bruteforce - find where the signal is supposed to go and jump signal with wires directly from the socket

Check the components near the port, I had once a cracked component.

After 2 failed attempts to repair this keyboard/mouse connection issue, I'm starting to think I may either need to just dump the board off "as-is", or go nuclear and try and tone out the data and clock signals further back on the board and jump all of it, if it's possible.

I tore it all down today and reworked everything...

• I toned out the data and clock connections to the IC shown in the image below right from a new PS2 connector.

• I toned out the +5V and ground connections to the cap shown in the image below right from a new PS2 connector.

• I reworked my bodge and made sure I had a solid connection to replace the missing via and lifted trace.

• Everything in the area shown was re-flowed.

All connections tone out good, but the Issue persists.

I have it "rigged" right now using a small piece of foam under the corner of the board to flex it ever so slightly to make a good connection.

All of this tells me the problem lies outside my work area (what you see in the photo).

This is one of the most frustrating repairs I've had to attempt in a long time. I may build a full external KB/M connector and put headers on the MB to connect it to. I like this board, and everything else works great, I really don't want to part with it, especially considering how much they are to replace.

Well, I made the external portion of my "nuclear option".

This quick and dirty connector is what I would attach outside my case, connected via wires to headers on a PCB (containing whatever necessary circuitry is needed that I may be bypassing) that I would put on the MB, then connect to wherever the signals go outside the possibly damaged area on the board. Hopefully...I'm still not 100% convinced the board has physical damage I can't see...so I'll likely look one more time before going nuclear.

It's been many years since I did any Jerry-rigging like this. 😜

I won't have time to dig into it for several days, so my plan of attack has plenty of time to change.

One of those SMD components might be cracked, no amount of reflowing will fix that. Cracked component would explain pcb flex fix.

rasz_pl wrote on 2024-08-04, 07:05:

One of those SMD components might be cracked, no amount of reflowing will fix that. Cracked component would explain pcb flex fix.

Yeah, that thought had crossed my mind, since I've only toned it all out on the board. Could be one of those inductors, indicated by the "L", so I assume they're inductors. I'll have to check those....again. 😀

Main problem with cracked components/solder joints is they seem fine when you put a probe down 😒 you have to measure away on the other end of tracks leading to those components and flex pcb to be sure.
For those inductors you can just jumper over them during testing.
Sadly I dont have access to TX97 boardview. both mouse and keyboard clock/data tracks should go directly to HT6542B in lower left corner http://www.s100computers.com/My%20System%20Pa … 0Controller.pdf U7 might be the hex inverter used for output. Also 4 resistors pulling those 4 signals to 5V is crucial for proper operation.

rasz_pl wrote on 2024-08-04, 20:23:

Main problem with cracked components/solder joints is they seem fine when you put a probe down 😒 you have to measure away on the other end of tracks leading to those components and flex pcb to be sure.
For those inductors you can just jumper over them during testing.
Sadly I dont have access to TX97 boardview. both mouse and keyboard clock/data tracks should go directly to HT6542B in lower left corner http://www.s100computers.com/My%20System%20Pa … 0Controller.pdf U7 might be the hex inverter used for output. Also 4 resistors pulling those 4 signals to 5V is crucial for proper operation.

Yeah, I really wish I could find a schematic or field repair guide for this board, but those are rare to see public these days.

The odd part is the problem effects both ports, not just one or the other, either they are both on or off.

I am assuming that means it's not the data line. The +5V and ground are common among the two and the fact everything literally shuts off makes me think it's one of those two, but I'm also not sure if they share the same clock signal and it's just split. On the board, the clock signal goes to two different legs of the CN1201 capacitor network seen in the blue/green box, the other two legs are the data lines.

I think I'm going to scope around and see what I see.

all 4 signals go to HT6542B so just locate which legs