Ergodox PCB problem

[GEIST]

I unsoldered the LEDs from the red marked switches today. Since I'm finished with this the keys in the green marked column behave weird: If I press one of them, the key sends out his whole row. So for example if I press the M key, it sends out 0nm,./.

dox.jpg (605.62 KiB) Viewed 3661 times

I already flashed some old firmware, to see if it has something to do with it, but it seems like a hardware problem. Looks like I broke something, but I don't know what and how.

[vvp]

Interesting. Do all the other keys work correctly?

[GEIST]

Maybe it sounds strange, but they do. Unfortunately I don't own a multimeter (or would know how to use it) to find some shorts.

[vvp]

Well, looks like you somehow managed to connect all column wires through a diode to your green column. Or maybe something else.
Maybe a picture of the other side of the PCB?

[GEIST]

Uh I have no idea how it could have done that. Unfortunately I would need to unsolder all switches on the other side, to be able to remove the plate. And you've seen, what I've already done to this PCB using my unsolder skills.
But only guessing won't help (I guess), so maybe I need to unsolder the switches, even if I'll destroy the PCB completely this way.
I thought maybe someone with a better understanding of PCBs could check my photo and the other side here

and tell me what to search for, cause I don't understand if I accidentally cut some trace or added one.

[anorak]

Maybe you can provide a better picture from the three switches you unsoldered the LEDs from? The red marked ones. The LED holes look somewhat dirty, maybe there is some solder left and causes a short.

[GEIST]

Good call and the PCB IS dirty. It doesn't look pretty good what I've done to the solder points.



By the way the keys worked yesterday for about 20 minutes like they should do, so I guess it is just a tiny contact which was added/destroyed

[vvp]

By the way the keys worked yesterday for about 20 minutes like they should do

Ok, that rules out what I though it might be. I would probably check better the area around Teensy/MPC. At the places where a plate is not present, you can check a PCB well for bad connections (or missing connections) with a strong light source. Look at the light source through the PCB. Well, it will not work that well with Ergodox PCB since it has too much copper fill. But you can give it a try. The flux residue can often be cleaned with alcohol if you want to see some traces better. Do not scratch the PCB. Use something soft.

[GEIST]

Thank you vvp
Unfortunately I can't see through the PCB, even in bright sunlight. But I'll get some isopropanol and clean it, to find something.
I checked the area around the Teensy/MPC but didn't found anything suspicious.
I thought maybe I accidentally connected something on the other side of the PCB, but since the switches are plate mounted on the other side it seems impossible to unsolder them, but I would have to, to see the solder points on the other side.

EDIT: If someone would ask me what I would need to change at the PCB to make one key send out the whole row I had no idea. That's my main problem. I barely understand how the PCB works, what the switches and the resistors do, but I have absolutely no deeper understanding what would happen if I change something. Maybe there is somebody how do understand more of how everything works and what could have been happening to allow this weird behavior.

[anorak]

I do not know the layout of the ergodox pcb. But for my handwired keyboards, a matrix scan works like this:

All pins connected to columns are configured as inputs with pull-up resistor activated. So input is read as high for now.
All pins connected to rows are configured as inputs with pull-ups deactivated.

When the matrix is scanned, the first row pin is configured as output and set to low (source sink). When a key in this row is pressed, the connected input pin of the column is pulled to low and is read as low now. All input states of all column input keys are checked. A low state for a column input pin indicates a pressed key for this row and column. Repeat for all other rows... Depending on the layout and firmware, rows and columns may be swapped.

There must be something that connects column 4 to all rows 1-5 when pressing one key in this column.
I wonder how this can happen by desoldering LEDs in column 6.

[GEIST]

Thank you anorak. I guess this HAS to be the problem, but I can't find something which looks like this. There isn't something on the bottom side of the plate and I tried to see something by looking between plate and PCB, using a flashlight, but couldn't find something suspicious too.
I already removed the three switches in row 6, but even from the top side everything looks pretty ok.

backsideswitches.jpg (622.78 KiB) Viewed 3492 times

So I guess the whole issue doesn't have a direct connection to this three switches in row, maybe something else happened, while unsoldering the LEDs in row 6.
Do you think it would help to buy a multimeter? I'm willing to try everything, since I can't use it at all in this condition.

[GEIST]

Just to finish this thread: Czarek from FalbaTech asked me to send it in (I have bought it used from someone here on the forum). The PCB had some kind of malfunction and he switched the PCB out. I had to pay for the PCB, but I didn't had to pay for his work, which seems more than nice. Especially since I don't bought it direct from him.