Author Topic: Troubleshooting faulty PCB traces (Keychron K4) Whole row won't register  (Read 1281 times)

0 Members and 1 Guest are viewing this topic.

Offline mcjazzy

  • Thread Starter
  • Posts: 4
Hi everyone, this is my first post here and hoping someone can help me with this problem.
I hope this is the right place to post this topic.

So a little backstory, I was installing Mill Max Hotswap Sockets on my Keychron K4. Managed to tore the PCB traces here and there. Resulting 3 whole rows of the keyboard stopped working. I googled a lot of things online on how to repair PCBs, and managed to fix 2 whole rows by creating bridges between the row pins with wires. But this last row just won't work even though the connection between the pins are fine (I tested it with a multimeter).

Here's some images to help with the story

Video here :
https://imgur.com/ywwrf3d

I've tested the connection between the row pins (point A through point B as shown in the image) and the connection is fine.
I've also tested the connection between the column pins (yellow lines) to each correspond nodes and it's also fine.
I apologize in advance for the bad soldering job, I'm a mess :-[


I've also tested the connection between the nodes and it shows no problem.


Here's a video of me testing the pins :
https://imgur.com/a/4KZM5pw
https://imgur.com/a/K2RGEDB

Full Album
https://imgur.com/a/06NnoV8

I've been troubleshooting for days and googling a lot but still have no idea what causing the issue :(
This is my first time troubleshooting a PCB so I don't have that much knowledge on PCBs.
I'm assuming the problem is somewhere within the PCB itself rather than the switch pins? Any ideas?
If you need any other images or information pls let me know, thanks in advance!

Offline Rico

  • Posts: 71
There is a trace that connect each row(and each column) to one of the pins of the MCU.
Looks like the trace that connect that specific row to the MCU is broken.

Offline suicidal_orange

  • * Global Moderator
  • Posts: 4370
  • Location: England
Rico is right, somewhere a trace goes from the non diode pins to the chip with a green line on it.  If there weren't jumpers in the way it might be visible - can you see round them on the board?  You'll either have to solder to the tiny pin on the chip or scrape some mask off the trace on the way down.
120/100g linear Zealio R1  
GMK Hyperfuse
'Split everything' perfection  
MX Clear
SA Hack'd by Geeks     
EasyAVR mod

Offline mcjazzy

  • Thread Starter
  • Posts: 4
I've tested so many points and I can't seem to at least pin point which trace is it that connects the chip with something related to that row.  :'(

If anyone have any slightest clue on which trace is it or which point to test with a multimeter, I would really appreciate it

I found some images online of the same board if it any help
https://user-images.githubusercontent.com/10539638/73797090-acda7d80-4774-11ea-83e1-90a89b533c83.jpg
https://img2.quasarzone.co.kr/web/editor/2002/2002obj___1288520770.jpg

Also maybe any clues here from someone reverse engineered another keychron keyboard :
https://github.com/alex-dinh/keychron-k8

Sorry it's my first time tinkering with PCBs
« Last Edit: Wed, 10 March 2021, 12:11:12 by mcjazzy »

Offline suicidal_orange

  • * Global Moderator
  • Posts: 4370
  • Location: England
Between your pic and the Korean one I can almost convince myself that there's a trace running from the left pin in box P3 to the right of the chip (2 in the pic with the circles) and the switch directly above it in the dead row.  Short a diode in the dead row to that hole and see if it types the right character.
120/100g linear Zealio R1  
GMK Hyperfuse
'Split everything' perfection  
MX Clear
SA Hack'd by Geeks     
EasyAVR mod

Offline mcjazzy

  • Thread Starter
  • Posts: 4
You mean something like this?

It does not seem to do anything
I also tried to short that hole in the P3 box with other points around that node, still no luck


You know what's the weirdest thing I just found though?
If I touch this part with just a little tip of my finger it seems to register the whole dead row for some reason

Here's the video

I don't know what's happening and just become more confused lol

Offline suicidal_orange

  • * Global Moderator
  • Posts: 4370
  • Location: England
Why did whoever have to put thick red circles on their pic - it would be perfect otherwise.

Lets stick to what we know - there is a trace going from the chip to the switch directly above, it's one of the ones that crosses your "press here and it [kinda] works" circle.  Looking on a proper screen not my phone there are 5 traces (I'm ignoring the one that stops just above the C8 label) running between the end hole in P3 and the corner of the chip but by the other side of the red circle there are only 4.

264193-0

Can you see the trace ending in a via (hole) just above P3, where the yellow line ends?  It could be connected to that row on the other side of the PCB so it would kinda makes sense.  If that trace does end there the next one in should be for the dead row.  You'll need something thinner than the wire you're using and short a diode to the cyan pin on the chip - maybe you could attach one end to a multimeter probe?

If that works you'll need to very carefully scrape the mask off the trace, I suggest doing it on the cyan bit where there's nothing below.

The other option is you've got a short to the via where the pink square is - that is where the P3 pad's trace ends.
120/100g linear Zealio R1  
GMK Hyperfuse
'Split everything' perfection  
MX Clear
SA Hack'd by Geeks     
EasyAVR mod

Offline mcjazzy

  • Thread Starter
  • Posts: 4
Man, thanks to you I finally able to locate the culprit.
Here is the broken trace I needed to fix :


Not the prettiest, but hey it works lol

Now there's no more dead row!

Thank you so much for taking the time analyzing and guiding me through this problem of mine. I greatly appreciate your help.
Case closed!  :thumb:
« Last Edit: Thu, 11 March 2021, 09:55:34 by mcjazzy »

Offline suicidal_orange

  • * Global Moderator
  • Posts: 4370
  • Location: England
Great!

If you're going to be taking the PCB out of the case again I would advise adding some hot glue to the last fix otherwise you could pull up the whole trace (I don't even see how you connected it but it works so there must be some solder there somewhere :)) )
120/100g linear Zealio R1  
GMK Hyperfuse
'Split everything' perfection  
MX Clear
SA Hack'd by Geeks     
EasyAVR mod

Offline Rico

  • Posts: 71
Great work!

Glad your keyboard is working nicely again :)