For reasons that will remain undisclosed until some time in the future, I recently had a need to panelize a few PCBs. Panelization is the art of taking PCB designs you already have, whether they’re KiCad board files, Eagle board files, or just Gerbers, and turning them into a single collection of PCBs that can be sent off to a fab house.
If you’re still wondering what this means, take a look at the last board you got from OSH Park, Seeed, Itead, or Dirty PCBs. Around the perimeter of your board, you’ll find some rough spots. These are ‘mouse bites’ and tabs, places where the boards are strung together to form a gigantic rectangular panel sent off to a manufacturer. You can check out this great interview with [Laen] from OSH Park to get an idea of how this works, but the basic process is to take a bunch of Gerbers, add tabs and mouse bites, solve the knapsack problem, and send the completed panel off to a board house.
Panelizing boards is something most of us won’t have to do often. Really, you only want a panel of boards when you’re manufacturing something. For small-scale production and prototypes, bare boards will do just fine. Simply by virtue of the fact that panelizing boards is far less common than throwing some Gerbers at OSH Park or Seeed, there aren’t many (good) tutorials, and even fewer (good) tools to do so. This is how you panelize boards quickly and easily using Open Source tools.
One of [CNLohr]’s bigger claims to fame is his process for making glass PCBs. They’re pretty much identical to regular, fiberglass-based PCBs, but [CNLohr] is building circuits on microscope slides. We’ve seen him build a glass PCB LED clock and a Linux Minecraft Ethernet thing, but until now, [CNLohr]’s process of building these glass PCBs hasn’t been covered in the depth required to duplicate these projects.
At the highest level of understanding, [CNLohr]’s glass PCBs really aren’t any different from traditional homebrew PCBs made on copper clad board. There’s a substrate, and a film of copper that is etched away to produce traces and circuits. The devil is in the details, and there are a lot of details for this build. Let’s dig deeper.
The next great advancement in homebrew electronics is an easy way to turn copper clad board into functional circuit boards. This has been done since the 60s with etch resist pens, sheets of etch resist rub-on transfers, the ever-popular photocopy and clothes iron, and now with small CNC mills. It’s still a messy, slow, and expensive process. [johnowhitaker] and [esot.eric] are trying to solve the latter of these problems with a mini PCB printer made out of DVD drives.
Playing around with the guts of a DVD drive is something [john] and [eric] have been doing for a while now, and for good reason. There’s a lot of interesting tech in DVD drives, with motors, steppers, and gears able to make very, very accurate and precise movements. Most PCBs aren’t very big, either, so a laser cutter that can only traverse an area a few inches square isn’t that much of a downside in this case.
With a small diode laser mounted to a CNC gantry constructed out of DVD drives, the process of making a PCB is actually pretty simple. First, a slurry of laser printer toner and alcohol is applied to the board. Next, the laser on this PCB printer lases over the traces and copper fills, melting the toner. The board is removed, the excess toner wiped off, and the unwanted copper is melted away. Simple, even if it is a little messy.
Of course this method cannot do plated traces like your favorite Internet-based board house, but this does have a few advantages over any other traditional homebrew method. It’s cheap, since CD and DVD drive mechanisms are pretty much standardized between manufacturers. It’s also easy to add soldermask printing to this build, given that soldermasks can be cured with light. It’s a very cool build, and one that would find a home in thousands of garages and hackerspaces around the world.
Everyone has their favorite process for PCB fabrication, as long as you’re a happy hacker I don’t think it really matters. But in this post I thought it might be interesting to describe my personal process, and some of the options available.
Making your own at home
Etching is the classic PCB fabrication option for the home hacker. It’s been many, many years since I etched a PCBs but it can produce interesting results. Some people don’t like it, and I’d personally tend to avoid it as a messy and finicky process. But, if you only need 1 or 2 layer boards with large features (through-hole components are best of course) it can be a viable option. In some cases, I think etched boards look awesome and are a great fit. One example is the skull etching shown to the right. The oxidation and discoloration of the boards adds to the design aesthetic in this case.
For those with a bigger budget a professional milling machine might be a viable choice. I’ve used an Accurate CNC in the past (LPKF and others make mills too), but this is an expensive option (no online pricing, but if $10,000 USD is a lot for you don’t bother). The accurate mill is pretty awesome, it can be fitted with a vacuum bed, automatic tool changer and vision system for alignment. The mill can produce high quality two layer boards with all the holes and vias drilled out. The final step of filling the vias is however manual, but compared to etched boards the results are pretty professional (the mill itself uses milled PCBs!). They claim a 0.1mm (4 mil) track size, I’ve never tried tracks this small but surface mount components were not a problem.
While a fun toy, it’s worth considering if you really need a PCB mill. The only case where they’re really valuable is if you want to be able to iterate over a design with less than a days turn around. This can be useful in RF or low noise designs where you might want to experiment with different layouts, but for other projects the price of a good mill can pay for quick turn around (1 or 2 days from order submission to delivery) on a lot of boards.
Years ago commercial fabrication used to be a very expensive and finicky process. For the most part you’d need to order a full panel putting the service outside of most hobbyists reach. Generating gerbers and drill files to the fabs specification could also be a process fraught with complication.
These days services that aggregate designs onto a single panel and break them out for distribution are common. For my work I mostly stick with OSHPark and SeeedStudio whose services complement each other well. I’ve also used Itead and found them compatible with Seeed (with the added benefit that they supply free boards for open projects).
Using OSHPark gives me the warm fuzzies. A child of the hacker community, born out of DorkbotPDX, all OSHPark’s boards are fabbed in the US (check out the great amphour interview for more details). Their services are limited to either 2 or 4 layer boards (always in purple, and always coated with gold (ENIG)), in 6×6 (6 mil traces with 6 mil spacing) or 5×5. I rarely attempt BGA boards so the 2 layer service works out great for me. OSHPark’s minimum order is 3 boards, which is perfect for prototyping. The gold plating also provides a nice finish, which both protects the board from oxidation and provides a nice surface to solder to. The main reason I use OSHPark however is that they’re cheap for small boards and have a relatively fast turn around (I recently purchased 3 tiny 20x15mm boards for $2.40 including shipping which was unbeatable). From OSHPark in the US to the UK my boards take about 2 weeks to arrive. They’ve also automatically upgrade boards to their super-swift service for free when there’s spare capacity. Their service is pretty slick, and provides a rendering of the gerbers prior to ordering as a final check which comes in very handy.
Seeed on the other hand are much cheaper for larger size boards and volume orders. They also provide more color and finishing options. The cheapest option at Seeed is green PCBs with HASL finish (hot air solder leveling). From Seeed, my boards usually take about a month to arrive (there are a few delivery options, but in my experience this is about as fast as it gets and faster shipping services often make using Seeed less attractive).
The image to the right shows a couple of very simple boards I had manufactured at both OSHPark and Seeed. I’ve never had a fabrication issues with boards from either service (though I prefer the ENIG finish).
Partly due to the limitations imposed by using commercial fabs I pipeline my projects. I send projects out to fab early in the design process and then switch to another design. When the board comes back I bring it up, bodge as required, and iterate over the layout. This works well with a two-week turn around, so I mostly use OSHPark while prototyping.
My boards also tend to be quite small (Arduino shield size or a little bigger). With small boards like this OSHPark is usually on-par or cheaper than ordering from Seeed (whose minimum quantity is 10 boards). With boards of about 100x100mm or larger I consider Seeed as they become significantly cheaper.
As a hobbyist I also rarely need huge numbers of boards, but for workshops when I need 10 or 20 boards I order from Seeed based on the final iteration of my prototypes. This is not only much cheaper than OSHPark, but I can get boards in a variety of colors to make workshops more interesting too.
This post has described some of the available options and my personal process. I hope it’s been interesting, but I’d love to hear about your favorite fabrication techniques, services and experiences both good and bad too. Please comment below!
If you need a circuit board now, you’re probably looking at a toner transfer process; all you need to make a PCB is a copper clad board, a laser printer, some special paper, and the usual etching chemicals. The quality of these boards is highly dependant on the quality of transferring toner to the copper, and getting the process right is as much an art as it is a science. A clothes iron is the easy way of transferring the toner to the board, but if you’re looking for repeatability, you’ll probably want a laminator.
Laminators, too, also vary in quality. The king of toner transfer laminators is the Apache AL13P. With four heated rollers and a steel chassis, it’s enough to do some serious heating. [mosaicmerc] came up with an amazing mod for his Apache laminator that takes all the guesswork out of the settings, and does it all in one pass for maximum repeatability and PCB quality.
The Apache laminator in question is a beast of a machine that drives four rollers with a synchronous motor and also has a ‘reverse’ button that sends the laminations out the front end of the printer. Stock, a toner transfer PCB would require dozens of passes through the Apache, but [merc]’s mod takes care of everything for you.
The addition that makes this possible is a small board with a PIC12 microcontroller. This microcontroller connects the motor driver board and the display interface together, triggering the reverse button to move the board 5/8″ forward and 1/2″ back, giving the laminator an effective speed reduction of 12:1. This method also has the bonus of not tampering with the motor or control circuitry, and allows for multiple passes in the same run.
With this modification, the Apache AL13P becomes the perfect solution to transferring toner to a piece of copper, with the ability to transfer 10mil traces on 1oz copper. The board also offers some other features like thermal sensor failure shutdown and a cool-down mode that overrides the heater. If you’re looking for an easy way to step up your toner transfer PCBs, you can’t do much better than this mod.
With the proliferation of desktop routers, and a number of easy methods to create PCBs at home, there’s no reason anyone should ever have to buy a pre-made breakout board ever again. The traditional techniques only give you a copper layer, however, and if you want a somewhat more durable PCB, you’ll have figure out some way to create a solder mask on your homebrew PCBs. [Chris] figured Kapton tape would make a reasonable soldermask, and documented the process of creating one with a laser cutter over on the Projects site.
The solder mask itself is cut from a piece of Kapton tape, something that should be found in any reasonably well-stocked tinkerer’s toolbox. The software for [Chris]’ laser cutter, a Universal Laser Systems model, already has a setting for mylar film that came in handy for the Kapton tape,
Of course, getting the correct shapes and dimensions for the laser to cut required a bit of fooling around in Eagle and Corel Draw. The area the laser should cut was taken from the tCream and tStop layers in Eagle with a 1 mil pullback from the edges of the pads. This was exported to an .EPS file, opened in Corel Draw, and turned into a line art drawing for the laser cutter.
The result is a fast and easy solder mask that should be very durable. While it’s probably not as durable as the UV curing paints used in real PCBs, Kapton will be more than sufficient for a few prototypes before spinning a real board.
When your project is ready to build, it’s time to find a PCB manufacturer. There are tons of them out there, but for prototype purposes cheaper is usually better. [Ian] at Dangerous Prototypes has just announced Dirt Cheap Dirty Boards, a PCB fabrication service for times where quality doesn’t matter too much. [Ian] also discussed the service on the Dangerous Prototypes forum.
The boards are definitely cheap. $12 USD gets you ten 5 cm by 5 cm boards with 100% e-test and free worldwide shipping. You can even choose from a number of solder mask colors for no additional cost. [Ian] does warn the boards aren’t of the best quality, as you can tell in the Bus Pirate picture above. The silkscreen alignment has some issues, but for $1.2 a board, it’s hard to complain. After all, the site’s motto is “No bull, just crappy PCBs.”
The main downside of this service will be shipping time. While the Chinese fab house cranks out boards in two to four days, Hong Kong Post can take up to 30 days to deliver your boards. This isn’t ideal, but the price is right.