KBD67 rev 2 PCB row 2 triggers entire column of keys

[DaxSama]

This was for my first build, using a tofu65 case. It was working fine for about 3 days until suddenly every single key in row 2 would trigger every key in it's respective column. For example, pressing "a" would trigger "2qaz" or pressing "d" would come out as "4edc". Every key that's not in r2 works perfectly.
So I tried re-soldering the entire board, and it seemed to work for about an hour and now it's having the same problem again.

This is the first time I've soldered so I'm not entirely sure what I did wrong. Is it because of the flux residue causing a short? I left most of the flux residue intact on my pcb since when I tried cleaning it, it just spread a shiny thin layer over the pcb.

Here are some pictures of the pcb, sorry that they're blurry.

[nevin]

double check you aren't shorting against the case or a screw in stab

[DaxSama]

double check you aren't shorting against the case or a screw in stab

I've tested the pcb without the case and it still has the problem, and the stabs have washers installed already, so I don't think they're the problem atm.

[nevin]

soldering looks good. don't see anything wrong there

just take the screws out and see if there's a difference (guessing you're using a plate and can't totally take off the stabs)

and that it seems to be intermittent, gives the hint that it might be a short somewhere connecting something it shouldn't

[DaxSama]

Just removed the stab screws, no luck.

Still have no clue where the short is, since it did work for a short period of time after re-soldering, and it worked for a few days after the first solder.
Any ideas of what I could do now?

[nevin]

can you see anything on the other side?

something with the firmware? did you edit or straight form developer? possibly a row or column crossed in firmware?

[DaxSama]

Can't really see anything out of the ordinary on the other side unless I remove the plate.
I did flash the keyboard with my own layout, but I've already tried flashing it back to it's original firmware with no results.

[nevin]

are there standoffs on your plate? looks like a previous version did and that's what was causing the shorts
https://www.reddit.com/r/MechanicalKeyboards/comments/ak7v71/kbd67_issue_multiple_characters_detected/

also make sure every screw hole is insulated well on both sides

[DaxSama]

There are no standoffs between my pcb and plate. I also used o-rings on both sides when I mounted the pcb to the case

[nevin]

something is shorting to ground on that row....

[DaxSama]

Yeah, I'm going to try to re-solder the board again and see if there are any results tomorrow. If nothing gets fixed, I'm just going to conclude that I either I broke something while soldering or I got a defective PCB. Thanks for the help though.

[nevin]

you can try & find the problem with a multimeter, using the continuity tester. (should make a noise if there's a connection)

i didn't say about the diodes because you said it was working fine for a little, so i'm thinking it an intermittent short somewhere

[suicidal_orange]

There's no need to desolder the whole board - the problem is on the row.  You could remove one switch at a time on that row until the problem disappears, but once you're missing half the letters how can you test especially when it's intermittent...

If it is a problem with that row pin (not a short) and it's a rev1 PCB (I don't see anything on it to say it's rev 2) you could cut the connection to the row and connect it to the Caps Lock LED pin instead, if you don't need it as an indicator.

[DaxSama]

you could cut the connection to the row and connect it to the Caps Lock LED pin instead, if you don't need it as an indicator.

How do I do that?

[suicidal_orange]

Look in the firmware which pin is connected to that row, find the pin on the controller chip's datasheet and then on the PCB (note the dot in a corner showing pin 1) and hope there's a visible trace going to it.  There is a trace somewhere, if it's not visible you might have to remove the switches near the chip to see it on the other side.  When you've found the trace get a sharp knife and carefully cut it without damaging the traces either side.  Finally check in the firmware or use a continuity tester to work out how the LED is connected and solder a jumper wire from the LED pad or the other side of the resistor connected to the other LED pad to the switch pin.  Then you will have to modify the firmware to use the new pin...

The most important thing is if it is a short this wont fix it so you need to be sure.

Also if your thread title is correct and it's a rev2 PCB (I missed this earlier) there is no caps lock indicator in the firmware so it's not an option. 

[DaxSama]

Ah, yeah I double checked and it is indeed a rev 2 pcb, so I guess it isn't an option. Thanks anyway though.

Haven't de-soldered anything yet, will do it this weekend and see if it's a specific switch causing the problem. Will update then.