Help with E8-V1 TKL PCB/Multimeter - Bad Column, Repeated Keypresses

[onmybikedrunk]

So I posted on Reddit a couple of days ago and didn't get much of a response outside of a single friendly and helpful dude... I'm gonna lay out this issue below in depth but really what I need help with is how to track down a suspected short to ground in one of the columns of the PCB.  I've never used a multi-meter before outside of home-repair stuff and I'm a bit lost.  For those wondering: a drop-in TKL PCB replacement is kinda out of the question as the USB port is a bit off-center and unique (hacking the case or bridging a usb daughter-board is a last resort).  Being that a lot of people got kinda borked with this GB, there aren't many people out there with extra PCB's either.  I have been in contact with Exclusive - the man that tried to make it all right after the original GB ended.  He's been a great help and is a really good dude.

See images below... There are super-hi-res PCB close-up images hidden under a "more" section and beneath the screenshot of the testing software.  The screen-shots are low-res and may prove helpful.  Hi Res PCB shots for reference OR if you can spot something I couldn't!  Black screws on stabs are plastic, brass are original metal.  At the time I took the pics of the PCB, the column in question was still purposefully desoldered for testing...just so yah know.  Ok, here we go...

Issue Description in Length:

This all started when I was cleaning up my office and ran over the keyboard with a VERY damp microfiber towel. It was plugged in (stupidly) and connected to a Windows desktop. I unplugged it (without noticing any issues at the time) to get a bit deeper in between the keys. There is ZERO chance any moisture penetrated the board and/or got to the PCB.  There is a possibility that ESD could play a role here...?  But yeah, I mean the rag was practically dry.  When I sat down in my nice and clean office this is what I discovered...
It seems like the following keys

Code: [Select]

F9 LShift RCTRL ' [ - are being repeatedly pressed over and over in super-human fast succession. This is according to Passmark's Keytest software. If using Aqua'S KeyTest it registers the same except as

Code: [Select]

LCTRL F9 RShift RCTRL ' [ - I would assume this is maybe the difference between BIOS and Windows code? The other thing that happens is that it seems almost as if Windows locks the keypresses once I plug this keyboard in. I have to physically restart the computer to get an alternate keyboard or Windows for that matter to work properly - which makes testing a PITA. For instance, if I plug in the keyboard, test it and then unplug it, then plug in a fresh keyboard (different controller, same cable - known good) now any keypress is as if I'm still holding down shift+control - so if I'm in a Chrome window and press T it will open a new tab and Windows won't let me double click on something like a folder. Now all my other keyboards work fine and I get this same issue across multiple machines, so I would assume it's not a permanent OS issue - the OS is just having a hard time catching up with so many rapidfire keypresses. All other keyboards with Atmega controllers work fine as well on a fresh boot!

What I have tried:
I tried putting it in bootloader mode, re-flashing firmware and resetting EEPROM, got a fresh hex file from Exclusive directly (he also looked for a replacement for me), desoldering the troubled switches (including right shift/CTRL AND left shift/CTRL) and tested, inspecting the PCB for any solder bridges or contaminants via magnifying glass, replacing screw-in stabs with plastic screws (where it came close to traces) and nothing changes... I can't spot a damn thing wrong!

Assumption:
Either a fried controller or, hopefully, a short to ground. Let me preface by saying other than soldering and watching Louis Rossmann on YouTube I have NO electrical engineering experience. I'm not a hobbyist (outside of keyboards) and I'm pretty limited to soldering/desoldering switches, making jumpers/cables and flashing firmware. I don't know the pinouts (or what they mean) or what makes the Atmega32a a good controller...so go easy on me.  What I really need is someone to hold my hand in finding the short via multimeter.  This is my chance to learn something new and if someone has the patience to pass on their knowledge I would be forever grateful. 

I don't wanna retire this entire board just because of a PCB issue. There has to be a way! I have faith in you all!  Thank you for your help and responses!

This is the Multimeter I'll be using.

More

[pixelpusher]

Don't know why it would cause issues now and not before, but your screw in stabilizers need a washer under them on the right shift.  That screw is contacting a pad.  Can use a piece of tape in a pinch.  You say you swapped them, but there's no way that one isn't making electrical contact.  So you need something under it.  Maybe the black screws in the photo are plastic?  Surely someone one here can help you fix it.  Gotta be a short, like you say.

Your link to the multimeter isn't working.

[onmybikedrunk]

Don't know why it would cause issues now and not before, but your screw in stabilizers need a washer under them on the right shift.  That screw is contacting a pad.  Can use a piece of tape in a pinch.  You say you swapped them, but there's no way that one isn't making electrical contact.  So you need something under it.  Maybe the black screws in the photo are plastic?  Surely someone one here can help you fix it.  Gotta be a short, like you say.

Your link to the multimeter isn't working.

Yeah it's a weird one...  Yes, the screws you are referring to are plastic (black).  The brass are the original metal ones and none are even close to a trace.  I usually add a small Teflon washer under all stabs no matter where they are but I guess I was out at the time of building this board and I must have forgot to go back in there.  Link is fixed 

[suicidal_orange]

First the easy test you probably tried but didn't mention - use a different USB port.  I suggest this because when you unplug a keyboard it should stop pressing anything (unless maybe if you have sticky keys turned on) so forcing windows to load new drivers can't do any harm.

To confirm a short you need to use continuity mode on the multimeter which is the one that beeps when the probes touch - there's a helpful picture confirming it's the one with the Ω and diode symbol.  Using this you can confirm that all the switches are connected on their pins that aren't connected to the diodes.  The column of keys looks normal, left control looks very strange but maybe it should be connected, or maybe it's a clue to the problem.  A link to the firmware would confirm if it should be connected or not.

Is it shorted to ground?  See if you get a beep when connecting the switch pin to one of the round switch LED pads.  If not try the GND pins on the chip - the dot on the diagram is dented into the top of the chip so you know which way round it is.  Three of the GND pins are in the middle of their side but one isn't, just to be annoying...

[onmybikedrunk]

First the easy test you probably tried but didn't mention - use a different USB port.  I suggest this because when you unplug a keyboard it should stop pressing anything (unless maybe if you have sticky keys turned on) so forcing windows to load new drivers can't do any harm.

To confirm a short you need to use continuity mode on the multimeter which is the one that beeps when the probes touch - there's a helpful picture confirming it's the one with the Ω and diode symbol.  Using this you can confirm that all the switches are connected on their pins that aren't connected to the diodes.  The column of keys looks normal, left control looks very strange but maybe it should be connected, or maybe it's a clue to the problem.  A link to the firmware would confirm if it should be connected or not.

Is it shorted to ground?  See if you get a beep when connecting the switch pin to one of the round switch LED pads.  If not try the GND pins on the chip - the dot on the diagram is dented into the top of the chip so you know which way round it is.  Three of the GND pins are in the middle of their side but one isn't, just to be annoying...

Wow, this has been a great help.  Yes I tried multiple USB ports and cables, even two separate Windows machines and a Hackintosh with the same result.  I got a bit distracted with other boards but I plan on testing today/tomorrow.  I’ll let you know how it goes.  Thank you so much for your detailed response! 

[Sup]

Can you show a picture of the back of your PCB? Some other guy also had the issue that one key activated multiple key presses. The reason was that there was a small piece of solder stuck on a trace. So maybe who knows

[pixelpusher]

Can you show a picture of the back of your PCB? Some other guy also had the issue that one key activated multiple key presses. The reason was that there was a small piece of solder stuck on a trace. So maybe who knows

well, i don't think he's desoldered it, so you won't be able to see much.  But that might be the next step.  It's scary, I know, b/c I've rebuilt this board and a canoe (notorious for having a bad PCB), but I didn't ruin either of them.  I do have a decent desoldering iron, though.

[onmybikedrunk]

First off, sorry for the delay.  I have twin 6 year olds at home during school closures and stay-at-home orders - life hasn't been that simple lately.

Can you show a picture of the back of your PCB? Some other guy also had the issue that one key activated multiple key presses. The reason was that there was a small piece of solder stuck on a trace. So maybe who knows

So I did desolder the keys in question, hence the missing solder on the pads in the photos... I did NOT desolder the entire board but I'm willing to do so.  I have a Hakko desoldering vacuum so it's not an issue. 

First the easy test you probably tried but didn't mention - use a different USB port.  I suggest this because when you unplug a keyboard it should stop pressing anything (unless maybe if you have sticky keys turned on) so forcing windows to load new drivers can't do any harm.

To confirm a short you need to use continuity mode on the multimeter which is the one that beeps when the probes touch - there's a helpful picture confirming it's the one with the Ω and diode symbol.  Using this you can confirm that all the switches are connected on their pins that aren't connected to the diodes.  The column of keys looks normal, left control looks very strange but maybe it should be connected, or maybe it's a clue to the problem.  A link to the firmware would confirm if it should be connected or not.

Is it shorted to ground?  See if you get a beep when connecting the switch pin to one of the round switch LED pads.  If not try the GND pins on the chip - the dot on the diagram is dented into the top of the chip so you know which way round it is.  Three of the GND pins are in the middle of their side but one isn't, just to be annoying...

So just a quick update... I tried all the things you mentioned.  All problematic keys in that column go to the correct pin on the microcontroller without issue - I actually tested all the columns and rows with the multimeter and all is well.  Side note: keep in mind that the problem-switches are desoldered and the problem still exists.  There is no short to ground that I have found - I tested this by both your example of the in-switch neg LED holes and by the GND on the microcontroller. 


Here is a copy of the hex that Exclusive sent me if you're curious.  In the AquaKeyTest the LControl was simply wrong.  It hasn't shown up in any other keytest software so I wouldn't pay any attention to it.


I've been in talks with a local guy that runs an electronics repair business and also happens to love keyboards... I'm going to have him potentially replace the microcontroller for me.  Do you think I should desolder the entire board as a next step?  Thanks again for the help - all of you!

[pixelpusher]

I've never had a problem like this that wasn't visibly identifyable.  So yeah, seems like the next step is to desolder the whole thing.  Maybe you'll get it apart and there will be something wrong you can see.  Not much to lose at this point.  You can't use it in it's current condition.

[onmybikedrunk]

I've never had a problem like this that wasn't visibly identifyable.  So yeah, seems like the next step is to desolder the whole thing.  Maybe you'll get it apart and there will be something wrong you can see.  Not much to lose at this point.  You can't use it in it's current condition.

Yep, roger that.  I ordered a new controller just in case and I'll post up the end result.  One minor annoyance is that I built this board with the very first round of Invyr Pandas and they have this SIP style cover over the terminals and they are a PITA to desolder.  Oh well.  I wanted to lube them anyway...  Thanks again for your help!

[pixelpusher]

I've never had a problem like this that wasn't visibly identifyable.  So yeah, seems like the next step is to desolder the whole thing.  Maybe you'll get it apart and there will be something wrong you can see.  Not much to lose at this point.  You can't use it in it's current condition.

Yep, roger that.  I ordered a new controller just in case and I'll post up the end result.  One minor annoyance is that I built this board with the very first round of Invyr Pandas and they have this SIP style cover over the terminals and they are a PITA to desolder.  Oh well.  I wanted to lube them anyway...  Thanks again for your help!

wait... that means you can just pull them out.  No need to desolder.  Of course, you'll have to put the same switches back in or similar switches with thin contacts.