PCB rework | Hacker News Reader

PCB rework

home.comcast.net

195 pointsjgrahamc49 comments

Wow, this is fantastic. This can only be done on 2 sided PCBs, since you couldn't connect the inner layers of the PCB patch back to the original PCB. But it reminds me of Kintsugi[0]. Perhaps if he used a solder that had gold mixed in or plated his etched PCB with gold, it would even look like Kintsugi.

[0] https://en.wikipedia.org/wiki/Kintsugi

Right! That is an awesome repair job but don't try to fix your RaspberryPi board like this. :)

inner layers can be repaired but it is even more labor intensive http://www.circuitrework.com/guides/4-2-6.shtml

things can get even crazier: http://www.circuitrework.com/guides/6-2-1.shtml

Interesting. I guess maybe you could repair a multilayer board that had a hole in it by milling down the edges of the hole in kind of a stair-step pattern to expose each inner layer and then attach your patch. That would be quite a job.

I bought a dead Korg Trinity synthesizer (for £50) a few years ago some chump had poured a cup of tea inside and it burned up a few traces, fortunately all power ones. I would have actually done those steps but after doing a year of rework on 4 layer boards (my first job!) I couldn't face the job.

Stripped it down and sold the keybed and chassis parts on ebay and made enough to buy a Korg Triton so I was happy.

This is somewhat off topic, but since you mentioned it and I rarely see anyone into such things: I have a Korg Triton Extreme with a dodgy joystick (it gets stuck on the pitch bend), how easy a job do you think that would be to fix?

That's handy: I fix them for a hobby so :-)

It's a pretty easy fix. I'd buy a new joystick and replace it yourself though rather than repair the unit. It's possible but I've never taken that assembly to bits. The Tritons are built to be repaired so it's pretty easy to replace the joystick assembly. If it's like the Triton Studio (what I have) or Pro which I suspect it is then it's a case of undoing the large base screws, taking the base off and the joystick assembly can be removed easily with a few screws around the edges. There will probably be a long cable that attaches it straight to the motherboard and that's it. Careful with the cable when putting it back as they can get trapped and sheared if you don't route them properly.

You can get the parts from ebay usually but you might have to wait a bit for one to appear. Either that or Korg can send you parts out from their service center. They're pretty good, reasonably priced and have bits going back to the 1980s in stock!

Overview here on how to disassemble (should be the same): http://www.bustedgear.com/repair_Korg_triton-pro_disassembly...

What a beautiful website.

Great link, I knew I'd seen something like this before.

I wonder roughly where the board cost makes this level of massively labour-intensive rework cost-effective. Unless parts/plans are totally unavailable, I'd have thought respinning the board would probably work out cheaper unless it's something really weird.

Thanks for the wiki link, very informational. I'm curious how to do Kintsugi, or if you can buy pottery with that resin.

It looks like the patch PCB is held in only by the soldered wires. Although he is right in using wires and solder, the first rule of soldering is that "Solder is not a mechanical connection". He should have held the PCB in with some epoxy, too. A thin run around the edge cracks would have made it 10x stronger - necessary because this is in a subwoofer, and there are parts mounted to the patch board.

He actually mentions that solder is not a good mechanical connection and said he used solderwick to reinforce the patch joints.

The braid is laid flat across the join and then soldered to both boards. The soldering braid is only fixed using solder. That's not mechanically strong. There's no way to wrap the braid around anything.

In my experience the solder-copper bond is stronger than the copper-substrate bond. I'd expect that to fail first. It's a very cool little project nonetheless.

Wow, this isn't just "PCB Rework", it's PCB surgery and transplant!

This seems to be a common enough problem that there is a service for it: http://www.ebay.com/itm/like/281486792952?lpid=82 "WE REPAIR JBL L8400P"

This fellow has too much time on his hands. :-)

Seriously, electrically this seems to be OK. I would question the mechanical strength of the repair. On the implanted PCB, it seems to hold a couple of rather large inductors and other smaller components. The PCB seems to be held up by the "solder joints" that are connecting the implanted PCB to the original PCB. Maybe after a few years, the mechanical stress on these "solder joints" will show up and another round of repair work is ensured.

No, he's reinforced the solder joints with wire and the larger ones with solder wick so it's going to be just fine. (That info is further down in the article).

That's a lot of work. Since this seems to be an ongoing problem with this particular board, it might be easier to just make up an entire blank board. You could move the big components over to a new board easily, and put in new resistors and caps.

Or redesign the board. It's a class D amplifier, which is a PWM device. This is a motor controller with delusions of grandeur. I suspect those caps are trying to filter the PWM signal down, and are seeing big inductive spikes from the speaker magnet.

If you just stuck a 600W power MOSFET amp board in there, you'd get as much bass as ever, with less hassle and a smaller parts count.

http://digitkits.blogspot.com/2011/11/600w-mosfet-subwoofer....

http://www.parts-express.com/1x600w-tas5630-class-d-amplifie...

600W linear amplifier? Sure, lots of bass, lower parts count. But the transformer will need to be larger, the heat sink will be much larger. If you call class D amps "motor controllers" then I'm going point out that speakers are just fancy linear motors. Bass distortion is practically inaudible anyway.

The picture Perfectly flush with the rest of the PCB gives me the shivers (https://home.comcast.net/~rburn/PCB-rework/images/fit3.jpg).

I once held a power supply like that and by accident touched both contacts of one of the main caps, still charged, with my index fingers. Burnt holes in my skin and gave me a shock I could feel up to my shoulder. My arm still had a funny feeling to it hours later.

NEVER will I hold a PCB like that again. Even when I now, 10 years after dat, am repairing any device my body remembers what happened and makes me extra careful, almost scared.

Remember kids: always get rid of the charge using a suitable resistor!

Try opening up a disposable camera and getting jolted by a shock from the flash's capacitor. It jolted me back and made a sound like a fire cracker going off. I had two burns and we were able to reproduce the shock by placing a flat head screw driver on the capacitor once it was recharged.

In my youth, we used disposable camera flashes as the ignitors for potato cannons. One lead soldered to a screw, the other lead soldered to a bent coathanger. It's pretty amazing how many times those flash caps can handle getting a direct short!

Very cool! Right now I need to create a PCB schematic given an original circular PCB (and get all of the spacing correct for the holes). I was thinking of using a scanner and am glad to see someone else is doing this successfully.

A useful tip: Most PCB entry programs let you enter coordinate locations of parts and pads by hand, allowing for exact duplication of an existing layout.

Reverse to schematic, or reverse to board-layout ("gerber") level? The schematic typically doesn't care about things like hole spacing, although it might be attached as footprint metadata to a component.

Once you have some nice clean scans which align properly, there exist programs[0] which will turn it into a netlist for you.

Then when you're at the stage of producing the new board-layout, printouts (especially on transparencies) at 1:1 to compare against the original board are invaluable.

[0] http://www.scancad.com/products/fab.php is one, I'm sure I've seen free/OSS ones though.

https://stacks.stanford.edu/file/druid:np318ty6250/Johnson_R... is more what I'm thinking of, but doesn't appear to provide sources for their work.

Found my new hero, simply astounding skills. The closest I got to this was to bring an extra power line from a PC's power supply to a graphics card that was drawing too many amps ... yeah very pale compared to this ..

I remember doing that too, after some heavy draw toasterised a few of the ATX connector wires.

I was quite surprised at the efficiency of the inner ground plane(s) in sucking the heat right out of the attempted joint, and ended up requiring 2 irons and a hot-air preheat to actually get it to reflow.

Wasn't my machine, but worked fine for years afterwards, I believe.

Yeah, I did install the first mod-chips on a Xbox that required a million soldering points way back when. But still, I am a noob compared to this dude ...

How lucky was he to have a service manual that contained an actual plot of the PCB? That's got to be really, really rare.

It used to be really common.

It's very common with older gear. I've bought some things in the past that came with a hand bound booklet with photocopies of the hand drawn circuit diagrams. It's still standard with obscure test equipment, but probably not for any sort of consumer product.

You can still find service manuals and schematics for a lot of electronics (including laptops and phones) - just look on the Chinese side of the Internet. Ironically enough this availability is often through a lack of security rather than official release as most companies are against repair of their products.

Any suggestions on what to search for for those of us that don't speak Mandarin/Cantonese?

I would guess search on part numbers printed on the PCBs.

"service manual" or "schematic" (with the quotes, to avoid any obnoxious query rewriting) and the name of the device works well, since they tend to use many of the English terms too.

Many older consumer products had both schematics and PCB diagrams in their manuals, if not in the operation manual, then the service manual.

I've seen lots of schematics in old consumer manuals, but not PCB layouts.

It's not really older consumer products, you can find brand new ones as well. It depends on the type of product (not the typical throw-away style) and the standards used by the company producing it. For instance gear like rotary hammers, hifi amplifiers, .. in mid to top ranges typically have ver, very complete service manuals. Including schematics.

An actual service manual intended for technicians who would be repairing the thing is likely to contain a lot more than the user manual would, although I have seen schematics/board prints in both.

Even without that available, stripping it of all components and running both sides through a high-res flatbed scanner works surprisingly well, although top-bottom alignment can be a bit of a bugger afterwards. (And wouldn't solve his problem with the gaping hole in the middle)

And it doesn't work on >2 layer boards, obviously.

That said, the part he ended up with didn't look too complex, and it's possible it might have been inferrable from either a schematic or by reversing the rest of the board to a schematic and making logical assumptions about what must connect where. Things like appropriate track sizing for the high power traces might be a bit more guesswork though.

It's a fairly common technique in board reverse engineering, although usually you have a couple of sacrificial units, so you don't have to worry about putting all the damn SMD components back on afterwards :P

This takes DIY to the next level! Can someone point me to more of this person's works?

In retrospect, was resurrecting the dying board with a "transplant" the best solution? Harvesting the components and etching a fresh board seems to be a time-effective and more robust solution. Maybe the author chose to do it anyway just because...

Note the large number of surface mount components on the one side of the PCB. Initially I though just like you did, then I noticed the component count and realized this really was the quickest way to get back to a working machine.

Besides, other components might fail during the harvesting.

I think I have found this guy. I went one step upper in the URL (https://home.comcast.net/~rburn/) and then walked down the links here: https://home.comcast.net/~rburn/metalsculpture/content/me.ht... The first Wayne Martin in Google search results for "Wayne Martin PCB rework" query is https://www.linkedin.com/pub/wayne-martin/49/4b2/506

This won't last.

While it's nice to see the amp in a working state again, he missed out on one important thing: The replacement inset is hold in place by thick solder bridges only. This fixation definitely won't last.

He'd better fixed it using loose wire.

The author mentions this and notes, " The gaps here were bridged using solder wick." This is a flat copper braid so should be pretty strong.

My bad, missed that part of the article... However I still doubt that the fixed part is as strong as it was originally. Especially that coil looks rather heavy

I love this. Thanks for documenting it, its a very cool experiment. I'd be waaaay to lazy to go through this.. seeing someone else do it successfully is great!

How did he cut the FR-4 so cleanly? Normally when I mill FR-4 it turns into an absolute mess.

Belt sander. Also, a sharp burr and very high RPM works OK for me. Just don't inhale the dust!

Reply on news.ycombinator.com