Author Topic: Troubleshooting a faulty PCB (K8 Keychron ISO)  (Read 1983 times)

0 Members and 1 Guest are viewing this topic.

Offline Pegart

  • Thread Starter
  • Posts: 5
Troubleshooting a faulty PCB (K8 Keychron ISO)
« on: Mon, 21 December 2020, 10:42:36 »
Hi, first time mechanical keyboard owner here. My Keychron K8 disappointingly shipped faulty. A set of keys doesn't send any signal to the computer and the LEDs don't work as intended (they do output all the colors though).

I've contacted customer support and they are going to send me a replacement and didn't ask for the broken one back so I've been thinking about trying to fix it, if at all possible. I have some experience with soldering but basically no ideas as to how to go about testing the PCB to check if any traces, caps or something else is faulty.

Photo that shows the LEDs not working correctly (brightness set at minimum):


Photo of the PCB:


Videos that better show LEDs not working correctly:

Full album: https://imgur.com/a/cL31KWo

As you can see the LEDs on the faulty keys do work but not as they should. They do output all the colors but they are usually at full brightness and even get lit up when other keys are pressed (video #1), depending on the light profile. Besides that the faulty keys themselves do not send any signal to the computer when pressed.

Any tips and ideas on how to start troubleshooting the PCB? I'm open to reading/watching through any extensive/descriptive tutorials that might possibly exist too!

Thanks!

Offline suicidal_orange

  • * Global Moderator
  • Posts: 4202
  • Location: England
Re: Troubleshooting a faulty PCB (K8 Keychron ISO)
« Reply #1 on: Mon, 21 December 2020, 12:01:40 »
A black pcb  >:D  Not your fault but annoying none the less.

Lets start by confirming the diodes are attached where it looks like they should be - plug in the USB and short the yellow dot to the cyan one (you can use anything - scissors, bent paperclip, any piece of wire...) and it should type a 0.  If it does short the yellow dot to the two green ones and see if they type anything.  If you get a B confirm it's a U shaped 'column' by shorting yellow to all the blue dots.  If they all work as expected solder a wire between yellow and pink and you're done.

It probably won't be that easy but without more info it would just be lots of 'if's so please post what characters you do get when trying the above.

LEDs?  No idea.  Seems strange that it's the same switches so maybe the same fix will do both?  We will see...
120/100g linear Zealio R1  
GMK Hyperfuse
'Split everything' perfection  
MX Clear
SA Hack'd by Geeks     
EasyAVR mod

Offline Pegart

  • Thread Starter
  • Posts: 5
Re: Troubleshooting a faulty PCB (K8 Keychron ISO)
« Reply #2 on: Tue, 22 December 2020, 14:34:15 »
Thanks for your time and effort to explain how to proceed. Much appreciated. :thumb:

This was actually the first time I've seen a black PCB in person and the first thing I've noticed was that traces are barely visible. It's really annoying.

On to the troubleshooting. Honestly don't know enough about electronics/PCB so I didn't dare go beyond first step because it produced different results to what you said it should. BTW, if shorting all other combinations would be safe to do I can check them anyway - unless there's no point since the orange+cyan produced different results. Sorry, I'm a noob, first time doing this. Generally speaking can I short two wrong points and actually damage the PCB?

Shorting the yellow to the cyan typed the whole row with slightly varying results. I was scared of pressing too hard and for too long so it might have been me doing a bad job when testing. Is it safe to connect them for longer (let's say a second or two)?

Switch hitter results (I forgot to change to 88 key ISO but I presume that just changes the layout nothing else):


I've opened the notepad while shorting the yellow and the cyan a couple of times in a row and this is what I got (each line is it's own shorting):
^1234567890ߴ
8^1234567890
^1234567890
^1234567890


Only the first sequence is complete and correct but the results not being the same each time might have been me shaking and pressing too lightly or backspace, insert, home, page up all getting triggered as well since they're in the same row. If it's ok, I can try again and just press firmly for a second if necessary?
« Last Edit: Wed, 23 December 2020, 04:35:57 by Pegart »

Offline suicidal_orange

  • * Global Moderator
  • Posts: 4202
  • Location: England
Re: Troubleshooting a faulty PCB (K8 Keychron ISO)
« Reply #3 on: Wed, 30 December 2020, 05:26:41 »
Usually you can't do any damage shorting switch pins but if they are somehow attached to the LEDs when they shouldn't be it might be possible on this board.  No harm done so far, and I was pleasantly surprised to see a whole row lit up, that makes things easier as rows are rows and columns are ... only slightly weird (why connect F11 not F9 to the dead column?)

I wonder if there's a short on one of the chips - one is on the extra blue PCB the other on the main board the other side of the text.  A nice sharp picture of these would be easier to check than looking at the chips themselves as the legs are tiny, if there is a short there anything could happen.

Also if you could get a picture with the shiny area (on the keys above the arrows on the first pic) on the affected column that might be useful though I don't think anything is close enough to short.
120/100g linear Zealio R1  
GMK Hyperfuse
'Split everything' perfection  
MX Clear
SA Hack'd by Geeks     
EasyAVR mod

Offline Pegart

  • Thread Starter
  • Posts: 5
Re: Troubleshooting a faulty PCB (K8 Keychron ISO)
« Reply #4 on: Thu, 07 January 2021, 10:28:42 »
Ok, I've finally had enough free time to continue with the troubleshooting. I've also retested the connections and tried all the other combinations per your instructions in the first reply and got some weird/interesting results. I tried with a paperclip and a wire.

The keyboard types different things depending if I use a paperclip or a wire which is... weird. Right?

This is what gets typed:

yellow + cyan
paperclip -> whole row
wire -> 0

yellow + right green
paperclip -> whole row
wire -> 9

yellow + left green
paperclip -> nothing
wire -> nothing

yellow + one to the left [ND74] of left green
paperclip -> whole row
wire -> +

yellow + any blue
paperclip -> nothing
wire -> nothing


The 0, 9 and + are all what should get typed when the corresponding key is pressed with my layout/language. Are yellow + left green not producing any signal a problem?

I also have a couple of better photos of the whole PCB and the small blue chip as you requested. I'm not sure which is the other chip? What do you mean by "other side of the text"? Do you mean on the other side of the actual PCB or did you mean the one that's only half visible on the first photo, that has some red dye on it or something else?

If provided photos aren't good enough, I'll take more!
« Last Edit: Thu, 07 January 2021, 10:31:14 by Pegart »

Offline suicidal_orange

  • * Global Moderator
  • Posts: 4202
  • Location: England
Re: Troubleshooting a faulty PCB (K8 Keychron ISO)
« Reply #5 on: Thu, 07 January 2021, 12:24:46 »
Is the wire insulated?  I can only guess the bare metal paperclip is somehow grounding through you because as you say it is very weird.  Please use the wire from now on because it gives better results, you can't solder a paperclip on anyway

So the (or at least one) problem seems to be that the diodes for that column of switches aren't connected to the controller chip.  You got the shine perfectly on the column left of the dead one so I traced that, it confirms the diodes are connected down the full column which is as expected.

In the closeup I can read enough of the chip on the extra PCB to find it and it does bluetooth (datasheet) so it is the other chip with the red dye on we need to look at.  I don't see any problems with it in the main pic but there's no shine around it and the red dye makes it unreadable - can you post a pic of that chip with the row of switches above, and if it's not visible what the text on it says?  Hopefully that pic will either reveal a short or we can trace part of the matrix back to the chip and/or find a datasheet to see where the data pins are.
120/100g linear Zealio R1  
GMK Hyperfuse
'Split everything' perfection  
MX Clear
SA Hack'd by Geeks     
EasyAVR mod

Offline Pegart

  • Thread Starter
  • Posts: 5
Re: Troubleshooting a faulty PCB (K8 Keychron ISO)
« Reply #6 on: Thu, 07 January 2021, 13:53:09 »
Yeah, the wire is insulated and the paper clip is the regular metal one. I'll use the wire from now on, it's easier to handle anyway.

I'll get better photos of the chip and surroundings sometime next week. Currently in self isolation as a precaution and won't be able to take good quality photos until then.

I did, however, check the text on the chip to potentially find a datasheet.

This is written on the chip:
HFD
HFD2201KBA
208ANWA0a

I'll post another post with the fresh photos next week as soon as possible.

Offline Pegart

  • Thread Starter
  • Posts: 5
Re: Troubleshooting a faulty PCB (K8 Keychron ISO)
« Reply #7 on: Thu, 14 January 2021, 18:46:26 »
--UPDATE--

I've finally got more and hopefully better photos.

Closeup of the chip with serial/model number:
259849-0

1st macro of the chip and surroundings.
259851-1

2nd macro of the chip and surroundings. A bit different lighting and contrast - maybe it helps.
259853-2

PCB - chip and the columns above. Hopefully it's good enough, it's hard to get a good photo of a black PCB.
259855-3

Another PCB photo with a slightly better lighting in the upper half.
259857-4

And one more PCB photo just in case if there are areas that are better visible than on the other photos.
259859-5


IDK how these forums handle photos but they can be zoomed in pretty far on the computer. You might need to save them on the computer to look at them more closely.

As before, if these photos aren't sufficient I'll try to take more with even better lighting or other areas of the PCB if requested.

I really appreciate you taking the time to help with the troubleshooting.

Offline suicidal_orange

  • * Global Moderator
  • Posts: 4202
  • Location: England
Re: Troubleshooting a faulty PCB (K8 Keychron ISO)
« Reply #8 on: Fri, 15 January 2021, 00:24:51 »
If you click on a pic on the forum it goes fulllscreen, if it can go bigger there's an X thing in the bottom right corner to make it full size.  Pics look good (for a black PCB - anything is better, though white not much)  Most of the traces seem to be visible on the front which is great, working out where they go on a phone screen not so much so this is one for later.
120/100g linear Zealio R1  
GMK Hyperfuse
'Split everything' perfection  
MX Clear
SA Hack'd by Geeks     
EasyAVR mod

Offline suicidal_orange

  • * Global Moderator
  • Posts: 4202
  • Location: England
Re: Troubleshooting a faulty PCB (K8 Keychron ISO)
« Reply #9 on: Sat, 16 January 2021, 13:07:00 »
So many traces :eek:

Is there any chance there's a short down the right (with the writing readable as the pics rotate) side of the chip?  Looks like there's something shiny on a couple of pairs of legs but there's so much shine not sure if it's solder or nothing.  There's also something that looks like a metalic hair on the left in a couple of the pics which again may be nothing.

Trying to follow traces as they change from being the shiny bit to the dull bit between while running under components is not proving too successful, I can't find a datasheet or pinout for the chip (it must be out there or no-one would have known it fits the spec they need...) and all the pins seem to go up and right while your problem is left.  So either you have a break on the other side after one of the many vias nowhere near the chip, or the pins across the bottom of the chip with no traces to solder to do go there.  I will keep looking in the hope of finding three pins in a row that go somewhere but it's not looking good.
120/100g linear Zealio R1  
GMK Hyperfuse
'Split everything' perfection  
MX Clear
SA Hack'd by Geeks     
EasyAVR mod