These Are a Game Changer for PCB Production

[@fkiesel9442]

Pro tip for the pogo pins: Don't solder them directly to the board, use the correct receptacle. This way you can easily change pins, once they get dull. Also, sometimes it is necessary to play around with different tip types and spacers to get a reliable connection. Having an unreliable test jig is just the worst, it costs so much time and headache.

[@max_kl]

Is it possible to use a standard 0.1" female header for that?

[@rpavlik1]

I've always heard that, but never seen pogo pin holders, and it sure looks like Adafruit just solders their pogo pins right into their test jigs.

[@wickedprotos1937]

@@rpavlik1 production "bed of nails" testers use socketed pogos. They wear out over long production runs. Also applying even pressure across the unit under test pcb take fixturing skill. As the pogos are increased in number, the force can add up significantly. I have found pogo pins small enough to fit in the .1" female headers and might be fine for smaller runs of testing/programming.

[@DavidD-xm4rz]

Check out company Name "PTR Messtechnik" or " PTR Hartmann". The sleeves "H 1010 C" or "H 1010 L" are very helpful and practical.

[@pflernak]

Not sure he is going to be testing the kind of volume where thats going to matter. NB Pro pro tip. Use slightly shallow receptacles for the pins. It can be a hassle to get the pin out if the pin head breaks and everything is flush with the receptacle.

[@gusbert]

The only thing missing is the provision of "tooling holes" on the board to be tested. These engage with steel pins on the test jig so the two can be aligned accurately. These tooling holes have to be specified with a tight tolerance.

[@mortensentim511]

You need a way to keep track of the boards. Either enter a serial number that is on the board into the test computer at the start of the test, or print a SN sticker for the board from the test jig. If you ever need to rework an already tested board once its been fixed the second method doesn't really work though as you lose all record of it being reworked. I would give all your boards a machine readable serial number (barcode/QR code) so you keep a full history of the test you ran on a board even if they're reworked. If you assign the serial right after reflow you can also track back to the machine the board was made on and see if the same issue is occuring on the same machine allowing easier troubleshooting.

[@rpavlik1]

You should be able to use the mcu serial number once that's installed, probably good to put it visibly on the board too as you point out

[@ashwin372]

Pro tip 👍

[@DanielMelendrezPhD]

What about flashing the SN into the device's EEPROM?, apart from the sticker. That way, if the sticker falls, you can still get the serial number from the device

[@bartoszkazmierczak7249]

If the MCU has an option to read it's serial number then it's probably the simplest way. You can also buy EEPROM chips with pre-programmed MAC adresses. iirc TI sells those. They don't cost much more than a normal EEPROM does and because each MAC address must be unique it's a great way to get some unique IDs for the boards.

[@thunderb00m]

Exactly!

[@CraigHollabaugh]

I first saw a bed of about 500 spring pins at IBM in 1985, they were testing very large disk drive circuit boards. Glad you found them ,they're awesome. I've used them on and off for about 15 years or so. Nice work on using the same design to run your tester. Thanks for the video.

[@MakenModify]

Developing test equipment for electronics production for a living there are only minor things I would do different. For the actual bed of nails I would solder in capsules. That way you can exchange test probes in case one brakes or you need one with a different tip. The capsules also help to get the exact depth on every pin. Testfirmware is a great decision, makes so many things easier. Jtag is mighty but also pretty involved to get started. The software side also sound good. We use NI Teststand as a test sequencer but that is very expansive an a simple python script should do the trick as you said. Customers love to have a test report, so good there ;) We log our results to a SQL database. It's always handy if you have an individual QR- or barcode for the serial on the individual boards themselves. Some board manufactures can add such a QR in production (you would only have to scan it). But that might be a bit overkill here. All in all awesome project and really well done :)

[@carlmen6567]

What's a capsule?

[@DavidD-xm4rz]

@@MakenModify Check out company Name "PTR Messtechnik" or " PTR Hartmann". The sleeves "H 1010 C" or "H 1010 L" are very helpful and practical.

[@MakenModify]

@@DavidD-xm4rz well, I know them, they are my main suppliers for probes (and sleeves) ;) (that is why i called the sleeves "capsule", since that is what they call them on their englisch site)

[@sameasone21]

@@MakenModify QA Tech and a few other probe manufacturers call them sockets

[@thunderb00m]

I am a professional who has been making bed of nails test boards for years. First I wondered if there is anything new I missed out on, but there really isn't this is the standard process. However, full respect to Stephen, there are not a whole lot of people who can design schematics lay out PCBs as well as solder and write firmware on a professional level.

[@decee1157]

Awesome to see a PCB testjig with pogopins. It has been on my todo list to experiment with this concept in a long time!

[@justingort1]

I encourage you to give them a try. I make jigs like these for 6 years now and love not having to connect wires. Also using a sleeve for the pin is a big recommendation over just soldering. Sometimes tin seeps into the pogopin making it useless

[@justingort1]

Added to that supporting the tips is important. They bend easily en then they don't spring back. Usually I make a cover plate that only let's 2mm of the pin protrude above it.

[@taisenification]

@@justingort1 Can you recommend some small pogo and sleeves? :-)

[@CraigBurden1]

One tip for your next rev of the bed of nails, mounting the pins in such a way that the solder joint takes the mechanical stress of the depression load, you will experience solder cracks which will give you a few headaches. I can't tell if you are using recepticals for the pins, but next time you should, you'll find they have a small ring around 3/4 of they way up their length and that should mate with the top surface of the BoN PCB, that takes a lot of the load away from the solder joint. Another thing to do is to make sure that the BoN can't flex at all, otherwise you'll hit the same problem again. When I build mine, I typically stack laser cut sheets of MDF or perspex between the two BoN PCBs to support them. I'm not sure what quantity you are intending to put through this jig, this may not be a problem if your quantity is fairly low. But I have successfully pushed 40000 devices through one of my jigs with the above method and never had a joint fail. Another tip, the recepticals allow you to remove the pins when they fail, which again does happen. Especially when you aren't using guide pins. You should use spring loaded guide pins and a clamping mechanism to ensure your board locates correctly and is depressed exactly parallel to the BoN other wise you'll wear out the pins even quicker. Anyway, great work!

[@moertelruehrer]

additional points for the Eragon book in the background :)

[@max_kl]

Where? Seems like I'm blind...

[@moertelruehrer]

@@max_kl all the time on the monitor bottom left of him :)

[@max_kl]

@@moertelruehrer Oh, right. I kept looking for an actual book

[@RobR99]

I worked QC in the 90s. We had a machine that had different fixtures for each board we made. They had a vacuum suction that pulled the board down into the pins to do automated in circuit testing of all the components. EDIT: I would agree with the other commenters that you should put the pins in sockets for easier replacement. They can easily break if you miss something on the visual inspection before testing that causes lateral pressure. Or uneven pressure from through hole leads. Was a common problem I had to fix at least one broken pin a week.

[@ToyKeeper]

Have been using pogo pin for years as an easy way to reflash firmware on tiny embedded devices without having to move or unsolder anything. They're wonderful things.

[@joseangelpenuelasmachado5047]

Awsome work!!! As others comments you could add some tooling alignment pins, maybe you will do on the Jig press you mentioned. Also traceability with a serial number, labeled or embedded on microcontroller, is a really good one to have. Also you could add a test byte on memory, if a unit passes you store a flag to confirm the board was tested and further tests or field users can be sure all potential issues were tested for that unit.

[@vinamarora7049]

That design decision to have all the test-pins on a separate board was brilliant!

[@maxmyzer9172]

4:44 for anyone wondering what this song you always hear is: Artist: A Silent Partner Album: Sad Past Released: 2014

[@wil7vin]

If your gonna run the raspi headless you might need some automatic script to pull the master version of the testing software automatically that way if it has wireless you don't have to get it out of the jig to update that part of it.

[@RavenBomb123]

Could also ssh into the pi to run the update. My pi servers are headless and I do everything on them from SSH.

[@TheEmbeddedHobbyist]

Funny I was using them 40+ years ago, you have to remember that the board must push down in a straight line. So your board clamps should not have a curve in its path after it starts to push the pcb onto the pins. as somebody pointed out don't solder to pins, if board placement moves you can easy bend the pins. Also one pin head shape does not fit all connection. Built a few custom ATE racks in the 80's for testing production pcb's most have used 1000's of pogo pins.

[@MaleLion.]

Same here, those things have been around for many decades.

[@AJMansfield1]

I remember having to set up a system like this that needed to use some ancient toolchain software -- each test run would involve the tester needing to spin up different Windows XP virtual machines in sequence, and use a script to move the mouse pointer around to different buttons on a bunch of different crappy GUIs to run a minutes-long test and programming procedure on each board.

[@hojirick]

I'm brand new to your channel. It's really awesome seeing your work! Hoping to learn more from your content!

[@mborowski2010]

Great video. I'm surprised I haven't seen your videos before. Now I will, I subscribed! You do beautiful work from designing to soldering to this video. I look forward to seeing more of your work!

[@crashbuilds]

Loved the way this was cut, and how you managed to make soldering engaging. One quick pro tip: don't bother with paste if you're not stenciling - it just balls up and creates metal fragments all over your board which creates cleaning time and potential for shorting. Instead, tin your pads with ordinary leaded solder, purchase a hot plate (learned this from Sparkfun) and just fire up the hot plate with all of the components in place. The whole board will solder at once and components will slip into place just like a reflow oven. I really like the mindset you employ, a la "if it ain't broke don't fix it" and it reminds me to just stick with things that work already, rather than constantly try to put my personal stamp on designs. There's already a ton of amazing projects out there on the open hardware scene, I should really focus more on supporting them than going off on a tangent. Thanks for the extra perspective!

[@yjweaver5108]

Ok but how do the parts get held in place ?

[@max_kl]

The music was a bit loud, seemed much louder than the talking parts. Especially the second track

[@joshuavincent7884]

I find this in a lot videos but actually this one was fine imo

[@sjoer]

2021 and you just discovered pogopins?

[@johnkaplun9619]

Recipts a niceidea because it's easy to associate with the physical board and keep track of everything.

[@dengyun846]

Great choice of music for the montage at 4:00.

[@jensschroder8214]

One tip. A big LED. Pulsing blue = test is running. Red = test failure. Green = test OK and done. You see what's going on.

[@juskim]

Nice! I'm going to take your idea of using 2 PCBs sandwiched to align the pogo pins vertically. :) I normally 3D print the jig and super glue them to align it vertically but your sandwiched method is definitely more easier to assemble!

[@danilolattaro]

So, why did you use a black magic probe to program the STM32F407, but a SWD programmer to program the STM32F103? You could have used the same SWD programmer for both STM chips

[@stephen_hawes]

Honestly, I didn't have the 1.27mm -> 2.54mm header converter to make the connection! 🥲

[@danhellgren5671]

I normally epoxy stainless steel pop rivets onto my pogo board to align the test board. Works perfectly.

[@robertparenton7470]

WoW! Thank You from Frisco, TX.

[@max_kl]

Nice! I'll try using pogo pins instead of a programming header in my current project

[@GENcELL2014]

I like Superbuttons made by HCD (High Connection Density, Inc.). They're a wire engineered to be a highly capable smd contact that sits in a double sided adhesive backed carrier laminated/adhered to test pcb with no soldering. They can be less than 1mm thick and 6A-12A per Superbutton. There's no time involvement trying to get contacts all parallel/perpendicular because Superbuttons are held together by the carrier which is just a sheet of plastic or composite with drilled holes for Superbuttons.

[@wesley00042]

Ha, I was thinking about printing test results right before you mentioned it. APC does that with their UPSes as well.

[@benjamincrall8065]

If the recipt printer is just for the user to know the test status, you could put a simple screen on the RPi (like you would with an Arduino or something like that) that could show that the board passed, and only print failed results

[@sio2groper410]

everett charles technologies has a wide range of probes and receptacles. Back in the day I had 1000+ probe in-circuit test fixtures loaded with their probes and had very little issues. The square end 0.1" centers receptacle tails will fit into a standard 0.1" IDC cable header OK. Board under test needs to travel parallel with the fixture base or the probes get a side load and either bend and then stick down or snap after a while. Tooling pins keep the board aligned with the jig and help to prevent that a lot.

[@strandtc]

This is awesome. You certainly have some awesome skills. I want to get into this more, and be able to design and make pcb electronics.. but have never found the right information to start learning. Any suggestions on where to learn how to do this?

[@webdes03]

Been watching these videos since the beginning. 10/10 for content as always. As a point of feedback though... not sure if it's something new with your setup, or that specific pair of glasses, but I found the reflection of the studio lights right where your eyeballs are pretty distracting, especially since you do such a good job of making eye contact with the camera lens, and your viewers.

[@randycarter2001]

If you're going to use a Raspberry Pi as for ATE control then hook up the 7" R-Pi display. It's not to expensive and will give you nice feed back and since it's a touch screen a great control panel. You can get very nice enclosures, like the SmartiPi case, that can be screwed down to a board along with the other PCB's for a compact all in one unit.

[@rpavlik1]

If you look at some of the Desk of Ladyada videos as well as their Made in NYC videos, you'll see a bunch of their pogo pin test jigs. Fairly similar to your basic plan, except they tend to use Teensy boards to drive the test so they can put the firmware on sd card and their swd dap library plus whatever test functionality is needed, have enough pins for a 16x2 LCD for results when the product isn't a display. Cool to see your slightly different approach. They also have a couple kinds of little clamps they use. Love the thermal printer idea. Definitely serialize the boards plz, and/or at least track batches. I have a ton of pogo pins (ended up getting 10x my order) but have only used them for Arduino Nano 33ble swd pinout, plus some battery boxes.

[@tofubl]

Cool stuff. But if you're using a Raspberry Pi anyway, why not skip the blue pill and use the RPi onboard serial and OpenOCD + GPIO SWD driver?

[@km5405]

you could also ship receipt with the board showing it went through the test correctly.

[@stephennixey]

Spring Pins I was using back in the early 1980s with a company called DITMCO Corp for PCB testing in a matrix .

[@daskasspatzle2396]

Does the test procedure include creating a serial number?

[@adamarmfield1069]

blue pills can have a variety of mcu's on them, some chinese clones, might be tricky to get a consistent supply, or for people to find good ones

[@jlegen]

Well, how many test jigs with a blue pill do you think he should build? Ideally, he‘s already done, and never needs any more blue pills… ;-)

[@عمرمطر-غ5ب]

You need to test this testing board 😺😺

[@tuure991]

Awesome stuff as always! Especially the interchangeable separate board for the pogo pins. :) A small thing that would make a big impact in the green screen scenes for me at least is if you moved the filming light higher or more to the side such that it doesn't reflect from your glasses and block your eyes every now and then.

[@chriskukowski398]

Make a screen with thin stainless steel, place on board, apply solder paste, squeegee across apertures, place components, stick in oven, voila!

[@darkshadowsx5949]

how are you getting components for these boards? there are so many things that are out of stock i cant even find suitable replacements for some of the things i need. i even have parts on back order since last year.

[@morrisonnx02]

seems like a really good time to add a barcode scanner to the pi and make serial number stickers so you can keep track even have a prefix for revision number or something

[@marek_ryn]

Just found your channel. Many of the pogo-pin tests can be replaced with JTAG boundary scan. The only problem is that software and JTAG programmers for this purpose cost money. But on the other hand fabricating testing rig for each project also cost money.

[@stevenpetryk6869]

Are those prescription safety glasses (or just stylish safety glasses)? I really like how they look. Where are they from?

[@RoyvanLierop]

To track the results of the tests I would setup a simple PostgreSQL database. Could even run in Docker. Would be a great learning experience if you have never dabbled with SQL before. If you go that route than PL/Python might be interesting too, running Python directly from your database. No script needed!

[@mohitgadhok4209]

Impressive design! As other have said best to use recepticals. Pins can fail at any time for a number of reasons.

[@daniellora7740]

It is cool! I will be waiting for the mechanical part video

[@sduartes2004]

you dont need to solder or use a PCB. You can 3d print the fixture to hold the receptacles and the pogo pins.

[@alexstone691]

Finally another channel doing some pcb dev, i love watching this as i cant do any myself

[@Slushee]

4:43 Of course that music is familiar! It's PhoenixSC's outro!!

[@lukasz_kostka]

That one "extra" UART is a GDB server. BMP won't allow you to flash / read at custom address. It only works with ELF files. And watch out for wrong value USB pullup resistor on those blue pills. Won't it be easier for you to flash target with some code that will do the test and report back ? Instead you have poop loads of pogo pins.

[@lukasz_kostka]

I'd sent a link to github but YT genius AI removes my comment with link.

[@rjordans]

Some alignment pins may be a good addition, otherwise looking great!

[@MohamedAfzal]

hi, your contents are awesome. I have a question, How do you make the board appear in custom name? are you using own VID and PID?

[@seanocansey2956]

Where are you getting your STM32's from? 🤔

[@julianbinder2371]

ordered them before they ran out

[@NavinF]

I didn't get a close up look, but he's likely using a GD32 or similar Chinese reimplementation of the STM32F103. These days it's impossible to buy a Blue Pill with the original ST chip and there's little reason to do so.

[@joelchaney5607]

Have your system also program in a serial number?

[@Velsethen]

Typically bed of nails are done becuase they wire wrap the pogo pins to easily change them when they break. Or inverse flying probe machines exist for a reason.

[@pamelax64]

Hello,sorry for a beginner question,what is and where you get this gray paste soldering,i had flux but not with this (silver effect). thanks in advance!

[@CdnPIMO]

I wonder if the sharp points of the pogo pins will wear out the pads on the bottom of the motherboard? Just my first thoughts with this, cuz I've actually been looking at this option for the same kind of PCB stacking function... Great video btw!

[@pflernak]

Well if you keep testing the same board over and over then sure thats an eventuality. Might want to have actual pins instead of pads then and use concave pogos.

[@draco5991rep]

I wasn't aware of pogopins before this video, they seem pretty cool.

[@conrad2468]

Steven, awesome work. I've been quietly following along for the past few months. One thing that could help you with your one-off boards is home made laser cut stencils. I don't think you have a laser cutter but they've gotten pretty cheap over the past few years. I personally use 5mil kapton tape and it sticks right to the pcb after some alignment. The kapton costs as much as the laser cutter but I've been using this method for years with the same roll of tape I started with! Just an FYI from someone who is burned out on hand soldering smd parts.

[@max_kl]

Since the boards are sponsored by PCBWay, it should be very easy to get them to send a matching stencil along

[@decee1157]

He just have a thing for manuel soldering. He was streaming the soldering process and told us that.

[@EraYaN]

Also for the shared server, might want to stick to plain HTTP/REST instead of plain TCP, will save you a ton of headaches. And you might need to find a way to identify every board with a serial number barcode or a programmed serial number.

[@narwodev]

Wow...nicely done soldering that STM32F4.... :)

[@pallas-cat-s]

Did you thought about protecting your inputs after the voltage dividers? Some Schottkys ?

[@0x3333]

Which PCB clamp/vise is this? Round one...

[@Dragonmastur24]

Eragon is hands down my favorite book series!!

[@SomeGuyInSandy]

From what I've seen, spring pins in the real world cause nothing but trouble. I assume they are fine for a temporary connection, like testing as you've mentioned, but the use cases I've seen (robot joints, etc.) all seem to suffer because of them.

[@starlingwatch3921]

Do you have a favorite jig to push the pogo board to the board under test? -Frank

[@platin2148]

Is the neo pixel directly controlled by the stm32 or is it going through say a i2c led driver? People here at work seem to like the led driver idea i just found it strange. Are there Unit-Tests for that Firmware too?

[@sylvainfalque726]

Where did you get your Stm32s from ? I'm so desperate finding those bad boys 😥

[@hansroemerszoonvanderbrikk7626]

The transistor principle on your shirt comes from The art of electronics

[@theindieg]

Great video! Aiming to make something similar for some projects that I have been putting together. Out of interest, what pogo pins and ribbon cable connectors do you use?

[@Scrogan]

Is that bluepill an actual STMicroelectronics STM32? Because a lot of them will have knockoffs on them, some of which have known defects.

[@enredao_electronico2737]

Nice !! When do You think a full tested kit of Index would be ready for sales?

[@Xambax1]

i dig that transistor-man art of electronics t-shirt :)

[@arxaaron]

Quick (and odd) OT question, but: Are you related to or do you know of William Hawes, the creator of Amiga's Arexx interprocess communications language?

[@OscarSommerbo]

I think it would be a good idea to test continuity on the board, where possible. That would catch bad boards and reduce time looking for incorrect faults.

[@AJMansfield1]

Continuity already gets tested at an earlier stage after the board is printed but before components are soldered.

[@DanielLopez-up6os]

Pogo Pins are indeed wonderful!

[@Jandodev]

I’m gunna make that receipt printer! I’ve never seen anything like that before and I want one

[@mvadu]

Thank you showing the journey of starting a hardware product from scratch.. Amazing.. On a side note your overall soldering work flow looks amazing.. If you don't mind can you make a video of how you handle MCU and large chips, as well as delicate pieces like USB C port.

[@Deveyus]

Can't say I agree with you committing the 'change after test' sin; response testing from a marlin flashed board seems drastically more sensible.

[@roseroserose588]

I guess it depends what you're testing - there's two things to check: hardware & software hardware tests need to happen for every board, and seem to be the focus of this setup. software tests are less important for every board really - if you test it on a known board, and it works, then you know every board like that will work. Plus i'm sure the marlin team themselves will do a extensive testing of their own which removes a lot of pressure from small operations like this, that's one of the benefits of using marlin over a fully bespoke firmware

[@EasyOne]

Very well explained

[@nicolasbrusco1997]

What kind of microscope are you using?

[@brighttech8922]

Would you help me create a jig for my pcb please?

[@adamcarver9057]

What PCB design programs are prefered?

[@mrvisual2482]

Time to get a small solder oven and stencils!

[@prashanthb6521]

Dude, you are a genius !

[@MerlinIslandCrabCoveRailway]

could you run JTAG instead

[@renjiecai7869]

9:28 Challenge: repeat "at the bottom of the motherboard" as fast as you can.

[@AndersJackson]

Store the test result in a database and print the key on the paper.

[@fischX]

Zebra recipe printers are fun to program at least the old ones with BASIC on board.

[@Chrisknot94]

I for one would like to make it known, that I am in fact neither a goblin, nor a ghoul. Best, Christoph

[@draco5991rep]

Okay orc, whatever 😉

[@tempbmx]

as 37 year old white guy from the mitten getting into this level of stuff after years of the steps prior, to it, your video was the least painful to watch.. maybe a few decibel lower on the techno but for all effects, personality, info, lack of scripting, and charm and charisma are concerned, much much better video than most out there. subd.

[@ghanrajsingh7347]

I wanna join you please tell me. How I can

[@VorpalForceField]

Nice Work..!