Cómo diseñar para evitar solicitudes de modificación

Chris Church
|  Creado: February 3, 2022  |  Actualizado: August 6, 2022

Tabla de contenido

<ul><li class="b-hide__item"><a href="#transcript">Transcript:</a></li></ul>

Como diseñador de productos, tú puedes evitar solicitudes de modificación que podrían repercutir en la producción o en los costos cuando estos pasen a la fase de fabricación. A continuación, trataremos el tema desde la perspectiva de un fabricante por contrato. Además, ofreceremos algunas propuestas que vemos habitualmente y que te serán muy útiles para racionalizar el proceso de fabricación.

Aspectos destacados:

  • El uso de un diseño eficiente para evitar que los cambios repercutan en los costes de producción
  • Los distintos tipos de solicitudes de cambios que se producen en el proceso de diseño y fabricación
  • La cuestión de la disponibilidad de los materiales y cómo afecta a la fabricación
  • Diseñar con vistas a una prueba
  • La importancia de la documentación

Recursos adicionales:


Chris Church:
So thanks for joining me today. We're going to talk about designing to prevent change orders, how you as a product designer can prevent changes from impacting the production and cost of your product once it's gone into manufacturing. We're going to talk about that from the contract manufacturer's perspective and give you some insights that we see on a very common basis that really can cause some issues with your product. Now, a little bit about myself, my name is Chris Church. I am the chief product officer at MacroFab. At MacroFab, we are a cloud electronics manufacturing platform. We work with 2,000 customers in the industrial controls, robotics, and automation area. We help them with prototype to mid volume production across 75 factories in North America. Myself, I have started robotics software and manufacturing companies. And for the past 12 years, I've either been involved on the customer side, the OEM side, manufacturing products or on the contract manufacturer side.

Now, our agenda today is that we are going to talk about the engineering change orders and their impact on your production. Then we're going to dive into three key areas that drive change orders at the contract manufacturer. Those are material availability, design for manufacturability, assembly, and test, and then finally documentation. A quick little primer on change orders as we're referring to them here. So we're really looking at these from the contract manufacturer perspective. In product design, there can be a number of change orders throughout the process. But the ones we're really going to concern ourselves with here today are the ones that are issued in conjunction with the contract manufacturer. Now, these are typically not required when we're in the quoting phase or the early product discovery. But once we go past the quoting, the price has been set any changes to the process or the materials, et cetera, typically we'll have a cost impact and a lead time impact.

And they'll require typically what we call an engineering change notice to be accepted, and then following that with an engineering change order. So the right way to think about this from the context of our presentation today is really around once that quote has been accepted, we want to that all the changes have already been locked in. And we're going to talk about some things that we can do as a designer to really reduce those from being issued after the quoting process. The first thing we have to think about here is where the primary stage drivers are. From the contract manufacturer's perspective, there are three stages in production where we tend to see these change orders. The first is in sourcing. And these are usually triggered by unavailable or long lead time components or even poor vendor selection or process specification.

And what I mean there is we could see, for example, an injection molding component or material that could, for example, be not meeting our quality specifications or they are unable to produce in the time necessary or the cost of changing as they're learning more about the design. Now, in sourcing the primary impact to the product is going to be on cost, but there are pretty significant lead time impacts. And we're seeing a lot of that here in 2021 with supply chain issues where components that we're designing around can suddenly go out of stock and come in with multi-year lead times. And we have to go back and find alternative components, which can really impact the lead time of that product. Then as we've gone through sourcing and the first production capabilities here, we're thinking about the first article.

This is in the new product intro phase. And this is where the manufacturer is really trying to put your process and your design to test and verify that everything can be built as specified. Really, the big challenges we're going to see here are around incorrect form, fit or function of components in that design, incomplete or missing documentation of requirements or significant challenges with the tooling or the assembly process. And one of the big impacts here of change orders in the NPI process are a significant impact to cost is what we see. A lot of times we will believe that a product takes 30 seconds minute to assemble and validate. And then we find out in the NPI process where we're doing time studies that it's really like four or five minutes greatly increasing the cost.

Additionally, we can see new tooling required here, major changes to tooling, which could even cause the existing tooling to be scrapped and started over with. Those are all very significant on the cost, and we want to find ways to avoid those. Then finally when we're in production, whether we're ramping up or in final volume production, we're going to see issues around the failure to scale the process. That is to say what worked at a small scale in NPI or in early production runs is just not really keeping up as we move it to a volume production. Another big challenge we'll see there is an inability to meet quality targets. This is where we see changes to the process or the design because we simply aren't getting the output we're looking for, the quality output. There's too much lost in the process. And typically here the cost is often not impacted as much as the throughput of the process.

So these change orders can greatly impact our ability to produce that the volumes require. When we look at 2021 the products that we've been working with through MacroFab, the two primary issues that tend to trigger ECOs for us have been either material issues or design for manufacturing, assembly, or test. And we'll see even the material issues in 2021 have been very, very significant for us. They've been nearly 50% of the ECOs we've seen for products in production. And we think about what can we as design engineers do to manage and control for these issues. The right way to think about this is to look at two particular axis. One is what we can plan for as design engineers, and two is what we can control. So obviously some things are completely out of our control in planning ability as a design engineer. And that's things like the ability to execute the process at the manufacturing floor. That is really fully in the control of the manufacturer. And things like tooling and fixture reliability, you really can't know a lot of those until you're actually running and testing these.

We can plan for a few things, but a lot of times our designs for tooling and fixture once they actually get in the hands of people working with them, they will find all sorts of challenges in those that we didn't plan for. But product specification changes, that is definitely in the control of the product design group or the product owners, but you really can't plan for that. So what we're going to look at today are three key areas. One is material availability. While we have very little control over that, we can plan and do things to compensate for the impact on our product. And then looking at designing for manufacturability and documentation, those are very well in the areas of both planning and control. Those are something that we can do upfront and we can have control over those.

Let's dive first into material availability. This is a big issue for everyone in 2021. We've all seen the supply chain issues out there, and we're seeing a lot of impacts across our customer base. Now, as I mentioned previously, it is one of the least controllable things we can deal with. But that's countered by the fact that most of the products we see at NPI have less than 5% of their build of materials with allowed substitutes. That means in a lot of cases the design engineers are either not having or not taking the opportunity to provide some alternatives to the supply chain teams allowing us to respond to those. However, one thing we tend to think about here is that once we've quoted something, we have those materials, we're fine, they can be gotten, there's no issues.

And that may have been the case in the past. But one of the things we're seeing today is even once you have purchase orders in for materials, that does not guarantee their availability. We're seeing material orders get canceled before delivery. That means the order was placed with a distributor, that order was placed with a manufacturer, everybody committed to producing that. But for whatever reason before that product is delivered that order actually gets canceled on us. And that is something we have to be prepared for going out there to be able to respond to that very quickly. The other thing that we have seen in a few cases, and this is rare, but it has happened is where we put in a purchase order for materials. Everybody accepts it, we're waiting for delivery, and now the price is changing.

One thing we don't tend to think about a lot is as a contract manufacturer we can't just authorize new prices for materials. When the material pricing changes, we typically have to go back to the customer and get that approved. Now, one of the things that we're seeing here is how often material design or material issues impact a product that's already in production. Historically, that's something we've seen very, very rarely. But in 2021 due to supply chain issues I think everyone is aware, at MacroFab, we've seen a 6X increase in those post-production redesigns due to material unavailability. This was a product that was either in production in previous years or had been in production in 2021, and then we were suddenly no longer able to get a core material for it. And what we want to do is think about how do we as a product designer, what are the tools in our tool shed here to help prevent these things from happening?

So one of the things that we want to think about are increasing our substitute coverage and our blanket PPV coverage. Now, what do I mean by blanket PPV? One of the things we think about a lot in manufacturing is purchase price variance. That is to say, what did we quote that material at, and what were we able to actually purchase it at? Historically, there hasn't been a lot of purchase price variance. If we are able to quote a part at 5 cents, we're typically able to purchase it at 5 cents. In today's market however, the lack of material availability and high competition for some materials means that the pricing on those materials can change on a day-by-day basis and the distributors and manufacturers of those components are not willing to lock in that pricing well in advance.

So what can happen is when that material goes to get purchased, the price can suddenly increase. So one of the things that we can do during the quoting phase is we can, between our purchasing teams and our design teams, we can agree to a certain level of what we call blanket PPV coverage. Now, again, this is typically on the supply chain team, a little less on the designer. But one of the tools right there available at the quoting phase is to offer a 3 to 5%, what we call PPV coverage. That says for certain parts of the build of materials you'll automatically approve a 3 to 5% price increase based on market availability. But as a designer, one of the key things that we have in our tool set here is to provide alternatives for things with form-fit-function equivalence. I took the time to, as part of this presentation to go back and look at about 500 of the production RFQs that we processed in 2021.

These are request for quotes, for products that are ready to go into volume production. We saw 75% of them had no BoM alternates at all. For those that we approved for quoting and had no BoM alternates, on 85% of those, we had to go back to the design team and ask for alternatives because materials were unable. And then finally of the quotes that were approved, we saw that 28% of those required an ECO for pricing or new alternatives due to unavailability during purchasing your NPI. So we know that right off the bat the more alternatives we can provide in our build of materials, the fewer the possibilities we have for ECOs. And one of the questions you might ask is how do we handle those alternates and pricing? Typically, there are a number of strategies. And it will be different based on how the supply chain team, what their general policy is on accepting this from contract manufacturers and how that contract manufacturer works.

But typically, you'll see either one of these three outcomes. So if you have multiple alternates in your build of materials, if one of those is low priced and highly available, the price of that component will typically be used in the quote. If very few of the components are readily available, typically one of the highest price components are readily available, what you'll see is the typical price for that bottom line item will be the highest component. Alternatively, we'll see average pricing across the alternative component set or in some cases, and we see less of this today, this was more common in the past is the build will be priced on the availability of the individual components in that set, aggregated, combined together. And one final price will come out of that. That is to say if I'm producing 10,000 units and I can get 2,000 of one component, 4,000 of another, and 4,000 of the final, we'll use the pricing of the actually purchase components.

Today, we'll typically see most common either highest price or average price simply because of the availability issues we often see. Other things that we can do as a designer is really considering families of active components. We see in a lot of cases that people design to the maximum capabilities. That is to say if I know that a particular, like here I'm showing an STM32F3. If I know that a particular chip that I'm around has 64K of flash memory, I might write my firmware a little sloppier and take full advantage of all of that 64K. But if I'm able to fit my feature set and to fit in any of that family, what I can then do is now pick from any of those. And maybe I could do cross compilation, et cetera. But that's one thing we can do to guarantee or to increase the probability that we can build our device.

Another common strategy, and we've seen this even in the automotive industry is degrade functionality to keep product flowing. That is to say if there are two variants of a chip, one has an optional feature that I like to have but it's not critical to my product, I design around that. And when that particular functionality is not available, that particular component is not available, I can degrade the performance. Now, I might have to sell it a slightly lower cost out there in the market, but having that can help keep product flowing. And like I say, we see this in automotive industries where they've actually left out certain circuit boards or certain features off cars to keep it being produced. We can see this in other products as well.

If that is available to your product owner and your team, take full advantage of it. Another thing that we see, and again this is when the product allows for it because, again, not every product that we have all of the tools available to us. If we have two different ways of achieving a particular set of functionality via two different sets of components, if we have the room and our design supports it, we can create two different networks and bridge the correct one in using a zero-on resistor where available. And that allows us to have pre-approved design variants that have different population lists based on which components are available. And those will typically be done for things where we have something like a variable power array or something like that. Something where we can actually switch that functionality on and off through populating components on our board. This is actually a fairly common thing that we see as a response, and it's typically around high power components, stuff like that.

Another thing that you can do is consider when there are two parts that have the same function but a different or compatible footprint, we can think about designing the product to support either of them. So on the right we see an image of a dual footprint, in this case it's an ATmega component where the MLF64 footprint sits inside of a TQFP64, TQFP64. In this particular case at build time, we can choose between either component which is available where that's possible. Now, this is not a very uncommon thing. Texas Instruments even talks about this and has an excellent paper, which we have the link here on the screen about how to fit different footprints within each other for families of components.

This is an excellent tool where you can do it, it gives you a lot of design flexibility. But it is complex, it is harder to lay that design out. And you do have to watch out for any issues with signal trace length, anything like that that might create a challenge for you. Now, we talked about electrical BoM items, but what about the mechanical? It's real easy to think from a PCB design, but a lot of our challenges also exist in mechanical components. And these are things like your screws, your heat sinks, enclosures in particular wirings, panels, buttons, et cetera. One of the key things we want to do here is reduce single sourcing where possible. We just had an issue with a customer a few months ago where there was one manufacturer for their product. It was an injection molded product. And that manufacturer went out of business a week before delivery of their product.

So now they're having to scramble, they're having to get the molds, get those over to another manufacturer and get that process run up as quickly as possible because now it's holding up the entire production. So thinking about having multiple vendors for any particular mechanical component is really valuable. And not every case is it possible, and we'll talk a little bit more later during the design for manufacturability perspective about the importance of having good relationships there. As a design engineer, consider how that product is going to be manufactured and assembled as you design it. I can't tell you how many times I've heard, "Hey, yeah, we're having issues putting these things together or getting to go together reliably. I'm having to do a lot of modifications on the bench here."

But I think it'll work better for you guys when you start doing it in production. Well, the answer is as the designer if you are having challenge, you're more prepared to work those out than anyone else. So go ahead and work those out during the design phase. If you're having issues with fitment of two different components, you can work with the manufacturers for those components. But we really need to work those out before we get into the NPI phase, otherwise we're going to have to come back and make some pretty significant changes. And depending on where you are in the tooling development, that could be very significant cost.

We tend to think it's easy to make some changes if we have to in NPI. And if it is about just the process, the sequence, how people are doing that, that is possible. But if it means changing your molds, that could add very significant cost and weeks or even months in some cases to the production timeline for that product. And one of the ways that we can really do this as a design engineer is develop some tooling and fixture prototypes for assembly as you're designing. So thinking about how things are going together, how this circuit board, if it has a very specific way it has to line up inside that enclosure, are there guides in that enclosure to do it? Do we have a fixture? Can we make a jig that helps them do that? Because if we're sitting there working it out by hand and having some trouble as an engineer, the manufacturing process is going to have similar troubles, and they're going to have to develop that tooling. And that will increase the cost if it's not known going into the product.

So thinking about our material availability, there is a strategy we can employ here. The first is to identify our critical components, whether they're electrical build materials or mechanical build materials. Then we need to identify the risk for those components. We can use services like Octopart, Findchips, IHS to look at the historic and present market availability and that pricing volatility. So if this material tends to spike and have big peaks and valleys in its availability, we want know that and have an alternative strategy for that. We want to avoid single source suppliers. So thinking about components which we can, especially mechanical components that we can only get manufactured via one supplier, we want to avoid that where possible. And thinking about the complexity of the product that we're building, how much it relies on those individual components and then develop our response strategy.

That is to say alternate components, alternate suppliers, and design variants as we discussed. Let's move forward now into the design for X topic, that is designed for manufacturing, design for assembly, and design for test. When we talk about design for manufacturing, we're asking how well our design maps to the processes, tools, and machines that are going to be used to produce it. Design for assembly is really about how does our design map to the processes executed by humans or in some cases they have very high volume products, the robots that will do that. And then finally in design for test, how does our design enable itself to be validated? And how does that reflect on the reliability and the outcomes of manufacturing? That is to say, what are the quality rates we're getting out of it? And if we think about where those three key areas tend to drive engineering change orders by production stage, we'll see that design for manufacturing is one of the biggest drivers in NPI.

That is to say we can't reliably get this process working, there's some challenge with that design. And then looking at the ramp up phase, that's where design for tests really starts to show its head. That is to say we're starting to see low yields coming out of production, we're having a lot of quality issue. That's what we tend to see there. And then later when we go into volume production, these three should have less impact because we've worked them out in the NPI and the ramp phases. Let's look at design for manufacturing. One of the key areas we see where challenges come in and from design for manufacturing is around pushing the process limits. That is to say if we're thinking about the operation of the machines, the fabrication of components, et cetera, the closer we get to the tolerances of that process or the particular manufacturer, we can work in, the more likely we are to have issues that are going to impact our ability to successfully manufacture that product.

Now, the challenge here is as an engineer, as a product designer is our tools give us the ability to do lots of things. It doesn't necessarily mean that those will translate to the manufacturing process we're ultimately going to select and the price point we want to hit for that or the throughput, for example. So thinking it from that perspective, we want to work with the different manufacturers that are going to be involved as early as possible in the design of the product. That means if you're doing something non-standard in the PCB fabrication, if there's something very special about your design, work with that PCB fabricator are early on in the design process. Get their feedback, understand what they think is achievable, what they feel they can do, and really listen to them.

If they say, hey, we think this is going to have pretty low yield, take that into guidance and expect that if you do follow down that path you are going to have yield problems with the PCB fabrication. Likewise, the PCB assembler you're going to work with probably has a very specific set of machines that have a specific set of capabilities. This is where we run into two possible outcomes here. One is that if you have a PCB assembler that you are very, very familiar with and you really like working with them, you may alter your design to better fit into their process. Likewise, if you have a product that needs a very specific design, make sure that that assembler can support it or select a different assembler.

And I'll give you one example we see here a lot especially with some of the smaller assemblers is very complex led-free dense BGA designs really going to require a nitrogen or a vapor phase oven process for high yields. If your assembler doesn't have that, then it's probably best that you go to a different assembler there. At high volumes, you're not going to get the yield you're looking for there. And mechanical fabrication, this is really, really important. It's best to work with, especially if you're doing injection molding, machining, work with and get feedback on your design from the fabricator as early as possible in that design.

What you don't want to be doing is coming out there and creating this really great design, you've worked it through, maybe you 3D printed it and then you start hearing over and over again, "Hey, I can't actually produce this at the level you're thinking because our gut response is we'll just try harder, we've done a lot of work. The design is done, we need to get this into production." That's not the place where you want to end up. So work with each of the manufacturers necessary to produce your product, get their feedback early in the design, incorporate it. And they can really help work out what's best based off of their equipment and their processes.

Now, when we talk about design for assembly, these are the things that really go into making that product. And this is a little less about the manufacturing process itself and more about all of the different elements as they come together to bring that final product in place. And one of the challenges that we often have as product designers is we assume that something is easy because we can see it and go, oh, okay, when that happens, I need to do this. And we can do that on the fly because we know more about this product than anyone else. The reality is the typical assembly worker doesn't have the same level of visibility of the product as you do, and they also don't have the same level of control and responsibility.

That is to say if something is not lining up, typically their focus is to be within spec. And when something is out of spec, they're not often given the liberty there to go and alter that process or do something other than the instructions they've been given. And that can cause our yield to drop very, very rapidly. One of the goals we always want to have to do ... One of the goals we have in design for assembly is really to focus on repeatability, how likely is it that we're going to get the same outcome every time we do the same step and to reduce the steps. So thinking as a product designer, what does that mean mean for me. I need to either reduce components or reduce process steps. So any way I can have a component preassembled that takes out one of the manuals steps there, that's something we want to do.

So thinking about having wire harnesses built, having connectors on the board versus say soldering wires directly to a board. Those are things that at volume that can really make a big difference. One of the big things that we often see here is the use of hot glue. A lot of products have hot glue in them because those are things that are used to adjust for problems in the processes or to reduce cost of materials. But the higher the volume they go, the more those can become repeatability challenges. So think about those steps, how do we design a product to reduce those kind of steps in there?

Poka-Yokes are really, really, really critical here. Anywhere a step can go wrong and that can have a critical impact on the output of the product, either its functionality or its yield, think about how you as a designer would put a Poka-Yoke into that process. That is basically a thing that prevents you from making a mistake. So that could be a go, no-go test fit fixture, that could be a piece of tooling or guide that prevents the person doing that step from making a mistake. So think about as you're designing the product, and if you would make a mistake in doing that, how do you prevent someone else from doing that? Start thinking about that at the design level because then a lot of times you can actually work those issues out at the design level and greatly reduce the cost of building that and increase the through throughput.

The other thing to really consider here is the process for building your product at scale. If you're building products in the hundreds of quantities or the tens of quantities, you have a lot more options available to you. You can do all sorts of things you cannot do at high volume. And one of those things that we see very, very often is the use of adhesives or solvents to connect two plastic pieces. That's totally doable. We can take cyanoacrylate or depending on the plastic, maybe some MEK, use that to actually chemically weld two parts together. But when we start thinking about doing that thousands of times over and over or tens of thousands of times, we start to see that, A, it's going to be more problematic because, we have fumes, all those things we have to deal with, those have to be controlled. And secondarily, the opportunity for slop, mess or mistakes in that process can result in a discarded process. That can also be very slow.

So if you're thinking about a product that is being manufactured at volume, consider a process of sonic welding or hot plate welding, which will change your design. If I'm in a low volume here and I'm doing cyanoacrylate to connect to components together, I can have a lot more flexibility in my design there. Whereas if I'm sonic welding, I got to make sure that I have the right fit between the two components that enables them to get a good friction surface and ensure a good weld. And if I'm hot welding, I want a product that can actually fit, or hot plate welding, I want a product that can fit in an automated welding fixture. So we have to think about that at the design phase because once we get into production, it's really too late to make some of those changes. We may not necessarily get the outcomes we're looking for.

And so if we look on the right side here, when we talk about the processes and level of scalability, adhesives and solvents tend to be for a very low scale production. We want to try to eliminate those wherever we can at a higher volume both from a labor perspective and a scrap perspective. Mechanical fasteners, that is screwing two parts of an enclosure together, that works both at low and high volumes. However, unless we're using some sort of robotic process to automate that, it can add significant cost at high volumes. And those things depending on the targets for your product, you may want to work out there. Sonic welding of two plastic components, that can go from mid volume to very high, it's a fast process. Now it is typically a manually operated process.

But the reason why we don't talk about that typically at low volumes is because the cost of getting into it, building the dye, building your horns, et cetera for that can really add some significant non-recurring engineering costs. And typically at lower volumes if you're only going to be producing in the tens to dozens at a time, the adhesive process is going to often be less expensive. Now, hot plate welding is slower than sonic welding on a per unit basis, but it has a benefit in that we can develop fixtures and their automated machinery for this where you can do 10, 15, 20, even 50 units at a time depending on the size and the shape of the product. And that enables that to scale up to very high volume as well. So when we're thinking about an enclosure and the ways to connect it together, we kind of go in that sequence, adhesives, mechanical fasteners, sonic welding, and hot plate welding.

Now, another common assembly outcome here is to protect the board. We have two ways that we see commonly doing that. Those are potting. That is where the board is fully encapsulated in a plastic like setting to protect that board from the environmental and conformal coating. Both of these work very well, although conformal coating does tend to scale much higher. We see a lot of robotic tooling there, a little less so in potting. So when you're thinking about if you have an industrial device that's going to need to be out in the field, conformal coating is something you're going to need. The processes are there. But what we want to think about on that process is how many components need to be protected from the conformal coating. So things like your connectors, microphones, sensors, et cetera, they still need to be exposed to the air. And so the more of those we have on the board, the more we need to think about how that assembly process is going to work.

Are those going to be hand taped? Are we going to need to create specialized fixtures? So those are things that we want to talk about early on in the process in design to make it easier there. In conformal coating, one of the very specific things when we talk about designing to make it easier, having a critical component that needs to be coated right up next to a connector that needs to not be coated is much more challenging than having connectors off by themselves and the critical components in another area on the board. So the more you mix those things, the more cost you add and the more challenges you add in that process. And that's what we mean about being able to think about how you're going to have that product manufactured at the design stage can help reduce those costs and reduce changes.

And any manufacturer you're working with can give you feedback on your design well before you get into the NPI stage. Now, designing for test. This is an area where it is really going to matter, just like in the design for assembly, the volume is critical here. But some things are going to be consistent. We need to make sure that our device is testable and that everything we are going to test can be tested. That means that if there are critical networks in our device, we need to have test points for those. If we're going to be using in-circuit testing, that is a very high volume process, we need to make sure those test points are readily a available and accessible by the in-circuit tester by a fixture.

With flying probe, we have a lot more options. We can bring those probes in at angles, we can touch on the legs of an IC, et cetera. But that's a much slower process than in-circuit testing. It's going to add more time and more cost to the actual manufacturing. But in in-circuit testing, we really need easier access to those, we're going to develop fixtures. And that's where you want to talk to your manufacturer as well as you're designing to understand, do I need my test points on one side of the board or on both sides of the board? How is that going to impact? If I'm doing a high volume production, is two sided or single sided is going to be more effective for my particular product?

Now, at a very, very low volume, you can do things like bench testing. That is where, we've seen this where people hook up. And this actually happens in some real low volume industrial products because trying to do a complex fixture for in-circuit test just doesn't make any sense because you're producing maybe 10 or 20 of these at a time. And this is where someone will actually come in and have ... In the design, you'll want to have hookup points for things like oscilloscopes, multimeter, et cetera. These can be purchased right off the shelf and bake those into the design when you know you're going to do bench testing. That's at a process level.

When we talk about in-circuit testing and flying probe, they're really testing whether or not we've assembled the product correctly, whether or not there are any defects in the surface mount or through hole assembly. But one of the big tests you're going to need to do in many products is a system level test. That is testing complex functionality, giving it inputs and outputs. Some of the in-circuit testers can do this, some of the flying probes can do this. But typically we'll see a fixture level test. And this can be either for the PCBA. And you've probably seen a lot of these where you put the PCB into a device, you push a button, some pin adapters come in and compress on that and then custom software runs to cycle and test that device.

Alternatively, we also see this a lot in devices that have been assembled, and we need to have an input come into that assembled device and then a measurable output coming off of that. Consider in your design how are you going to test that finished good. If it's fully sealed up in an enclosure, how do you validate that the enclosure assembly process didn't damage the product in some way. This is where a lot of people tend to think, hey, if I tested at the PCBA level, we're all good. Well, the reality is a lot could still go wrong in final assembly. At this point, you want to think as you're designing it, how am I going to validate this? Am I going to use fixtures? Are we going to use flying probes in circuit testing? Have I enabled that particular validation process?

Because what'll happen is if you don't do this, as we're getting into that ramp, we're going to start seeing a lot of fallout. We're going to start seeing real low yields or low throughput. And then everyone's going to say, hey, we need to get this back up. And now we got to come back and make some changes to that product at potentially the most expensive point in time to enable us to fully test those kinds of problems in the process. So listen to that feedback from the manufacturer, reach out to them earlier in the design process and talk about what the testing strategy for your product will be given what needs to be tested and the volumes you're looking to produce.

So again, when we think of design for X,, what are our strategies here? So in design for manufacturing, we want to identify our critical vendors. We want to get their input and then scope our design to their recommended processes and specifications. At design for assembly, we want to consider the scale at which we're going to produce our product and design for that. We want to reduce the complexity and manual processes involved in there. As much as possible, we want to reduce the number of steps building this particular product. And we start that process design in parallel with the product design. So as we're designing our product, we have to think, how is this product going to go into manufacturing?

Get input from the manufacturers, hos is this going to be assembled? And adjust that design as we're going rather than creating that perfect design and then coming back and having to modify it as we realize that it can't be manufactured or assembled the way we're thinking. And then under design for test, just to reiterate again, if we'll notice on all of these as designing for our process at scale, identifying that critical functionality as we design and making sure we have a mechanism to test that functionality in process at our current scale using the right tooling for that volume and that test.

Finally, let's go to documentation. Typically, there are four types of documentation we want to provide with any product. Like for example, we may not have packaging instructions if it's not packaged. But in general as we go into production, we want high quality documentation. Those are assembly diagrams that show us how the product should be made, how things go together, the general fit and function, the process documentation, how should we do all of those steps,. The test documentation, what needs to be validated? What are valid outputs, what are valid inputs? And then if they are to be packaged, what is the packaging documentation? How do we put all this together?

What we typically see, this documentation is started after the design is nearly complete or completed. We're going back in time and saying, okay, I've got this product, let me document how to build it. However, what that often ends up with is us making assumptions that, hey, I say to do this, but how do I actually validate that, how do I achieve that? We can look at those assumptions and go, well, they will figure that out. Well, a lot of cases that can't be figured out, it'll end up being a problem area. So documenting as you design really helps to identify where you're going to run into challenges, where there are going to be critical process problems. And let you go ahead and address those earlier in the design phase rather than having to come back and fix that later. Think about what's functionally critical to your PCB design. Do you have a procedure to verify it? Can we access the critical elements to test that?

And if I'm using a fixture, can a fixture be effectively designed? I recall an experience with a product not too long ago where the product had been designed, it had been in manufacture before. There were some yield problems with it. So they decided to add a few additional tests to reduce those yield problems. However, the design didn't change, there were no actual test points to execute those tests. During NPI, it was brought up that, hey, this is going to be very, very difficult to develop a test fixture. That customer had said, hey, let's just move forward, figure it out, go forward from there. Ultimately, the fixture couldn't be reliably developed to test the way they wanted to, and we had to have a design change.

This added a month and a half to the overall production time for that product, and that's something that could have been avoided early on by looking at that critical test and saying, hey, let's go ahead and design in the ability to test that. Now, another thing that we want to do here is if we're not doing a lot of our documentation, if we don't have within our organization a standardized document set that we can pull on that's been through a lot of production cycles, it's okay to ask your contract manufacturer for example documents and what they would like to see and then work with them throughout the process getting their feedback as you're building that. If you have any questions, I will be here for the next half hour or so. Happy to answer any questions you have, just put them there in the chat window and we can talk about them. Thank you.

Sobre el autor / Sobre la autora

Sobre el autor / Sobre la autora

Chris is the co-founder and Chief Product Officer at MacroFab. With over 20 years in technology development and leadership, he has started and led companies in the SaaS, Robotics, and Manufacturing spaces.

Recursos Relacionados

Volver a la Pàgina de Inicio
Thank you, you are now subscribed to updates.