In-House Manufacturing for Electronics Projects

The news media like to embellish the advances and proliferation of affordable 3D printers, sometimes calling them ‘a factory in your home’. While this concept of being able to avoid buying products at a store and building them at home sounds ideal, I personally don’t believe we’re anywhere near it being a reality. I purchased my first 3D Printer in 2012, built my first CNC router in 2007 and purchased my first CO2 laser cutter in 2014, so I have some experience in automated DIY fabrication. 

While we are not at the point of home factories being a reality, I do think we are getting close to having affordable automation in our offices. I say affordable specifically because, given enough money, you can set up a factory in your office, but it’s probably not going to be cost effective. 

Utilizing open source projects and technology makes it very viable for a hobbyist, small business or small division of a larger company with limited resources to build complex jigs, prototype enclosures, and even assembling prototypes or small batches of PCBs. By using these technologies, you can enhance your productivity and work more efficiently.

I feel one of the major issues with the ‘factory in your home’ idea is the lack of electronics manufacturing and assembly capabilities. We are a very long way from the average user being able to make circuit boards that can compete with those coming from a board manufacturer. Just stocking a decent range of passive components required to build electronics takes hundreds or thousands of values of capacitors, resistors, and diodes. However, a serious hobbyist or a business likely already has these components on hand for prototyping.

There are some interesting methods for making PCBs at home, such as printing them with conductive ink using a dispenser like the Voltera V-One, milling them, or chemically etching them. Those methods are fine if you want a single layer board, however, vias and plated through holes for two-sided boards continue to be a challenge for prototype assembly. The V-One is probably the best of those options for plated through holes, as it inserts copper rivets into the drilled holes. The smaller vias can be created by using smaller rivets or filling drilled holes with conductive ink. Nevertheless, the conductive ink still has issues with solder wetting and high resistance. Rivets are popular for homemade boards, but are still relatively large compared to what a commercial board house can offer. I haven’t seen a good solution for boards with four or more layers, and with the low price of prototype quantity boards out of Asia, the only real advantage of making them in-house is turnaround time.

3D Printing

Low-cost 3D printing has come leaps and bounds in the past decade. The industry has moved from the early rep-rap style 3D printers, that spent more time being tinkered with than printing, to ultra-reliable, multi-head, multi-material printers. There’s a multitude of options available for economic, reliable and precise printers on the market. Beyond the filament deposition printers that most people are familiar with, there are desktop resin printers which offer incredibly high-resolution prints that match injection mold quality. Whilst these printers can’t compete with an injection mold on cost-per-part at volume, a high-end printer costs less than even a small injection mold. They also offer more flexibility than the mold, allowing the production of low volume customized enclosures, or one-off fixtures.

Having a 3D printer in the office allows for in-house creation of novel jigs and fixtures, enclosure prototyping, and mechanical fittings production for your electronic products. If you don’t have skills in mechanical design, software like Autodesk’s Fusion360 is a very powerful option that is free for startups and makers. 3D printers are now so easy to use that the largest learning curve involved is in the mechanical design of parts. Once you have learned mechanical CAD software, however, you’ll find endless ways to utilize 3D printing to create parts that help you prototype, reduce lead time in projects, and allow you to deliver a superior product.

If you’re going to be working with injection molded enclosures or parts, using a high-resolution resin printer such as the Form 2 or 3 will give you better results than using a filament-based one. If you are printing jigs and fixtures, a filament printer, such as a Creality Ender-3 Pro, is much cheaper and offers the ability to work with a wider range of materials possessing different mechanical properties. For me, a typical jig is a bed-of-nails test jig to line pogo-pins up with test points on a board, complete with a printed clamp to hold the boards down. Using exports from Altium, this is quick to design in mechanical CAD software and assemble once printed. A printed jig is also much cheaper and more mechanically sound than a custom PCB made to hold the probes, and with an integrated clamp mechanism, allows you to work on the board hands-free. For production stage programming, you can use a small vacuum pump to hold the boards down on the probes for flashing firmware and burn-in testing. The vacuum allows very rapid swapping of boards in a production line.

My co-workers have printed handy jigs to hold cut tape for hand assembling boards, and custom clamps to hold boards as they are worked on. These can be cheaper and more ergonomic than commercial off-the-shelf products, and could solve problems that they can’t. Using clamps built specifically for a board can be more effective than using commercial options, especially for very compact designs that have connectors or edge features which off-the-shelf options would interfere with.

3D printers can also be useful for printing support features used on a PCB. For example, I have printed custom LCD screen mounts or standoffs, button caps, and board guides for sub-boards. When your volume is too small for injection molding, 3D printing can offer a viable option for adding custom features to your product. If your volume does allow custom supports or mounts to be injection molded, a high-resolution 3D print allows you to better check component clearances on real hardware than simulation in CAD does. By doing this printing in-house, you can save lead time that would otherwise have been wasted waiting for sample parts to arrive.

When I bought my first 3D printer, the low-cost market was just starting to offer some pre-assembled options. Back then, the 3D printing community assumed that only the best setup printers would run jobs flawlessly each and every time they were used. Today, a cheap, relatively low-end printer is expected to be highly reliable and rarely fail a print unless due to user error. Software we use to generate G-Code is vastly more powerful and reliable, which significantly contributes to the reliability and quality of prints. When even relatively small prints can take several hours, and a large print might take 30 or 40 hours, reliability is very important. If a print fails on hour 39 of a 40 hour print, it’s not going to help your productivity much. If you are looking to purchase a 3D printer, there are many review websites which will have hands-on tests with the printers they review.

If you don’t already have a 3D printer in your office, the Creality Ender-3 Pro is a very inexpensive, reliable option that has many good reviews. A more expensive printer with dual extruders, such as the pre-assembled Flashforge Creator Pro, might appeal more to those looking to start their first prints within minutes of receiving the printer. For the highest print quality and testing the fit of parts designed to be injection molded, the Formlabs Form 3 dominates the high resolution market segment.

SMT Pick and Place

Pick and Place machines for SMT assembly are typically large and expensive, and consume a lot of power and compressed air. They are designed to be integrated into an assembly line with large reflow ovens and place tens to hundreds of thousands of parts per hour. The current state of the SMT assembly machine industry is not too dissimilar to that of 3D printers 15 years ago. However, like 3D printing, the maker movement is making steady progress towards reliable, cost effective desktop SMT assembly machines. While there are only a few kits available on the market, you can readily source all the parts to build an SMT Pick and Place machine based on open source designs.

There’s a significant gap in the market between hand assembly of a single board and mass production of thousands of boards. If you have parts which are difficult to place by hand, paying to have a production line set up to run just one board, or even a hundred, can be financially impractical.  If you are making a hundred boards, hand placing all the components can be tedious for a small bill of materials, or unrealistic for larger bills with many parts to place. This difficulty with low volume and prototype production is what the open source projects set out to resolve. DIY machines can’t begin to compete with the speed of assembly line machines, nor are they intended to. However, a DIY machine can cost less than a single feeder for a line machine! 

For assembling a single board, or even a hundred boards in one day, placement speed is much less important than ease and speed of setup. For small production runs, iif a machine takes several hours to set up, load components into and begin running a job, it doesn’t matter how fast it can place the parts. The most popular open source SMT Pick and Place machine projects have software that is designed to import and set up a board quickly.

The two main open source projects are LitePlacer and OpenPNP. LitePlacer is a single-head machine kit and associated software, whereas OpenPNP is purely focused on software. OpenPNP can control a LitePlacer, and a wide variety of home-built machine designs. One of the most popular starting points for a dual head machine is Anthony Webb’s machine on Hackaday. There isn’t a kit available for this machine; however, community member Peter Betz offers a fully machined set of plates and head, which are the hardest parts to build yourself. The remaining components are primarily easily sourced extrusions, fasteners, and components used by 3D printers. Both OpenPNP and LitePlacer have thriving communities that help support new users and work collaboratively on challenging features such as low-cost feeders for reels.

As these machines are intended for low volume production, they typically utilize cut tape strips, trays, and tubes of components as a source of parts to populate the boards. The software allows you to teach the machine where these components are, and will automatically detect if there are no more parts available. Automatic tool changers utilizing industry standard heads, such as those from Juki and Fuji, enables the same wide variety of components to be placed as those supported by an assembly line machine.

Just like a line machine, LitePlacer and OpenPNP have up and down facing cameras to optimize component placement. The down-facing camera is attached to the pickup head and is used to identify fiducials on the board to correct for board rotation and position errors. The down-facing camera also identifies components in cut strips and loose parts for pickup. The upwards facing camera recognizes components that have been picked up, to adjust for rotation and the component center position. By using machine vision, a low-cost machine is able to precisely collect and place components. 

OpenPNP and the LitePlacer software both accept standard pick and place exports from Altium to be told component positions. It's relatively simple to set up and configure the placement job in the software. Once the job is set up, the machine will place components without requiring user intervention, allowing you to get on with your day. If you typically have high component count boards, the time taken to build a machine can be quickly recouped through the enhanced productivity as the machine assembles your devices, or the time and cost savings from not sending the board to a contract manufacturer.

I asked Jason von Nieda, the creator of OpenPNP, what drives his work on the project and he told me:

I think it's important that the technology to build things is available to everyone. In software, Open Source compilers and operating systems have created an explosion in the ability for people to program. 3D printers and Open Source modeling software have made it easy to make physical things at home. Building electronics at home is still somewhat difficult because as technology advances the parts get smaller and harder to place. It's important that hobbyists, tinkerers, inventors, DIYers continue to be able to build these things on their own.

I feel this is particularly true in a commercial environment, where design requirements tend to drive for smaller and finer pitch leadless components. Even those with the steadiest hand and best eyesight struggle to place a 0.4mm pitch LQFN package quickly and accurately. If the board is crowded with closely spaced small parts, it can be a challenge to get each of them placed without disturbing adjacent ones. Machines are much better at high precision repetitive tasks than humans are, so why not consider bringing one into your office to help you out?

Reflow

Once you have placed the components, you will need to heat the board to reflow the solder. There are plenty of options for low-cost in house reflow, depending on your budget and requirements.

Hot Air Rework Tool

The most basic way to reflow solder paste on the board is to use a hot air rework tool. By directing hot air to small sections of the board at a time, you can melt the solder section by section. This method works poorly with low-melting-point plastics such as those used in some connectors. It’s also ineffective with components that have large footprints, or boards that have significant thermal mass in the ground plane. It's crude but very cheap and effective for most designs.

Electric Skillet

If you head to the kitchen appliances aisle of your local supermarket or homewares store, you will find flat heated surfaces in the form of electric skillets. Typically, only the area above the heating element will get hot enough to reflow the solder, and there is very little thermal control. Placing your PCB over the heating element and turning the skillet on will typically give you an acceptable ramp up to the point of reflow. If you have a substantial thermal mass on the board, you might need to nudge the reflow over with a hot air rework tool. The skillet is a practical pre-heater for the board, allowing rapid reflow of even large areas using the hot air rework tool. You do need to ensure the solder melts reasonably  quickly once it’s at maximum temperature, or you will evaporate all the flux. Using a skillet is also a very crude yet effective method for the vast majority of designs. If you’re looking at trying a skillet out, ensure its plate is flat, and it has some form of temperature control knob.

Hacked Toaster Oven/Benchtop Oven

Benchtop toaster/pizza ovens have long been favorites of the DIY community to melt solder. There’s a variety of DIY controllers available that will read a thermocouple and use a solid state relay to control the oven’s heating elements. The best-kept secret of the benchtop reflow oven community is the ApolloNG picoReflow controller utilizing a Raspberry Pi and a MAX31855 thermocouple breakout board. This controller goes above and beyond what most others offer with thermal profile control and a web interface. If you’re looking for a budget oven to get perfect results on any board, look for one with both top and bottom elements. The elements should have shields over them, which reduces the amount of radiant heat for more precise control and less melted connectors. Ideally, the oven will also have a small fan built in that can help circulate air to ensure you don’t end up with hot or cold spots. While this solution may look crude, the results can be as good as or better than costly benchtop commercial ovens.

Commercial Ovens

There are benchtop reflow ovens that cost tens of thousands of dollars, but they are not what this article is about. You can find cheap reflow ovens from Asian suppliers on the usual marketplaces. Although these products are built specifically for reflowing PCBs, I have had very little success with their stock configurations. The heating elements generate a great deal of radiant heat that the temperature sensor won’t pick up, and the interior has significant hot and cold spots. Replacing the controller with picoReflow is a significant step to improving these ovens, as is adding some shielding over the elements. The main advantage of these ovens is their significant insulation and drawer that allows smooth insertion and removal of circuit boards. I know many people have had good experiences with these low-cost ovens, however, if you’re not afraid of rewiring a toaster oven, you can save a lot of time and money for a similar or better quality result.

Even though ‘a factory in your home’ may not be a near-term reality for the average person, there are some exciting technologies to explore. Meet like minds online at the Altium forum, or in person at AltiumLive.