Thermalwrong wrote on 2023-09-27, 20:29:Today I made some real progress with my MS-6168 motherboard which would not power on after I bought it from a scrap lot and reca […]
Show full quote
Today I made some real progress with my MS-6168 motherboard which would not power on after I bought it from a scrap lot and recapped it. I made the mistake of not testing it first and not recording my progress a few years back so it's interesting to pick it back up and figure out what I was doing.
ms6168-repair-2023 (1).jpg
Maybe it had one visibly blown MOSFET running the CPU power section because I'd swapped that out but it still wouldn't power - pulled the other MOSFET, which tested as two resistors so that's fautly too. It has a high-side and low-side MOSFET, run by the SC1153CSW controller chip. From checking with my thermal camera, I knew that at least one MOSFET got really hot with no CPU fitted.
Decided to replace with new-old-stock original parts from a UK seller and these new MOSFETs test okay so I fitted them.
Ran the board for a bit and initially it was giving "--" or "8C" which means nothing on my tester card, I was worried that maybe the northbridge was damaged or something else broken.
Power consumption with the new MOSFETs fitted is ~13w using the PicoPSU and the bench power supply, so I can easily see if there's anything crazy going on with power draw.
The board has sat for several years though so I thought I'd try reseating the CPU card and setting the multiplier right for the CPU. There we go, now it boots up to "D3" - very curious that it does not beep at all, even with the separate speaker fitted. I bet I broke a trace around there someplace.
But with RAM fitted, the board now POSTs and there's output on the screen which has no errors, so the following parts are working:
Northbridge - great, there were some scratches on it from where the heatsink was removed so I was worried. Heatsink is now re-fitted.
Voodoo 3 - working and the fan works properly, power consumption seems to be normal
Voodoo 3 memory - good so far, no noise on the display
CPU & Slot 1 socket - good, the socket did need contact cleaner sprayed in it to work properly. Just testing with a P2-350 for now but would like to run a coppermine CPU on this board once it's working. In fact that's why I have 2x MS6168 boards, one is a V1 and can only do katmai at best, but this board when it works should support coppermine as it's a V2.
Memory - also good, counted up to 128MB
Then I decided to re-check the thermal camera to see what's going on - one of the MOSFETs is running at 120C while the other is cold - finally I'm using my thermal camera for the job it's intended for. It let me to zero-in on where the faulty components were visually.
ms6168-repair-2023 (4).jpg
That's bad though so I shut it down and I pulled off the MOSFETs to re-test them, they're both fine so lets look at the controller next. The SC1153CSW gives a good reading of mega-ohms between ground and DH - driver output for the high-side MOSFET, when it's running the DH pin shows something like 4.5v. Checking the DL - driver output for the low-side MOSFET, that reads 22ohms from it to ground / source. When running, the DL pin reads 0.05v
With the SC1153CSW removed the 20 ohms short is gone but very curiously, the corresponding pins on the DC/DC controller also do not read as short once it's removed from the board. So I don't know for sure whether it's a shorted component or a short somewhere else on the board.
ms6168-repair-2023 (2).JPG
Checking my good MS-6168, the low side MOSFET's Gate leg does read mega-ohms to ground / source.
I really hope the new SC1153CSW chip fixes this!!
I'd post this in the other MS6168 rebirth thread but it's not been posted in for a long time so I don't wanna bump it. My experience with this board is different any-ways - I got it in a scrap lot something like 4 years ago and after replacing the obviously bad caps, it'd turn itself off like it had a short. I had no thermal camera back then and less diagnostic ability so it's sat in a box for a long long time, not possible to meaningfully test it all those years.
I got the new SC1153CSW and installed it - the reading between source and gate went from 20 ohms to 3+ mega-ohms with the chip replaced!
The MOSFETS were reinstalled and that took ages, I was worried I overheated them but they're fine.
The one capacitor I destroyed its ground leg of, I ran a thick wire on the back of the board to the next closest capacitor's ground leg.
Upon initially setting this back up I was really worried because it would not start up - that's my fault but not damage to this board! oh no, the other day I forgot my bench power supply has two outputs and both were hooked up. I was testing a laptop so the PicoPSU got 24v volts and started smoking - I thought it was just a TVS diode I destroyed but no, it no longer turns on.
It's tragic and dumb that I broke a PicoPSU because of a silly mistake like that, but the PicoPSU is replaceable and this board aint.
So here it's running from an ATX power supply - tested CPU voltage before installing a CPU and that's a steady 2v, the MOSFETs are both running cold, where previously the high-side MOSFET would be 80C by now.
Put the CPU in and it's POSTing 😁 It's already POSTed before but now the MOSFETs run COLD with the CPU running, as it should be! In the previous test the high-side MOSFET hit a max of 120C before I shut it off.
The board is fully re-capped and needs the Voodoo 3 fan quietened some along with a better BIOS, right now it's got the Packard Bell AMI bios which isn't great.
I'm guessing that before this board was sent off for recycling, it just stopped working one day - it ran for enough years that most of the caps were bad, then one or probably both MOSFETs blew and damaged the vcore dc/dc controller chip along with it, or maybe the controller started it??
I'll have to make a proper build around this soon 😀 Maybe a video memory upgrade too...