Next I removes all of the offending capacitors an measured them. Several of them had very clearly gone bad. This one I'm testing should read 10uF but instead reads 1.5uF.

Capacitors!

These three, to be exact.

But only a few of them.

A short while later it is repaired! But will it work?

This processor will do!

Hhmm... the computer will not boot. There is no beep or POST code. This processor has been tested previously same is known to be good.

Upon closer inspection of the troublesome slocket I found several broken traces. There is more repair to be done!

Zoom and enhance.

It's only a small hole, just flick the edges of the traces up 45 degrees with a needle so the electrons have a ramp to jump it. 🤣

One trick to remember is, if you get multi stranded wire, with thin strands, just strip an end and splay some strands out, while they are still attached to something you can hold, you can tab four strands down one side of the break, then you can cut them off and solder them the other side.

edit: also if you stagger the points where they are soldered up and down the lengths of the traces a ways, you have less trouble with running the blobs together if your solder wire and/or iron is big and clumsy.

Clean by scrubbing with alcohol and scrape away at the solder mask to reveal the traces.

I have a spool of copper wire wrap that is 0.17mm in diameter. I thought that this would be a good choice and prepped the area for work.

I found the wire was too thick and the surface was heavily oxidized which made it difficult to solder as well, even with flux. There was also the risk that the connection would be very poor.

Instead of used bits of file pure copper solder wick. These strands are 0.7mm. Perfect!

Here is the result after repair and cleaning.

The scale is very difficult to convey and the job is equally difficult to photograph. Here is a photo next to my thumb for comparison.

There are other damaged traces but I don't think any of them are broken. I'm going to try using it again and see if it works.

Well done, very neat and clean... ima send a small container, just a half size sea-can, of boards with a few dings to patch... j/k

Hey, nice job so far! Hope it works!

The system will now try to start but there is no startup beep and the screen remains blank. My diagnostics card reports code C1C0 which suggests that the system is halting as it tries to detect the system memory, I think. The error code is the same when I try the slocket in two different motherboards. This CPU works fine in other slockts and I know the RAM is fine.

I've looked over the slocket under my microscope for some time and checked continuity on every trace that looked suspect but there were no other severed traces.

It's possible that one of the pins from the socket is making a poor connection. Some of the joints looked a little suspect but reflowing them didn't help.

I'll have to take some time to think about what to do next.

You didn't actually sayyyy that you patched that other small crater off to the right on the first damage pic. Just wondering... also the nature of those, they don't look too much like those particular traces overheated then blew, they look almost like they could have blown through from beneath.

None of the other damage to the board was enough to severe another trace. I scraped away at anything suspect and checked continuity between anything that looked suspect, including what appeared to be another crater; that trace was deformed but not broken.

Can you try the slocket on another board to verify that it works?
I have seen C1C0 message when slocket has connection problems to Slot1 connector on the board. Since it does not have retention brackets you can try to push the slocket in the Slot1 either side during power-up and see whether it will boot.

Excellent work anyway, but what experience do you have with the quality of Hyncdz capacitors? I confess I've never heard of this brand before.

Not that I think this info will solve the problem but in all 5 versions of the same 'Matrix' slotket that are currently on offer on ebay (including one that is presented as new) the C58 position is left vacated. And there seems to be nothing connected in that section from C58 => R13 anyway.

On the archetype A-Max version of this Matrix card there was a mosfet on Q1 and the rest of the string was also populated.

I poked at this a bit more today without success. I tried removing the capacitor as suggested. Checked a couple more traces and in different boards but it refuses to work. It's possible that there is some other damage inside the PCB that cannot be seen.

The solder joints checked out fine under my microscope but I'm not reflowing 370 pins just to be sure. I'm certain that the problem is elsewhere.

Ydee wrote on 2021-09-08, 08:49:

Excellent work anyway, but what experience do you have with the quality of Hyncdz capacitors? I confess I've never heard of this brand before.

I don't need to use high quality capacitors for jobs like this. Especially if there is a high chance that they will get thrown away.

These are generic capacitors from Aliexpress and work fine. I do test them first for capacitance to make sure that they aren't defective or fake but I've never encountered a bad part.

Any chance of a picture showing the whole of the PCB? Given your magnifier then I can't imagine anyone will see anything you haven't, but might be interesting. I think the traces you repaired so far are on the data bus. Have you got a slot - socket pinout to check continuity? Any chance the damaged surface traces ended up shorting to an internal power/ground plane?

Success! I was poking around at it today, checking various traces and it started working.

I don't have any more photos of the repair but here are some shots of it working. Of note, the slocket works fine with and without a capacitor on C58. I don't think it's needed though, when comparing it to other identical slockets on eBay.

snufkin wrote on 2021-09-08, 16:19:

Have you got a slot - socket pinout to check continuity?

I do not. Can you link one?

Kahenraz wrote on 2021-09-08, 16:43:

Success! I was poking around at it today, checking various traces and it started working.

Nice.

Kahenraz wrote on 2021-09-08, 16:43:

snufkin wrote on 2021-09-08, 16:19:

Have you got a slot - socket pinout to check continuity?

I do not. Can you link one?

Sorry, no, I was only asking as I thought it might be useful for troubleshooting and that someone else might already have done the work to make up a table. If you hadn't already fixed it (dirty pin contact?), and such a table doesn't exist, then I might have had a go at making one. I guess it'd be a fairly tedious job of just going through the slot and socket pinouts and matching them up.