fab panel size globally, a lot of the bigger volume shops also do sub-cuts. Maybe 21... And some can do over sized. 21 by 24, is common at quite a lot of volume shops.
So we're talking a really big team, just to conceptually dimension this product, set it up with rails or space on the edge of the board for conveyance down the assembly lines, and make sure that you're accommodating overhanging and special parts. That should be done at the beginning of the project, not once you're ready to release your fabrication drawing and your Gerber files to your fabricator.
Carl Schattke:
All right. Thank you. That play between design and manufacturing is a lot of communication and there's a lot of different way that that information can be communicated. It can be through email or it can be through Excel files, or it can be through PowerPoints. The drawings for the panel drawings. Who typically does the panel drawings? Well, it could be the PCB designer. It could be outsourced to the assembly vendor. They might say, "I'd like my panels arranged like this." And they might come up with the assembly drawing for the PCB manufacturer. But somebody has to be responsible for it and it's typically... There are a lot of assembly vendors that like to design the panel or that's something that the PCB designer has to do. But it has to be clearly delineated, whose going to do it. Usually it's a commercial decision and a physical design decision as well.
Sometimes you don't want to do panels at all and you'll just use a smaller panel for the fab, or it might be one board per panel. That's another way to save money in your project.
One of the players on your defense is a field application engineers. What do they do? There's two types of field application engineers. There's ones that they're suppliers. There's ones that component suppliers that would then come in and work with your team. So an FAE from a part manufacturer would come and meet with the electrical engineer, give him the specs on the part, tell him how the part is used. They might help to bug the part in the lab. They might look and do a design review on the part. They might recommend alternative parts. They might recommend parts that are under a shorter lead time. Right now there's a lot of problems in the industry with long lead times. Your field application engineers are going to be able to tell you, "How do I solve this problem with no... This part's a year out. Well, how do I get that part?" "Well, you can't get that part but here's another part that you could substitute. Here's a way that you could use that alternate."
So field application engineers could be captive with a component supplier or they could be captive with a distributee of components. Then there's field application engineers that are associated with your contract manufacturers. Julie works for a fabrication contract manufacturer. People come to her company wanting to build something and she makes sure it gets built right. She's going to give feedback on what can be done differently. Make sure you're doing this. Make sure you're doing that. She's going to give stack-ups to the PCB design team. She's going to be giving rules to the PCB design team for proximity, for aspect ratios, for plating. All these kind of things or rules that would come from that contract manufacturer FAE. Those are your defensive players.
Reliability engineers, each factory that's building stuff has a team that's making sure it gets built reliably. They're running tests on the process when they build printed circuit boards. I've heard anywhere from 60 to 375 separate operations in the construction of a circuit board. So depending on how automated that shop is... But all of those processes need to be managed. The plating chemistry needs to be checked daily or in a continuous process. There's time tables for how often drills are replaced.
Carl Schattke:
There's timetables for how often drills are replaced. There's timetables for how often machinery needs to be maintained. There's timetables for meantime between failure of different equipment. If you're running a factory to build printed circuit boards, it's not unlike having a car and needing routine maintenance. I got to change the oil every 3,000 or 5,000 miles. Well, there's things in a factory that need to be done every so often that need to be planned and these reliability engineers in those factories are making sure their equipment's going to be up, that it's working in a reliable. And there's engineers that are working on the quality of that product. Do you want to add anything to that?
Julie:
No. It's just interesting a typical fabricator ... I think our San Jose plant, they do over 200 chemical titrations every day. There's a lot that goes into building a printed circuit board, and you've got a lot of reliability engineers, and a lot of processes that have to be controlled and maintained through using SPC data, and monitoring and measuring. There's a program called Exact that we use that the material is actually like ... for a six-layer board, we buy two cores, which are double-sided material. Then we put copper on the bottom, pre-pregs which are the partially cured material. A core that we already etched the top and bottom circuitry on, pre-preg and other core, pre-peg, and your copper. Shoot, I just lost my train of thought. Where was I going with this?
Carl Schattke:
We'll go back to reliability engineers then. One of the other-
Julie:
I know what I was going to say.
Carl Schattke:
Oh, go ahead.
Julie:
Anyway, so those cores and during lamination, that material shrinks and we have to actually grow our artwork. We have to pre-predict how much of that material is going to shrink through our processes. We don't take your artwork and build it one to one. We actually have to increase it a little bit based on the material type. Some of the feedback in the information tools that we're using is that program called Exact where for every stack up that we do, we're putting in this information for all our cores and all our materials, and then we're using x-ray targets and measuring the panels, so that we can get an idea and that we for every material and different families of stack-ups, how much the material moves, so that we can predict it correctly for the next time we laminate with that material. So-
Carl Schattke:
I want to talk to that. When they take your design in as a designer, and they bring it into their fabrication facility, they know what the etchback factor is. They know when they etch it for this weight of copper. They know how long it's going to be in the tank. They know what the profile's going to be. Also, when they put the layer sandwich together of all the different layers of materials, and they image it and everything, and before it goes into lamination press, they know when it gets laminated, it's laminated at a high temperature. It's going to expand, and then they cool it slowly so that it doesn't potato chip, and get weird dimensions.
Now, they know how much it's going to expand, and they know how much it's going to contract based on their history of building product. You couldn't just start out and start building stuff without that kind of characterization. It's really they have the knowledge to do that. They don't really burden the designer with that. They just look at the target ... they get the target where it's supposed to be. If it's out of target, they can't sell you the boards. They're going to be real accurate about that.
Yeah. Now, the quality engineers are people that are looking to make sure that that board meets the quality requirements that were on the drill drawing. That could be like plating finishes. They're going to be doing test for cleanliness. They're going to be doing test dimensionally, make sure that the outline is routed within the specified tolerance. They're going to be making sure that holes are drilled to the right tolerance. They're going to making sure that the IPC guidelines are followed, so that if it's class II, you can't have more than 20% of the trace missing. If it's IPC class III, then they're going to make sure that drills aren't outside of the via holes. There's a lot of rules that they have to follow. They have software that helps them.
If you go out onto the floor there today, you're going to see a lot of expensive tools that do this kind of work. Millions of dollars for some of these pieces of equipment, but it enables them to build printed circuit boards that are reliable, and high quality. That ultimately gives them the lowest cost. Even though those tools cost a lot of money, they're cost effective, because as they improve their yields, and reduce the rate of rejected parts, the less scrap they build, and the more that they get paid for, the higher the profitability of that company. It's a direct impact on your manufacturer, how well you design the board is ... If you design a board that's conservative, they're going to get a very high yield on that board. And subsequently, your cost of that board design is going to be far less.
You're going to pay that supplier less for a board with elementary type board technology versus a board with very fine pitch, fine line, close tolerances. That board is going to be a less reliable board, just based on the design that you created. The designer has a big impact on the profitability of the company, because the more ease with which it can be built, the lower cost you're going to get for that board. Over time, they're going to know what their yield rates are. They're going to adjust the cost of that board either up or down based on what they see. They might come back to you and say, "Our yields are terrible on this. We need to make some changes if we're going to keep building this in volume." They'll make recommendations to improve the yield.
A lot of what they do is with that initial check of the design is to help ... they're asking you to do things that are going to help their yield. It might be for copper balancing, or it might be for plating finishes or thicknesses or material choices. There's a lot of leverage that can be pulled to make the right decisions in that area.
Julie:
Good. That's about what we had left. Now, we'll talk about special teams players. ICT test manufacturing, what do they do? They develop the test fixtures, the test boards. They give data back to the ... when you get an assembled board back, they tell you it's a go or no go. There's stuff wrong, if there's stuff wrong, they say, "Go fix it. Don't use this board." Or they say, "It passes. Let's go put it in our product." The application engine, they're going to get the data from the design team. Application engineers are going to look at specific situations for a product. The sales staff is going to be giving feedback to the manufacturing and design, what they see from what are people saying about the product? What do they like? What do they not like? What could be done better?
Very important part of the loop is sales staff. Tool vendors are going to give us better choices for new tools. That's why they have this event with lots of machines downstairs. New tools, better technology, more advanced, better processes. Management is in that special teams area. We talked about what they do previously, but they make sure that people have the right things to do the job that they need to do, and that things are done in an efficient way. Management's going to be interfacing with everybody. They're going to be solving hard problems. They're going to be bringing people into solve hard problems. They're going to be getting rid of people that are problems.
Schematic review teams. A schematic review team is typically the smartest brains of the company. They'll come in together, and review a design and see if they ... they're going to poke holes in the engineer's work. They're going to come up with things that the engineer and the PCB designer did that don't make sense. They're going to be taking a hard look at least aspect of that, and the data they get in is that finished ... or think we finished it design. They're going to make suggestions for product improvement on that. That's with the schematic review teams. I'm including schematic review and design review in the same bullet point here. FAEs are going to be able to come in and look at design and say, "You really should change the way that this is arranged, because that's going to have too much noise." For instance, on a power supply, it's going to bounce all over the place and radiate energy. It's not going to be good for you.
It's not going to be as accurate as it needs to be. The return pass aren't going through other things. You got to get these other signals out of the way of that. They're going to be able to spot good practices and bad practices in your design. It's really great to have field application engineers that are intelligent enough to take a look at their product and your design and make sure that it would work. They can be a tremendous source for feedback on your design. Assembly vendors are also the ones that have to build it, and if something's hard to build, well, they're going to give you feedback on what it is that's hard about that product. They can make suggestions as well.
It's like these parts are too close. We get shadowing on this component. You put this tiny part in between two tall parts and we're not able to get the ... "We can't get energy down there to melt the solder in that place. Can you move that to the other side?" Sometimes a simple thing can fix a hard problem. A good assembly vendor's going to be giving feedback to the design team. Conformal coding is a process to add some goop on top of circuit boards that's going to protect it from the environment. There's a few different ways that that's applied. We typically don't want it on connectors that have holes in them. We don't want it to stop making electrical connection where we need one, but we do want it to stop making a connection between caustic air and the surface of our circuit board. We don't want sodium nitrate in the air to dissolve our circuit board.
I heard of one case where some circuit boards were shipped over to China, and the pollution in the air was so bad that within two months, boards that were working here in the United States for over a year went over there and two months later, they were failing. They ultimately found out that the problem was the air quality. The air was so caustic and the pollution so bad that it was actually destroying printed circuit boards. You have to be aware of the environment, and be aware of the environment in the different geos that you're shipping to. Box builders are the people that are going to be taking the finished assemblies and the boxes or the enclosures that the mechanical engineer designed, and they're going to be putting that project together. It might take several assemblies and put it together.
Basically, they're giving you the box, or they might even put it into a box, and have something that's ready to ship to a customer. That's what box builders do. Their input's going to be all the finished assemblies and what they need, and the drawings from the host company, and then also they'll put shipping together. Shipping's a big deal. I worked on a project very hard for a few months, we sent it out to a rapid prototype supplier, we had it assembled, we're trying to meet a show at COMDEX. The boards came in from the shipping, and the box was crushed, and more than half the boards in the package were destroyed. A lot of hearts sank real heavy when we saw the box and opened it up. Fortunately, there were a couple that were still good and we made the show. But we were supposed to have back-ups, and there were no back-ups for that show.
Because it wasn't shipped properly. Shipping's a big deal. It has to be packaged and wrapped tight. On our assembly drawings, we specify how things should be shipped, what kind of boxes they should be. Like that kind of information can be conveyed to your supplier and they have to meet it if you specified it. If you didn't specify it, they can do whatever they want. Like come in bent like those boards we got at a company I once worked for. Inspectors would be people that look at stuff and fix it. Yeah. All right. That's good. Equipment managers, tools and resource providers. These are other special teams players, information technology that make sure that software works at your company. That it's not getting hacked into, and things work.
IT make sure your computers work online. Online calculators, there's a lot of really good tools for PCB design online. Help you figure out a lot of stuff. Your vendors have them, but also there's some custom stuff out there. I'm not really going to get into the details at search engines, Gerber viewers, data translators, librarian wizards, and calculators, DFM validation tools, RF modeling tools, IMC chambers. There's different lab equipment, vibration testing, destructive testing equipment. Crash testing, safety engineering, ICT test. There's a lot of different tools that are resources that are available to us in the industry to check stuff.
There's also failure analysis lab. Let's say you have a design, it fails, you can't figure out why. Well, there's people that specialize in figuring out why your product didn't work. There's labs you can go to. There's consultants you can hire that are very adept and astute at finding those kinds of problems within your product. They might be expensive on a short term thing. They can save a product and ultimately get it to market successfully. We also have facility engineering, making sure that you have a place to build what you're trying to build.
In the design profession, we want to use all the resources at our disposal. That's why we presented so many different people, call them players that are coming to the game. You need to know your players. You need to know their strength, and you need to know their weaknesses for what your product's trying to achieve. Getting back to our game mentality is like make sure you know your players, make sure you know what they're good at. Make sure you know what they're not so good at. Make sure you ask a lot of questions. Close that communication loop with the ask, tell, ask.
Tell them what you need, ask them if they understand it, make them tell you what you need them to understand so that you know that what they're telling you is what you told them.
All right. Now, we're going to look at the playbook. This is different BKMs. Print circuit board design, how that happens. We kind of talked about that already. Board manufacturing, assembly, we already talked to that. Talk to our special teams players. All of these teams have different ways they go about doing that. We're calling that our playbook. Ownership has a playbook, management has a playbook. There's different playbooks. We're going to look at a few playbooks now. This is where we're going to get into the design and more the manufacturing. We're going to look at the manufacturing playbook. Julie's going to share a lot of that with you.
I'd like to take a-
Julie:
Five or 10 minute-
Carl Schattke:
Let's take a 10 minute break right now for everybody. I've got quarter 2:00. Let's make it a nine minute break. Come back at five to the hour. All right? You got nine minutes to take your break. Hopefully, that's enough for you. We'll all take a nine minute break and then we'll come back and we'll finish up our presentation. Please be back at exactly five minutes to the hour.
Carl Schattke:
Some of the bullet points you have are small.
Okay. We're going to start back. All right. Thanks for getting back to the room on time and coming back to your desk if you're online. When we had the break, I was talking to some people, and one of the things that was talked about was an FMEA study. That's a fair modes effect analysis study. If you have a process, this is a good way to examine what's going on. You can associate risk factors to cost, the potential for an error, the impact of that error, and then you can brainstorm all the things that could possibly go wrong with what you're trying to build. Look at ways that your process could fail or break down as you're trying to build that.
Then you can multiply those together and come up with a score for each of those items, and then you can come back, and assign a cost to fix that. Sometimes it's really easy to fix a problem. Oh, all we need to do is have them check that they got that data, and send us an email back. Oh, okay, well, now we know they got the data. That was the easy one, right? That's kind of the low-hanging fruit. Sometimes it might be in order to check this, we need to design a million dollars worth of software, and that's pretty hard, and we don't know if that's going to really have a decent ROI on it.
FMEA studies are a good way to take a detailed look at your process, and then mathematically figure out what would be the best things to fix first, and just give everybody on the team an idea of what it is that could improve your process. This does work. I worked at a company where we had a process that had a 50% failure rate. We went in, redesigned the entire process, and we had two consecutive quarters with no failures whatsoever.
It can be a really effective way to changing a process, and improving it so that you don't have problems that you previously had. You can engineer out the things that could be failures. That's the value of the FEMA studies, to figure out those hard to solve problems. All right. We're going ...
Carl Schattke:
For the next part of the session, I'm going to give a PCB 101 that shows the processes that we go through with the fabricator. I'll try to give examples of design concerns or design constraints that are effected by the processes. But that might drag me out too long. I'll try to keep it quick. This intro to printed circuit boards will start with the basics of raw materials and then go through pre-production engineering and the manufacturing steps, which are to process the inner layers, go through lamination, drilling, plating, soldermask, silkscreen, inspection and shipping. You'll see some of the in process inspections that we have such as automated optical inspection for the internal layers after we print and etch the course.
What is a printed circuit board? It's basically ... it's a board that supports components so that we can assemble components and put them together. It also has all the connections between the components. Because printed circuit boards have given us so much more connectivity, we can get a lot more features, and a lot more capabilities on a single device than we used to be able to. When Carl and I started in engineering about the same time, or the same age, and I was actually wire wrapping boards at Hughes Aircraft when I was in college. Now, we're going four lamination cycle printed circuit boards that are actually running the autonomous control systems for automobiles these days. Things have really changed.
A circuit board has the traces, which connect things together, just like wires do in a cable. Copper pads that we can connect components, so that we can solder them to the boards, via holes, which are the through holes or blind and buried vias, which are normally smaller holes, and they're just used for connections. Then we also have plated through holes which vias are also plated through holes. But those would be for standoffs for connector pins, component pins, and things like that. And then there are other features on printed circuit boards. Their IPC defines three classes of printed circuit boards, class I, II, and III, and then aerospace and defense has its own addendum. And then automotive also has an addendum, which it modifies some of the inspection and solder mask requirements, but basically class I is for general electronics products. The rules are not as strict. The annular rings don't need to be as big around the pads. The boards are not as high liability, so not quite as much copper.
Julie:
The boards are not as high reliability, so not quite as much copper. And if the board fails in 30 days and that's what the warranty on the part is, you're fine.
Now, class 2 is for dedicated service electronics products. Think photocopiers, machines that need to last a long time. But actually, the majority of all printed circuit boards that we use and we see, except for the consumer products, all the devices that we're using, these computers and everything, are all class 2 boards.
Class 3 boards are high performance. Might also have to withstand a really harsh environment, like vacuum in space or something. I'm an IPC trainer, and one of the key issues of class 3 is that uninterrupted performance is critical. And then there are the other class 3A-
Carl Schattke:
Like life support kind of products, right?
Julie:
Yes, life support. But a lot of medical equipment, but that's not life support.
Carl Schattke:
And safety equipment too.
Julie:
Safety equipment. It's surprising, though, because automobiles are considered so high reliable, and it's very expensive to support full class 3 because of all the extra inspections and all the ... Sometimes the designs can't quite support it. So the thing about class 3 is, to meet a true class 3 design, the designer has to meet the class 3 design rules. And the biggest one that really impacts the design, especially in smaller BGAs, is that the vias have to have bigger annular rings designed around them. The pads have to be larger for us to be able to achieve a full class 3 annular ring, which is 2 mil on the external layers and 1 mil on the internal layers. Because for classes 1, 2, and 3, we don't change our process. The design is changed to give us a bigger pad that's a target for that drill, so if the drill is misregistered a little bit because of process tolerances, material shrinkage, and things like that, we still have annular ring going all the way around the circle. And that is by design, not by process.
IPC also defines rigid board types. This is focusing on rigid boards only. We're not going to go into rigid flex or flex. Type 1, single-printed circuit board. Type 2 is double-sided printed circuit board with vias in it. Multilayer. Type 3 is multilayer, but without blind and buried vias. Type 4 is much more complex. It can have ... Is this a pointer too?
Carl Schattke:
Yes.
Julie:
Okay. Can anybody tell me what kind of via this is here? Is that a blind via or a buried via?
Carl Schattke:
Buried via.
Julie:
Buried. It took me about 50 years to remember this. So I always just remember that buried is buried inside the board. So this is a blind via, right? You can see it from the outside, but it doesn't go through the board.
You can see here, then, we potentially have two types of blind vias. This one's a laser microvia, and this one is a mechanical drill. So this is laser, this is mechanical. And when you're working with your fabricator, we can help you plan the best stackup for the lowest cost, because in this case, we have to laminate these four layers together, drill and plate, but for this hole right here, it's a free lamination. We already have the final lamination done, and we just drill and plate. We don't have to add another lamination cycle.
The units of measurements that we're always talking about will be in thousands of an inch. Micro-inches is thousandth of a mil. And then we also have millimeters. So IPC always specifies millimeters, and then in parentheses, mil. Next slide.
So, raw materials. This is what they look like. We have the fiberglass, the epoxy resin, and the prepreg. The fiberglass actually came out of the textile industry, and that adds support to the materials. The laminate manufacturers take the fiberglass, and they actually roll it into vats of the epoxy, and then bring it back up and they partially cure it. We call that B-stage, because it's not C-stage, which, C stands for cured. That's how I remember it. So what happens is, when we put all these layers together, the cores are already cured, but the B-stage, the epoxy's only partially cured, so it melts, and then it flows in between the copper and the internal layers and everything, and it acts as the glue layers.
So the copper foil goes on both the external layers, and we sandwich prepregs between two layers of copper foil to make the core, which is C-stage, because the laminate manufacturers generally make this. But if we're in a pinch in our prototype shops, we can make our own cores if the lead time is too long from one of our suppliers on a quick turn. So we have basically the prepregs, the copper foil, and the cores.
This is a stackup ... Oopsy, wrong. My goal is to minimize my whoopsies when I give live presentations like this, because I always manage to hit the wrong button. Okay, so as I was talking before, you can see this is a stackup. This is a core here. This is a core here. We're taking your circuit designs, and we're basically etching the circuit designs on those layers, etching the circuit designs on those layers, before we stack everything up.
You can see the different types of material here. We've got the copper on the bottom, prepreg, core, prepreg, core, prepreg. So there are different ... Some customers require two layers of prepreg, because there's some things with regards to reliability and anti-CAF, conductive anodic filament formation, if you have two plies of prepregs. But these days, remember when I said that if we have a half-millimeter pitch part, we need to use laser microvias? Well, laser microvias are only about, maximum, 6 mil normally, which means that we can only tolerate about a 3-1/2 mil dielectric. So a lot of times, if we're using a real small laser microvia, like 4 or 5 mils, we can only stand one ply of prepreg. So this is showing two plies here and one ply here.
Now, one thing that I'd like to point out here is, do you see here we're talking about the copper thicknesses? Let's see. We have half ounce, half ounce, half ounce. And up here, we're showing our thicknesses as half ounces ... Oopsy. This is half ounce, and it's showing as 6/10 of a mil. That's standard for copper. But then see how we get up to here? We've got half ounce copper, but we've got 2 mil finished copper thickness.
A lot of people don't understand that the reason we show either copper foil or final thickness is because after we've laminated all these layers together, then when we drill the holes, we actually use the top and bottom layers as the electrodes for ... They are the anodes in the copper plating tanks. So we actually end up plating parts of the outer layers also. Usually, what we do is, to start plating the holes, we'll do a thin coat of what we call panel plate, where we plate everything. Top, bottom, inside the holes. Then we will go back and we'll put an image on where we protect some of the foil, and we add additional plating where our circuits, our pads, our annular rings go. So that's why we end up with more copper on the outer layers, even though it shows up as a half ounce of foil. A lot of people don't know that. If you already know it, sorry for the repetition. But it's not something intuitive, unless we've explained it to you.
This is just an ... Oops. Shoot. That has a ... I'm going to flip off that. That was just an example of a stackup. But here's another example of a stackup. Try to figure out how many lamination cycles this is. Can somebody tell me?
Carl Schattke:
Two.
Julie:
Two lamination cycles?
Carl Schattke:
Four.
Julie:
Four? Okay. Four is correct. Oops. Because we had to start with these layers here, and drill and plate this hole, and then also drill ... Well, we would drill and drill, and then drill this laser microvia. So that's a laser. So that means that this dielectric has to be pretty thin, right? And then we add material, and then we start doing these laser microvias, add more material, do this laser microvia, add more material, do this and this.
Carl made a good point earlier. In a lot of really advanced HDI boards, we have to kind of niggle our plating processes to achieve plating both in the through holes and the laser microvias. A lot of the HDI boards these days do not have any through holes that are plated. They might have the tooling holes, but they don't have the plated through holes. That usually happens at smaller designs that are using, like, half millimeter pitch or less BGAs.
This is what the material looks like. Remember when we were talking about the difference between standard FR4 and ... This is an anisotropic material.
Carl Schattke:
On that last slide, can you talk to the cost difference between staggered and stacked microvias? And how available that is in the marketplace?
Julie:
This is going to be significantly more expensive than a board that just has this one drill hole, because when we're talking about major processes for a multilayer circuit board, we etch the internal layers, stack up all the layers together, go through lamination, which melts those glue layers, the prepregs, under high temperature and pressure, then we drill and plate. But every time we have another lamination cycle, we have to add material, go back through lamination, drill, plate, and etch. So it's a major cost to add every lamination cycle. So if we have a multilayer board like this and we have to add lamination cycles, every lamination cycle could cost up to 30%, depending on the complexity of the board. So you're going to really significantly increase the cost.
Now, for very high-volume production, where a dime here or five cents there makes a difference, one thing that we can do is half millimeter pitch. We can actually do offset vias instead of stacked vias. And offset microvias like are shown here are actually considered more reliable. They get better test data. But they're also a little bit less expensive, because we don't have to go through the extra plating time to solid copper plate those vias. So when you're working on cost optimization, stackups, really working with your fabricator to find out what the best options are for these kind of designs. Good question, Carl.
Okay, so this just shows the material. These are anisotropic materials, where they have the glass, and then they also are impregnated with that resin system. This is what the prepregs look like. So in the stackups, you'll see different numbers on your prepregs if your supplier gives you those. Basically, what it's doing is it's designating the glass type. The bigger the number, the thicker the glass is. If we've got a big board that we're ... You know, it's 200 mil thick, 10 layer, and we're trying to build up those layers. We're not going to do it with 10 106 layers between each dielectric, or between each copper layer. We're going to use a bigger prepreg.
So that's one of the other things that we have to consider in stackups, is these open weaves are not that good for signal integrity, because the DK is different between the open space and the knuckle of the glass. So you're kind of seeing speed bumps and things like that. So signal integrity engineers, they take that into account. So there's a lot that goes into a stackup. Probably-
Carl Schattke:
That'd be perfect for in between a power and ground plate, right?
Julie:
Yes, it'd be perfect. And the other reason we use this is we have to use 106, it's very high resin content, if we have thick copper layers. We'll see as we get into this, if we have 4 ounce copper layers ... Say after processing, a copper 1 ounce is 1.2 mil thick. So if we have four of those, that's 4.8 mil thick. Well, if we're etching circuits out on layer two, and then we put prepreg against that, we've got 4.8 milli-inches thick of a topographical surface because of the etch out of your copper. And all those spaces have to be filled in with resin from the prepreg. If we don't have enough resin, the glass is going to hit the copper, and that can cause failures, especially at high voltage. So those are some of the things that we take into account when we're creating stackups.
This is just showing what the prepreg looks like. It's this color because in real life it's this color. It has a shelf life. If you get a quick turn quote back and it's an exotic material and they say, "Well, we can build it in five days because it's single lamination or two lamination, but we can't get the material for four weeks," the reason we didn't have every prepreg in stock is because this material has to be kept at controlled humidity, controlled temperature, and it only has a 90 day shelf life. The cores can last forever, because even though they're not in the best packaging, or we could leave open packages of cores, when we prep the cores for etching the internal layers, the first thing we do is go through a micro-etch to get the oxidation off the copper. You'll see that.
When we were talking about those holes in the prepregs, this is spread weave glass. We use this for our high-speed circuits, because the DK doesn't change between the glass. You don't see the knuckles as much.
Of course, C-stage. This just describes that we can get thicknesses of cores anywhere from 2 mil all the way up to 59 mil, because an 062 board was common, so a two-layer board would use an 059 dielectric thick cord half copper, and then after we drill and plate and we get that extra plating on the top, that's how we get our 62 or 63 mil board.
This is what the core looks like. Oopsy. You can see that it's got copper with the yellow in the middle. It's just a sandwich.
Okay. Copper thickness. There are many copper thicknesses. The standards are always 1/2 ounce and 1 ounce. We're also using 1/4 ounce and 3/8 or 1/3 ounce for plated layers on the microvias. Remember, even if we have a microvia layer going from layers one to two, two to three, layer two, even though it's an internal layer, it's still a plated layer. So for us to be able to etch fine circuits in those plated layers, we start with thinner foils.
But we do have to be concerned with the thinner the foil and the lower the tooth, the smaller the peel strength will be. If you're trying to use a big fat-tipped soldering iron to remove a small component, you might accidentally pull the pads off because the peel strength is smaller.
So everything has a tradeoff. Everything has a consequence. Because we're talking mils here, you know, one ... My hair is probably what, 2-1/2 to 3 mil inch in diameter? Everything we change in a circuit board can have another consequence. I say that I spend my whole life splitting hairs or dealing with unintended consequences. And then when you've been thinking like that forever, you approach the whole life that way, and you really tick off your in-laws by asking, "Well, why are you doing it that way when you could do it this way?" They think I'm the biggest busybody, and they say, "Your customers must hate you." It's like, no, I get their permission to tell them that I think there's a better way to do it. So it's really interesting, my personal life versus my professional life. You guys all tend to love me. My family all thinks that I'm a busybody.
So, pre-production engineering. This is specifically for TTM. Because all our customers have proprietary information, we have a system called Dash. It's instructions for loading the gerber files. It's a really easy way to get the gerber files to us. We take that information and pre-CAM does a design rule check using our design guidelines for the factory that we're fabricating for.
Remember, it's like, you know, the clock Big Ben is a huge clock. You're not going to use the same tools and the same technician to repair or maintain Big Ben that you're going to use on your Rolex or your Breitling watch. The same thing is for a printed circuit board job. The circuit board fabricator who tends to do heavy copper, big old honking power planes and things, may not have the equipment, the refined and enabling equipment, that allows it to build the advanced HDI boards with half millimeter pitch BGAs. So we have to pick our sites wisely, because the sites have different tools. They have different DRCs, which overlap, but one fabricator is going to have better capabilities for fine lines and spaces than another fabricator.
Carl Schattke:
Or aspect ratio too. That's sometimes a big differentiator between suppliers.
Julie:
Yes. Good point, because I always say that our best aspect ratio, which is ... Does everybody know what aspect ratio is? Okay, so-
Carl Schattke:
It's a ... Yeah, go ahead and tell them.
Julie:
It's the ratio of the board thickness to the drill diameter. 10 to 1 or less is our preferred aspect ratio. That means at a ratio of 10 to 1, if we have a 100 mil thick printed circuit board, what is the biggest drill that we want to drill in that board?
Carl Schattke:
10 mils.
Julie:
What was that?
Carl Schattke:
10 mils.
Julie:
Correct. So what's-
Carl Schattke:
He said 10 mils.
Julie:
Yeah, and that's correct.
Julie:
Oh, yeah, I'm sorry.
Carl Schattke:
We got online people that need to hear about it.
Julie:
Yeah. That's what aspect ratio is. But that's for mechanically drilled drill holes. Now, remember laser microvias and other blind holes, which we can do controlled depth blind drills using a mechanical controlled depth drill, those, because they're blind, which means that in the plating tanks, the solution can't ... You know, we can't agitate it through the hole like we can in a through hole in a regular board. Then that aspect ratio is 0.8 or 0.75 to 1. Where that's really critical is, as the dielectrics get thicker, those drill diameters get huge, and they start chewing up your real estate under your components.
Carl Schattke:
Yeah, and there are some vendors that specialize in high aspect ratio boards, and they can get up to 30 to 1. But they have special plating tanks and special plating chemistry that allows them to do that, so that the fluids will flow into those holes better, and then they have agitation technologies which move the fluid more into those holes. It's a much more expensive process and wouldn't be used in high volume, but for particular applications where they need that, then there are vendors that support that kind of technology.
Speaker 6:
And we do ... Oops. And we do have sites that support that kind of technology for the back planes, for all the infrastructure and telecom guys. But for our automotive factories, we may not go that high of aspect ratio because we're servicing a different market.
So, pre-production engineering, besides doing the DRC checks, we do the stackup and material allocation, and then we also do the computer-aided manufacturing to program all the machines, the C&C machines, the lamination press cycles, and things like that. We create the drill files, route files, and the electrical test data, depending on whether we're going to use a fixture or use flying probe testing. Then we set up CAD reference files for each layer for AOI, because we're checking our layers by automated optical inspection after etching.
Then we do the pre-production planning, which is the traveler generation. That's the how-to instructions that go with every lot that we run through the shop, and then every operator has to sign off on his part of the production. And then the job gets released to the floor.
Some of the things that we have to do when we tool these is ... You can see this is a standard production 18 by 24 inch panel. You can see how we've step-and-repeated a single design. I don't see any arrays on here, but there could be rails on each of these, or they could've left, like, 5 millimeter clearance on these so that the boards could run through assembly at the end. But this would be pretty small. If I saw this going into production or fabrication and I knew it needed to be assembled, that's one of those times I'd be raising my hand, "Hey, guys, how are you going to put this on your SMT line?" Because this is such a small part, nobody's going to want to load those into SMT by hand. To have to build fixtures and use tooling to put them into tooling gets expensive.
But anyway. The reason we have some keep-outs. You can see that we can't use the whole panel, because IPC requires that we put test coupons on here, because we need to do all these validation checks so that at the end of the day, when we get these out, we're going to microsection these test coupons to make sure that in the cross sections, the axis view, we got the copper thicknesses right, the dielectric thicknesses right, that we got enough hole wall copper plating and things like that.
Then if you have controlled impedances, we don't measure them on the board. We actually do them on test coupons. So one of the ways to make your design as simple as possible is, if you have like five different controlled impedances, test coupons eat up your production panel. So if you're in a high-volume manufacturing, try to limit the number of impedances that you have per layer so the coupons don't get too big, because we have to use longer coupons for more controlled impedances. Or if we have too many controlled impedances, we actually have to add more coupons, and we end up losing space on our production panel.
This is the single and double-sided PCB process flow. You can see just four simple boards, we've got ... 18, 19, 20, one, two, three ... 23 ... Oh, it's right there. I didn't see it. 23 process steps. So there's a lot to create a printed circuit board. This is one of the things that I think is ironic in purchasing, is between contract manufacturing and printed circuit board fabrication, contract manufacturing obviously has quite a few processes, but they don't have as many sequential processes as we do in fabrication. Their biggest problem, I think, is getting a complete kit to the floor. So when we're talking lead times for assembly versus fabrication, we have a lot more steps, and once we get the material, we only have to buy the prepregs and cores and things like that. They have to load a huge kit, but they don't have as many steps as we do. So it almost seems like it should be easier sometimes on a really complicated fabrication design to take up extra time out of assembly than it would be fabrication, but I'm sure a contract manufacturer wouldn't agree with me on that.
This is a multilayer PCB process flow. You can see what we have to do is we add the internal layer steps, because before we can go into lamination to get all those layers together, we have to etch the circuits off out of the internal layers. So these are all the steps in the internal layers, and then we start lamination layup. It means we put all the layers together and go through lamination. We'll see all these processes as we go.
Then we go to the next slide. Now this is if we have sequential lamination. You can see, every time we add a sequential lamination process, we're adding a heck of a lot more steps, because we have to go through all of these steps for the first lamination, and we'll call that a sub-lam or a sub, and then we have to go through the same lamination steps again for the next lamination cycle. And those repeat until we can get back to the final lamination processes of drill, plate, etch, and finish our parts.
This should kind of given you a common sense of idea of whether your board is going to be really complicated. It's always based on what drill structures do you need to fan out your most difficult component?
The internal layer processes. We keep our cores just on shelves. This is what a stock room looks like normally. I've gone through a lot of ... I did this for a while. An auditor. You know, people come in to audit printed circuit boards. It's really difficult when you get one who doesn't know that the cores do not need to be maintained under environmental control like the prepregs do. So know the difference between your cores and your prepregs.
The first thing we do is we actually preclean these, and then we go through a dry film lamination. What dry film is, is it's a photo image of a laminate that it actually adheres to both sides of the core, and then we're going to image your circuit pattern. And then wherever we shine light through that circuit pattern, it bonds that film to the copper on the layers, and then we go through developing, which actually dissolves that dry film so that when we get out of this process ... This is the imaging process. We can use photo imaging. In this case, think of what your image is going to be. Where you have your circuitry, is it going to be clear film or is it going to be black and block the light?
Carl Schattke:
Block the light.
Julie:
Where your circuit is?
Carl Schattke:
Yeah.
Julie:
No, it's the opposite. We-
Carl Schattke:
Yeah. It's called a subtractive process, right?
Julie:
Yes, it's a subtractive process. So what the-
Carl Schattke:
There are additive processes-
Julie:
Yeah, there are, but-
Carl Schattke:
... but the common one is subtractive.
Julie:
The reason I asked is because I was never a photographer. My first hands-on job in printed circuit board shop was, they had lost their ISO certification and I had to go figure out what all the equipment did and tag it. All the tags had been taken off. I didn't understand these processes. So if we understand the film process, we know where we are in a printed circuit board shop.
In this case, we're going to shine light through the film and bond it to the copper, because after we do that ... Here's what it looks like. We're shining light through the openings in the film. And you can see it kind of discolors the dry film. So you can see where our circuits are going to end up, because we're actually going to protect the circuits, dissolve away the rest of the dry film, and then we're going to etch. And wherever the circuits are protected, they stay, and the rest of the copper foil gets removed.
You can see here we've got dry film on top of here. After we go through etch, we're going to remove this. We've protected the copper. Then we strip. We strip the dry film, and then we end up with our copper. Now, in the outer layer process, it's the opposite, because we're plating the circuits up, and we're protecting this base foil here.
So those are some of the critical things that it's good to understand so that you understand why fabricators say, "Well, you've got 2 ounce of copper, which is really thick. We need to make these lines bigger, because if the lines are too small in too deep of copper, we can't protect them long enough to etch, and they just all etch away."
Okay, so after we do our cores, we do what's called post-etch punch, where these are actually the tooling holes for registration when we go into lamination. We prefer to do post-etch punch because the material does move a little bit after going through etching, which is a high-temperature chemistry process that actually eats away the copper.
Then we go to automated optical inspection, which, it compares the gerber file design to the circuitry on the sides of the circuit board. It does top and bottom. And then the machine actually puts a blot of kind of an inky powder wherever there's a mistake, and then it goes to a second station for an-
Julie:
... there, wherever there's a mistake, and then it goes to a second station for an operator, a live operator, to actually determine whether it's really a quality problem, or just a dink, or something. In this case, you can see that it can determine/detect the opens, which IPC does allow for class II, that we can weld certain opens, and it detects shorts. A lot of times these are just really easy to flick off with a razor blade, if the operator has a good hand.
You can see my red highlighter is moving all over the place, right, so I have a really hard time flicking away those things with a razor blade. I'd be afraid of dinging the board. When we're talking about yields and design, even well designed boards will still tend to have some percentage loss of yield, and our biggest loss of yield in every circuit board factory, is handling, just because these boards get conveyed through so many processes. They get handled, they get dinged, a corner can nick another corner, if they're in racks and everything. Even if you have a really really well controlled perfect process, handling is a scrap cost for us.
After we've got the cores, we go into an alternative oxide. What this does is it just puts a little bit of tooth in that core. The reason we want tooth is because when we go into lamination, and that epoxy in the prepreg melt, we get more surface area for it to bond to, so that instead of a shiny coper, we've got a little bit of tooth so that we know that these boards aren't going to fall apart later. But when we're talking about high speed circuits, every little bump is going to ruin, is going to potentially add loss to the circuit.
Again, signal integrity and oxide types are a major science. Between our shops, depending on what we're using them for, we do have slightly different oxides, and we have low tooth oxides, and normal tooth oxides, and everything, so in high speed circuits, that's actually one of the things that should go in that 3D model for the simulator that sometimes people don't think about that, and their results come out a little bit off, compared to the actual board.
Oops, where are we? Dry film applied. Okay, so this is just showing what the actual panels look like, starting from that internal layer, again, these are just... then we image, so you can see where we have our circuits that are imaged. Then we develop. We rinse off that dry film. We're protecting the circuits that we want to keep, and we're going to etch the copper away. Then strip.
Strip is one of those things we develop a lot. We develop solder mask, and we develop drive film. We strip. Strip means to take any surface protectant off, so we strip dry film, and then later you'll see that we also use tin as a resist. Resist and strip are kind of two good things to know in printed circuit board fabrication. Anything that, a coating that resists is protecting what's underneath it from the future process. Then when we need to remove that after we're done with the process, we strip it. You'll see resistance striping used in different processes for different materials.
Then we do the post etch punch. You can see we do a lot of safety precautions. When you look at these... when we look at these holes, usually there's one offset so somebody can't accidentally turn them 180 degrees out when they go into lay up. When we are talking about all kinds of protections and everything, there's so much good engineering that goes into all these processes to make sure that something isn't set up right, or incorrectly. So automated optical inspection, you can see opens and shorts, and that's what it looks like. How many of you have been to a printed circuit board fabrication site and toured the facility? Okay, good, more than half of you.
Julie:
When you were there, could you really figure out what they were doing? You see the equipment, but it's hard to real... unless you understand these processes, it's kind of hard, at least it's hard for me, maybe because I'm a girl, but it's hard for me to figure out what in the heck they're doing those processes until I learned these. That's why I think these are good presentations.
Okay, so lay up is just where we put all the layers together. Because lamination cycle takes about four hours from start to finish, so we don't just laminate one press, or I'm sorry, one panel a a time. We actually do a set thickness of panels. We have the thermal and pressure profiles set for those thicknesses for that class of material. If we're only running a one panel job, we'll really run two in case one of them scraps, and then we'll fill it up with dummies to get to that height of thickness to make some consistency and the press cycle. A lot of times you'll see six up in one, we call it a book then, and so we'll have the big platen, which are half an inch thick stainless steel call plates, and that's with the press puts down to laminate the material. Against the platens, we might have like six panels with thin aluminum separator plates for good heat transfer between the panels.
Then besides having books with multiple panels in them, most presses also have multiple openings. We can see that this one has four openings, but we only have things in three of the openings right now. Then we're putting the fourth one in.
This is one of those things that I thought was really fascinating. These books, and these call plates, the call plates are half inch thick stainless steel. They weigh like 60 pounds each, or something. Operators used to have to hand carry these things, so people come up with conveyors, and these are just balls, so that it's easy to convey. This didn't used to be in circuit board shops when I started, so these are all kinds of automated, or electromechanical solutions that make life easier for operators. Then the door shuts.
This is what happens, we have hot pressure, and high temperature. If you're familiar with the led free profile, the temperature actually kind of looks like a lead free profile. The laminate, the material suppliers give that to use.
Then after we go through lamination, which is approximately four hours, like I said, the prepreg, since its melting, it squeezes out the ends. It squeezes to the outside of the board. The copper is over sized, but you can see just a little bit of that dried prepreg that squeezed out, so we have to go through routing, and we rout the panel and we put round corners on it so those sharp corners don't ding adjacent panels in handling.
Now we go to our outer layer processes. We have all our layers put together. The first thing we do is we have target fiducials inside the layers, and we use an x-ray tooling drill which actually measures how those are offset inside the layers, and then it optimizes and takes the best fit for all of those to figure out where to drill the tooling holes for the drill press, so that the drill puts the drills in the right place, because on the outer layers, when we're drilling, it's just solid copper on top and bottom.
Then we go into a mechanical CNC drilling. Somebody has to preload the cartridges first. That's what they're doing here. Then like Carl had said, for every drill size, every drill type, and everything, there's so many hits. Like maybe a 6th millionth bit, two human hairs diameter, may only be able to go like 300 or 400 drill hits, where a bigger drill might be able to go 1200, or 2000 hits.
We talked about aspect ratio already, and here is a good diagram of it. Aspect ratio is important for our ability to plate, because if we have too small of a hole, and too thick of a part, we're using the top and the bottom as the electrodes, but on an 8, let's say this is an 18 by 24 panel, our clamps are only on one side. If we can't get enough current, we are not going to get enough... we can't get consistent plating all the way to the holes, in the center of those holes, and our plating, theoretically, should be the lowest in the center of a fabrication panel.
In plating, we don't shoot for a range, like 1 mil plus, or 2 mil. A compliant pin manufacturer will give us 2 mil, plus or minus 8/10th of a mil for copper plating in the hole walls for their compliant pins. But really, what we're trying to achieve is that we meet the minimum customer requirement in the worst case, even if it turns out that in other places that are easier to plate, like bigger holes, that we have a little bit more extra plating in those holes.
Dry film application, now we've drilled our holes. We forgot a process. We have to go through scrubbing to get any deburring out of those holes, but then we go back into the dry film application, and we're doing a similar thing. Hang on, did we do electroless? I'm sorry. I was talking about aspect ratio. I forgot electroless plating copper. You can see inside the holes, we've got metal, glass resin metal, glass resin, these are internal layers. Electrolytic copper doesn't stick to the resin. What we have to do is we go in through electroless plating, which has palladium, which sticks and metalizes everything, so we put a real thin seed coat of plating on all the holes through the electroless plating process.
This is really interesting because you have to get really good consistent plating on everything inside those holes, and it might only be 40 to 60 micro inches thick, but those plating lines, if you go to a good electroless plating line, you'll see that it stages through lines, and the panels, they'll go in, and then they'll come up, and then they'll get vibrated in the tanks. They will also get swished back and forth at a set angle that is optimized to get the most bubbles out of the holes, and the most chemistry solution flowing through the holes. Then they'll also drop them. They'll raise them, and then they'll drop the whole panel inside the tank. So there are different ways of agitating and shocking to make sure that no bubbles are entrapped, because if we get a bubble in here, we're not metalized, then the hole won't get metalized through the electroless copper plating.
So, we've gone through the electrola... I'm sorry, through the electrolytic copper plating. When we're doing our final copper on the outer layers, not only do we want to get that 8/10ths, or 1 mil finish copper plating in the hole walls, we're also going to plate our outer layers. Instead of plating the complete outer layer of thick copper, and then having to etch through a thicker copper, we're going to partially plate it, which is panel plate, and then we're going to protect base foil, only etch our circuits up, so then when we go through etch, we're only having to etch through the thickness of the base foil, not the base foil plus all the plating. That allows us to do finer lines and spaces on those outer layers.
This is the same thing. This is what an electrolytic copper plating line looks like.
Julie:
Pardon me?
Carl Schattke:
A really big one.
Julie:
Yeah, this is a big one. This is a pal line. We have a lot of these. A lot of your smaller shops, you're using some NPI shops, they won't have these because they don't need it for the volume, but you can see how it can run this many one, seven? Is that six or seven? Seven panels at one time.
We've got the copper, electrolytic copper plating, the panel plate. We still have the dry film on here protecting the base foil. Then we go into tin plate. The tin is going to stick to the exposed copper. The tin actually acts as the etch resist when we go to etch the board, the final circuit. We go into tin plating. Ut-oh. This guy's stuck. My presentation's stuck.
Carl Schattke:
Try click on the screen. It gives you your right click on the screen.
Julie:
Okay, there. When we did the internal layers, it's called develop, etch, strip. That's all one machine. In this case, it's strip etch strip, because we've gone through plating, but we have to strip the dry film, then we... so here's the dry film. We strip it, then we go through copper etch, and then we have to strip the tin.
Then we go through automated optical inspection again, for opens and shorts, to verify that our circuits are correct. To tell you the truth, a lot of times, I would say some fabricators, if they're limited on their AOI machines, they won't do outer layers because they're going to do 100% electrical tests, and those are easy to see visually. Then if they get an open or short on the external layers, then they can just see it visually without having to go to AOI. But we are doing 100% AOI now, but I don't think 10 years ago, we may not have been.
Hello? Do you want to come up to the microphone so others can hear you, please.
Carl Schattke:
Which one is better, vertical or horizontal plating line?
Julie:
I'm not sure, because I think it depends on which process it is. If we're talking laser microvias versus plated through holes, we actually... I'm trying to picture-
Carl Schattke:
So-
Julie:
Do you know this Carl?
Carl Schattke:
... versicle or horizontal is going to, if you use versicle or horizontal, the question is do you want the fluids to be able to get in and out of the holes, or do you want it to sit there. They're going to change the process based on what's going to work best. It's a depends kind of answer.
Julie:
That's a really good point, because we were talking about our F circuits, and we want good clean etch. When we go through etch, we do both the top and the bottom of the circuits at the same time. Etchant solution can kind of pool on the top side, while the gravity pulls it down on the bottom side. One of the improvements that started over in high volume manufacturing in Chine is vacuum etch. That solution on the top side, as it goes through the etcher, gets sucked up. We're starting to migrate these, import these machines here in our volume shops also.
It's interesting, that's one of the solutions, like the difference between horizontal and vertical plating, the people continue to design equipment that makes processes better, which is one of the reasons I love engineering so much, because there's just so much innovation in the world, and it's amazing what we do in this industry.
After we have etched our circuits, we go into solder mask coat. We have to clean that copper first, and then we just, here in North America, we screen print it on through a screen. The thickness of the screen sets the thickness of the solder mask. Solder mask is a thick goopy paint. Solder mask rooms always end up dirty after a while, because the stuff just gets everywhere. Green, it's a liquid photo imageable product, again, which means that when you shine light through it, it bonds it to the surfaces underneath.
Solder mask goes on both sides, the whole board, and then we go back, we image the pattern that we want to leave stuck to the board, either through a film or laser direct imaging, and then we develop it off. Similar to how we do dry film to protect our circuits during etch.
Carl Schattke:
Julie, what would be a typical resolution for that process? Like how accurate is it?
Julie:
In etch, our imaging is probably only accurate to maybe a quarter of a mil. When I first started here, and I was creating stack ups, and I had controlled impedances, I was like niggling the line widths down to get as dialed into 100 ohms as I could, but there's actually some resolution on this kind of equipment. That's a really good question.
One of the things that Carl brought up is there's so many advantages to laser direct imaging because we can get better registration, we're not using film, and film, remember, as your shining light through it, it is going to spread a little bit, right, so we don't get quite as good of accuracy with film. Film tends to shrink, or expand, under thermal conditions. We can get tighter tolerances using laser direct imaging for our dry film, and for our solder mask.
Initially, it wasn't really that applicable to standard high volume printed circuit boards, because when these lasers came out, they were like a million dollars a piece, and they were taking longer to image than just shinning light through a film for a few second. Now as circuits are getting more advanced, and things are finer pitched and everything, we're having to continue to invest in capital equipment, to add more laser direct imaging capability to our legacy high volume shops that didn't use it 10 years ago. In this case, and these are... it's hard to do it quick, but when you go back, you can see-
Carl Schattke:
I think we've got about 10 minutes.
Julie:
Oh, we've got about 10 minutes. Okay, we're almost done. Basically, you understand the imaging and developing process is the same as for dry film. Okay, then we put our legend/silkscreen on. Similar process, or we use direct imagers, kind of like inkjet printers. Those take a long time, but one of the things about them is that they're accurate, and we can also individually serialize panels and things like that. We're actually, where we were not using inkjets for anything but NPIs over in our China production plants, we're actually starting to integrate it because as we get into all these high volume, high reliability electronics, we need to make sure that we can manage our lots, and that we've got traceability back to components for the bad times when there is some sort of quality concern, or we need to look something up.
One of the ways that we do present, or protect both ourselves and our customers is no matter how big an order is, even if it's all supposed to ship in the same month, we're going to break it up into a smaller panel lot, and maintain separate lot numbers. Let's say one board failed out of a lot, and we found out that a drill broke, and that that's why that board failed, then we know that we can contain that lot and that we only have maybe one or two more panels, if they were drilled at the same... drilled on top of each other, and we don't need to worry about all the other lots from that same date code. We do that for recording, and also for customer and factory liability. The inkjet printers help us a lot with that kind of recording.
Then we have to apply the surface finish. We hardly ever use hot air solder level anymore, although I think some of you in the military here are still using it. Electroless nickel immersion gold, this just shows the different kinds of lines. Immersion silver, this is a good one that we use a lot for the high speed circuits, because as we talked before, the nickel and immersion gold has ferromagnetic properties. Then OSP, which is copper, it's an organic solderability preservative. It's kind of copper based, and it provides a really really good solder to copper bond.
Then final fabrication, we depanelize the boards into the form that we are going to ship to you, whether it's a big board with clearances on the edges, or whether it's multiple ups, smaller boards with rails on two sides, or four sides, to go through assembly. Then we do flying protesting. We do electrical tests, flying probe, or we could do fixtures.
One of the things as we get into smaller and smaller components, we do have minimum rules for the fixtures. The pins can only be so close together, and the pad openings can only be so small. If we see a solder mask opening on a test pad in high volume in China, we're going to come back and say, "You know what, we need a bigger pad to be able to hit it for that test fixture." There are tons of design rules that all have a good reason for why they're that design rule in the first place. It's because of our capability. Then we do a final inspection, package, shipping, and ship the parts.
Carl Schattke:
Awesome. Thank you. That's the big playbook for manufacturing. Then we have different kinds of plays that we run. High voltage boards. A high voltage board, what are your chief concerns? The dielectric between the layers? The voltage across the surface? We have a clearance between one copper piece and another, and then we are also concerned about creepage clearances as well, and that would be the distance across the surface. We can increase that distance by routing slots into the boards.
Sometimes you'll see a high voltage board and you'll have a lot of slots in it, that's to improve the creepage distance, because the arcing will take place across the surface, but not as much because the air is a low dielectric than the substrate that it's bounded on. We can have contaminants on the surface, so we put slots in to make high voltage boards work.
High speed boards are typically finer pitches, lighter weight copper, so that we can have finer line details. We're not dealing with a lot of high current on high speed boards, we're typically interested in the signal integrity, the flatness of the board, that kind of thing, to make it work.
High functionality boards, some boards are going to combine technology. We might have a board that has a controller on it, but it also needs to deliver a lot of current for an application. Those boards can get pretty tricky, and they have their own challenges with plating. Make sure it's heavy enough to carry the current, but light enough to be able to carry the kind of circuits, so a real fine pitch part can't be heavy copper. You need to be aware of what your manufacturer's capable of. If you have a real fine pitch part, that should not be on heavy copper. It needs to be on like half ounce, or quarter ounce copper, and then built up a little bit from there. You're not going to start with a two or three ounce copper if you have a very fine micro pitch part, sub .5 millimeter.
Then high density products are going to be another type of play that we would run. High density, in order to increase the density, we need to change how close parts are mounted on the board to each other. As we do that, then we need to change the technology and how we connect to those, because we lose the ability... if we want to put all the parts next to each other, that means there's no room to escape and put a [via 03:00:21], right? That means we're going to change the technology to [viapad 03:00:24], or there might be... that might indicate, like on the BGA that we go to, a microvia that's going to be able to drop down to the plane.
We're going to change the board technology to be able to increase the density. It's going to come with a cost increase, but it might decrease the cost somewhat by making it smaller, or it might make a product possible. If you think of your phone, or a watch, or something like that, it's going to have the very highest density, and they try to pack as much functionality in it as they can, but they're going to be using a very advanced kind of board technology for that.
When we talk about high value, that's about getting the most value out of what you're trying to do for the application. Value is going to be based on a lot of different factors, not just cost. The value might be in having the portability, or it might be in having the functionality. We're always designing plays for high value, but whatever play we run, it has to work. At the end of the day, we're engineers. We build stuff, we make stuff. It has to work for who's going to be using it, and we want to find the optimal way to do that, to increase our profitability, increase the longevity of that product in the customers hands, and increase the functionality for the customer, increase the reliability for the customer, increase the... We want to build better stuff, so that it's going to be a more profitable product, it's going to result in higher shareholder value as we go through that.
We also design for different field conditions. Indoor products need things that are different than outdoor. We have cold, we have heat, vibration, shock, whether it's in space, whether it might need clean room conditions because of the fine geometries. It might go into cosmic environment, like I described earlier, so we always want to design with an awareness of where the boards are going to be used, and we need to research that enough. If we start a project, we should definitely know what areas that board is going to be utilized in. Our field conditions are very important for us to come up with a winning game plan.
Then there's clock management. Our project managers are going to be checking schedules and aligning priorities following the critical path to make sure the projects can be built as quickly as possible. They need to look at how long different things take, and arrange the schedule so that things can be built as quickly as possible. This is not only on the design side, but it's on the manufacturing side. A lot of moving parts. A lot of details have to get transferred between the different people working on stuff, so we want to make sure that we do that effectively. That's what skilled engineering's about.
Then we score ourselves on what we measure. We measure cost. We measure lead times. We measure repair rates, service time, downtime, profitability, ROI, ROP, return on productivity, on reliability, anytime between failures. There's a lot of ways that we're scored, but ultimately the industry's going to score us.
What are the penalties on our games? Well, late delivery is a big penalty. Shipping errors. Custom hold ups, "Hey, I don't know how this is done." Crashes, fires, earthquakes, typhoons can change our supply chain. Fails, and DRCs not passing regulations, those are all bad stuff.
Our team concept is winning teams cooperate, they learn from each other, they learn what people do. Losing teams have a silo mindset. They stay in one area. It's more about me than we. They don't help each other out. I hope you're able to go back to your team, bring up some of... take some of the things that you learned today, and hopefully your team's more of a winning team, and you improve your processes back, when you get back to where you came from. We hope that what we worked with and taught you today adds benefit to what you're doing. If you have questions, feel free to reach out to us. Always reach out to your suppliers. If you're a supplier, reach out to people that you're working with, because that communication loop, at the end of the day, is really what closes out and makes it great. Yeah. I think we're going to end it there.
Julie:
Yeah. I think.
Carl Schattke:
Yeah, thank you very much for your time. All right.
Julie:
Okay.
About Author
Related Resources
Table of Contents
<ul><li class="b-hide__item"><a href="#transcript">Transcript:</a></li></ul>
Take advantage of the world's
most trusted PCB design system.
One interface. One data
model. Endless possibilities.
Effortlessly collaborate with
mechanical designers.
The world's most trusted
PCB design platform
Best in class interactive
routing
View License Options
Back
Thank you, you are now subscribed to updates.
