Trying to repair a G80-11900

[berserkfan]

I decided to set my hand at repairing this board in order to learn to follow PCBs with a multimeter, but once again I've gotten lost.



There are 5 unresponsive keys here.  Aquakeytester can't detect them.

This is a single sided PCB so nothing on the other side.

All 5 unresponsive keys have continuity with other responsive keys if there is a visible trace from one joint directly to another joint. So I guess none of the visible traces are broken inside.

But there are still a lot of solder joints not obviously connected by a visible lead to another joint. I don't know what they are supposed to be connected with, so I can't test if that connection exists. So I can't find where the continuity is broken, and proceed to bridge that with a wire.

Anyone would like to try his hand at this Geeky Where's Wally game?

[suicidal_orange]

You know I'm always up for these!

Does this board have diodes?  There are spaces that look like there should be, but some of them aren't connected to anything and the top switch you've put a box around is connected to the "diode" you'd expect to be for the switch next door.  Also some switches are directly connected to another switch on both pins so I'm guessing no diodes.

If there aren't any diodes then those spaces must be for other side connections, so it's not a single sided PCB and will be more fun to work out...

In this one I'm checking how the designer is thinking - it's friendly to point the hidden connections at each other, but if they're all short and straight that's ok too.  Please can you confirm continuity between the same coloured dots?



I've also circled a dodgy looking solder blob that I'm pretty sure intersect with the connection to the other dead keys - could be a factory repair that's failed?

[berserkfan]

Thanks suicidal orange.

This is a keyboard that has been used by the owner for a while, so I think any dodgy solder blobs haven’t caused problems.

Also in real life it looks well soldered.

You’re right! Board does not have diodes! All the things that look like diodes, turn out to be just metal pins!

[phoenix1234]

But there are still a lot of solder joints not obviously connected by a visible lead to another joint. I don't know what they are supposed to be connected with, so I can't test if that connection exists. So I can't find where the continuity is broken, and proceed to bridge that with a wire.

The trace looks fine. There is nothing wrong with them.
At least I can tell based on my old PCB.


I think the problem is like what suicidal_orange highlighted or probably it is just switch failure.

[suicidal_orange]

I've tried to map the trace that joins all the broken switches but it doesn't seem to connect to the controller anywhere - is there definitely nothing on the other side of this PCB? The bits I can make out don't seem to have changed between rev 01 (Berserk's board) and rev 02 (Phoenix's) so this was the best pic to trace it on



It seems that the bottom switches near the middle of the board on this trace work (although I won't pretend to know how) so if you touch a wire from the other pin on a broken switch to the connected pin on one of those it should work.

[phoenix1234]

Here is the full pictures of the board with the traces lead to the controller.


This is a single layer PCB, on the other side, there are only a few fixed jumpers.

[xorian]

You’re right! Board does not have diodes! All the things that look like diodes, turn out to be just metal pins!

I recently picked up a G80, and a lot of them have switches with integrated diodes.  So those extra two pads below each switch that look like they're not connected to anything may be connected to a diode that's inside the switch.  With a multimeter's diode test setting you should be able to check this by trying both directions between those two pads on one switch: one direction will show a connection (mine beeps) and the other will not.

[user 18]

I've tried to map the trace that joins all the broken switches but it doesn't seem to connect to the controller anywhere - is there definitely nothing on the other side of this PCB? The bits I can make out don't seem to have changed between rev 01 (Berserk's board) and rev 02 (Phoenix's) so this was the best pic to trace it on

Show Image

It seems that the bottom switches near the middle of the board on this trace work (although I won't pretend to know how) so if you touch a wire from the other pin on a broken switch to the connected pin on one of those it should work.

You have a vertical green line in the area between the pads for jumpers on the far side of the board that crosses between two traces. The lower left quadrant of your green line isn't actually on the same trace as the rest of it.

[suicidal_orange]

I've tried to map the trace that joins all the broken switches but it doesn't seem to connect to the controller anywhere - is there definitely nothing on the other side of this PCB? The bits I can make out don't seem to have changed between rev 01 (Berserk's board) and rev 02 (Phoenix's) so this was the best pic to trace it on

Show Image

It seems that the bottom switches near the middle of the board on this trace work (although I won't pretend to know how) so if you touch a wire from the other pin on a broken switch to the connected pin on one of those it should work.

You have a vertical green line in the area between the pads for jumpers on the far side of the board that crosses between two traces. The lower left quadrant of your green line isn't actually on the same trace as the rest of it.

Damn, and it's bedtime so too late to fix it!  That explains my confusion at it not going anywhere...

[berserkfan]

You’re right! Board does not have diodes! All the things that look like diodes, turn out to be just metal pins!

I recently picked up a G80, and a lot of them have switches with integrated diodes.  So those extra two pads below each switch that look like they're not connected to anything may be connected to a diode that's inside the switch.  With a multimeter's diode test setting you should be able to check this by trying both directions between those two pads on one switch: one direction will show a connection (mine beeps) and the other will not.

G80-8200 and 8113 have real diodes. Not the 11900s.

[suicidal_orange]

Looked at it again and found the closest working switch - the green dot pretty much in the middle should connect to the top half of the green line above - please poke it with a wire from the other pin to check as I may have gone wrong again

This trace runs off the left side of the pic so still don't know where it goes to the controller but that's better than ending randomly!  The soldering on the "diode" on the switch to the right doesn't look great so try reflowing that first, but if not run a wire and all should be good.

[berserkfan]

sorry for not replying; something was weighing on me heavily these few days. Got betrayed by a 'friend' and now just kinda disillusioned by it all. I'll get to work testing that PCB in a couple days; hope to be feeling better then.

[XMIT]

I just fixed a magnetic valve keyboard that had a gap in a trace. From that board, I learned a few lessons:

- try to isolate the scope of your repair to just what is broken. In my case I had a dead row, and, there is a single trace that goes all the way across the board to detect rows.

- a multimeter in continuity mode (goes "beep" when you touch the leads together) is your friend. That same multimeter in resistor mode (say 0-20 ohms) can tell you when a trace is weak. This can explain some glitchy key switch behavior.

- you can (carefully!) scratch away some solder mask to detect continuity along a trace. This lets you hone in on where a trace is broken and where a repair is needed. When you are done, you can paint over with clear nail polish to prevent the trace from corroding. The nail polish comes right off with acetone. Test this on an unused part of the board first to see if acetone damages the solder mask. It should not but it might.

When you do find a gap, sometimes you can just bridge it with a solder blob. I used solder and a tiny piece of wire. I really wish I had some flux for this but it seems to be holding up.

Here are some photos of the successful repair that I did. It is still a work on progress. (Turns out I'm fresh out of clear nail polish.)




@berserkfan: Don't give up even when the odds look overwhelming.

[berserkfan]

sorry for the late reply guys; someone I though was a friend screwed me over bad recently and I didn't feel like coming back to the forums

I followed suicidalorange's advice, and replaced the iffy switch on the right of the green dot switch, and problem was solved! The weird thing was, that switch worked fine on its own. It also worked fine with multimeter testing. Just that all the other five faulty keys traced through it, so I'm guessing that there is something wrong I don't understand going on. The iffy switch had one loose solder pad for the 'diode' pins. I suspect that when this switch was pressed, it was OK. But when other switches were depressed but this switch was not, the contacts were not made so no electrical signal passed through it.

thanks to all your help I learned much about reading PCB lines along the way. now Gutz has levelled up again!

btw anyone who is reading this, if you can see my signature, just keep in mind that there are people on this forum whom you should be wary of. just because someone has been around for a long time, doesn't mean that he's a good guy. It just means that he likes this hobby.

[XMIT]

Glad to hear this was fixed. Often times you can isolate a failure to a particular row, column, switch, or region of switches. Use this information to help cull the list of potential problems and move you toward a root cause.

[suicidal_orange]

Glad to see you got it working, scant conciliation for the loss of a friend though.  I thought he was a good guy as he taught me how to debug matrices, guess I was wrong

[berserkfan]

Glad to see you got it working, scant conciliation for the loss of a friend though.  I thought he was a good guy as he taught me how to debug matrices, guess I was wrong

Suicidal orange, there is a big difference between someone who will teach you anything just because he loves to talk and show his knowledge, and someone who is genuinely committed towards the welfare of his friends. You also see that in teachers who lecture and walk out of class vs teachers who stay on their private time because they care about their kids.

Some people with good language skills are on the autistic spectrum. Give them a direct question or comment on a topic of their liking and they can express themselves for hours. But it doesn’t make them more honest or ethical or better people. So don’t be doing them altruistic favours such as bailing them out when they’re financial trouble, making deals intended to benefit both (such as sending someone money on promises that he can send you stuff later or perform services for you) or they’ll disappoint you because they plain don’t value friendships and can’t care a damn about making good on what they owe their friends.

Rest of my reply is sent via PM.

As for the G80-11900, if you can think up WHY replacing a working switch made 5 non responsive switches work, I'd like to hear it. You were right, but I don't understand WHY. So it's about learning how to fish rather than getting a fish from you.

[neverused]

The "diodes" in the switches are not diodes, but wires. If one of the pads is faulty any subsequent switch after that fault will not work. I just took one of these apart last night and noticed this.