Help needed with not working switch on PCB (Leeku 1800)

[Madhias]

Recently I have soldered all switches and some LEDs to my Leeku 1800 PCB, and all switches and LEDs work except the RETURN/ENTER key of the numpad, and the LED (or switch?) of SCROLL LOCK. The LED / SCROLL LOCK problem is not that important for now, but the RETURN switch of the numpad.

So far I have re-soldered and swapped the switch, and re-soldered the diode. Today I bought a multimeter (yep, I had none), and now I am struggeling how and what to check. I did test with the connectivity check the faulty switch up to the controller, where I do find its connection and the multimeter beeps. What else do I have to see if it is connected? How do I check or find the faulty connection? Why some connections of switches / columns beep, and some not, but the switches do work all? Is it the multimeter, which does not recognize each connectivity? My plan would be just to make a wired bridge - if that solves this problem, but where to?

I may have to read more basic electronic information first!

[Madhias]

I don't know if it helps when I test the connectivity of each switch to the controller, to know about the general matrix of the pcb, like I read in sixtys guide. But I don't know why the connection of the faulty switch to the controller then works - should this connection be down completely in this case?

[Madhias]

Added a picture to the opening post. As you see the SCROLL LOCK LED section is really messed up after resoldering and soldering for about 3 times. In the past with faulty switches I had always luck and just resoldering the switch or diode solved my problem. Not this time!

[flabbergast]

Few things:
- the way I test for connectivity with a multimeter is to set it to measure resistance (usually 2k) - and then when there is a connection, is should show resistance very close to zero.
- if the switches register in the firmware, then there's not much point in testing connectivity (they're fine)
- if the connection from the bad switch to the controller is fine, try just shorting the pins of the switch with a wire to see if it registers (this is what the switch is doing when it's pressed and not faulty).
- by the way the connections from a switch should be like this: one switch pad to one pad of the diode; the other pad of the diode to a diode pad in the next switch in the row or column (this depends on how is the matrix wired); the other switch pad to next switch's pad in the column or row (depends on the matrix). Note that the 'next switch' might be also further away, as the matrix layout may not follow the natural rows and/or columns.
- you can test if the actual LED is OK with the multimeter on the 'diode' setting (it should light up a little bit)
- from the picture it does look like you may have lifted one of the LED pads, so that one might need a jumper wire to somewhere
- if the KENT LED is not one of the individually controllable ones (i.e. just one of the 'backlight' ones), then the connections should be: one LED pad connected with the same pad on all other LEDs, the other LED pad to one pad of the nearby resistor, the other resistor pad to the same resistor pad on all other LEDs.

[Madhias]

Many thanks for your answer, flabbergast! I'll come back in a few minutes (I hope), but the LED works now! I checked in Windows, when updating the firmware, and there the LED was working. Perfect.

[Madhias]

Few things:
- the way I test for connectivity with a multimeter is to set it to measure resistance (usually 2k) - and then when there is a connection, is should show resistance very close to zero.

Thanks for that, I tried it out right now, actually I like this method since it is quiet! I can't disable on my multimeter the beep sound, and this night I was therefor in the kitchen, not to wake up someone!

- if the switches register in the firmware, then there's not much point in testing connectivity (they're fine)
- if the connection from the bad switch to the controller is fine, try just shorting the pins of the switch with a wire to see if it registers (this is what the switch is doing when it's pressed and not faulty).

I don't know if the switch registers in the firmware, but when I measure from the switch to the controller and its corresponding pin, the connection is OK. I tried to shorten the switch, but it does not register.

- by the way the connections from a switch should be like this: one switch pad to one pad of the diode; the other pad of the diode to a diode pad in the next switch in the row or column (this depends on how is the matrix wired); the other switch pad to next switch's pad in the column or row (depends on the matrix). Note that the 'next switch' might be also further away, as the matrix layout may not follow the natural rows and/or columns.

That's a part where I am not 100% sure - I already found most of the rows, and some switches are really far away. The faulty switch is connected with the other switches of the bottom row, and in its column there only seem to be 2 other switches (and two empty positions), and when measuring from the diode of the faulty switch to the controller, the connection is working.

I think I have to write all connections down, to be sure how the matrix works and if there is a path not working - but I am not sure when the connectivity to the controller is OK, the path then is OK?

- you can test if the actual LED is OK with the multimeter on the 'diode' setting (it should light up a little bit)

Great, I did not see the LED lightening up, since the PCB was always with the backside on top! As I tested right now all LEDs light up when testing with the multimeter. That would have saved at least one soldering session, since I was not sure if the LED was correctly attached (its direction of the electric pole is important?). But all LEDs work again, so this problem is solved.

My plan B would be a 'ghetto' modification, and just make a bridge to the main ENTER/RETURN key - yay, I have two on this keyboard! What could go wrong? If I do so, the switch (faulty one) should be not connected at its current position, and be like a total separate switch? Would it be enough to eliminate just the diode then?

In general it is OK to have two switches attached to one position? Or could I catch fire? As I tried out right now, it works:

[DanielT]

Just a stupid question, have you checked on the firmware side ? I had a problem with one of my keyboards, right Ctrl was not working, I swapped the switch twice and messed the pads and it was still not working only to discover that the key was not mapped to anything ...

[flabbergast]

I don't know if the switch registers in the firmware, but when I measure from the switch to the controller and its corresponding pin, the connection is OK. I tried to shorten the switch, but it does not register.

- by the way the connections from a switch should be like this: one switch pad to one pad of the diode; the other pad of the diode to a diode pad in the next switch in the row or column (this depends on how is the matrix wired); the other switch pad to next switch's pad in the column or row (depends on the matrix). Note that the 'next switch' might be also further away, as the matrix layout may not follow the natural rows and/or columns.

That's a part where I am not 100% sure - I already found most of the rows, and some switches are really far away. The faulty switch is connected with the other switches of the bottom row, and in its column there only seem to be 2 other switches (and two empty positions), and when measuring from the diode of the faulty switch to the controller, the connection is working.

I think I have to write all connections down, to be sure how the matrix works and if there is a path not working - but I am not sure when the connectivity to the controller is OK, the path then is OK?

Yes, if the connections to the controller are fine, especially if the other switches on the same "row" and "column" are working.
If the connections are fine and the key doesn't register even when you manually short the switch pins, I'd guess the problem is with the firmware.

My plan B would be a 'ghetto' modification, and just make a bridge to the main ENTER/RETURN key - yay, I have two on this keyboard! What could go wrong? If I do so, the switch (faulty one) should be not connected at its current position, and be like a total separate switch? Would it be enough to eliminate just the diode then?

In general it is OK to have two switches attached to one position? Or could I catch fire? As I tried out right now, it works:

There is no problem with this, electrical or otherwise. (The only thing is that you can't program the two keys to do different things, they'll always act as one and the same switch.)

[Madhias]

Just a stupid question, have you checked on the firmware side ? I had a problem with one of my keyboards, right Ctrl was not working, I swapped the switch twice and messed the pads and it was still not working only to discover that the key was not mapped to anything ...

I have checked again with the JigOn tool, and there are just two ENTER positions (not three), and remapped them (actually changed to something else, saved, and remapped to ENTER again) - but that did not helped unfortunately.

There is no problem with this, electrical or otherwise. (The only thing is that you can't program the two keys to do different things, they'll always act as one and the same switch.)

That's great! Another user at GeekHack suggested to just connect the faulty switch to one of the empty ENTER locations above or below the faulty switch location - I tried that with a bridge handheld right now, and that works - so I will just solder two cables to the switch location below. That's so far the easiest option and I will go for it now!

In the end I really don't know what the problem could be, and I think it is not the connection up to the controller - but if it is the firmware, all other users should have the same problem?

[Madhias]

This is the solution for now:



Easy in the end, but I have learned a lot, thanks again flabbergast for your great explanations!

[elRetto]

there must be something wrong with that batch of pcbs. two of my keys aren't working either. NumLock and numpad 2

[ScarletBegonias]

This is bizarre!

I've been working on my Leeku MX1800 L3 over the past few nights and last night I found that my Print Screen key was not registering. I dug my DVOM out and both the switch and the diode tests OK. I replaced them both, just for grins and still no Print Screen.

It still has the original firmware, so I hope to have a look at that later tonight, or tomorrow.

[ScarletBegonias]

I'm not sure if I had a problem with Linux not recognizing the scan codes the past couple of nights, or if the on-line tool I was using was causing the issue, so I'll have to look a bit deeper into the issue.

I'm on a Windoze box tonight and the good news is that JigOn recognizes all of my keystrokes. I also downloaded Switch Hitter from Elite Keyboards. There is no 1800 layout to choose, so I left it at 104 ANSI. It doesn't have any problem seeing my keystrokes, either.

I did notice some anomalies: PgUp is mapped to End, Insert is mapped to Delete, Delete is mapped to home, nothing is mapped to Insert, etc. But, I should be able to easy enough to correct these issues with JigOn.

tldr: If you have keystroke issues with your Leeku PCB, fire up JigOn, before you make a big project out of it.

[phosphorglow]

(...I just want to comment on the photography...: lines/angles/framing, even for simple technical reference snapshots...)