dead TT motherboard repair attempt

[jookie]

Hello all,

I've bought a broken TT motherboard in a hope I can get it back to life, but almost as expected, it turned out not to be not a matter of changing few caps. But hey, who doesn't love a challenge?

I've cleaned the board, removed 2 MB ST RAM expansion card, removed VME cables and PCB, removed FPU (in case it was causing problems), connected working PSU (with 5V, 12V, -5V, -12V, PowerGood signal), and the board is dead, some chips are heating up.

I've downloaded the TT service manual, I've checked the '3.3 Troubleshooting a dead unit section' to find that:
- 32 MHz clock (pin 6) on CPU is fine
- STATUS (pin 74) on CPU is high
- HALT and BERR lines on CPU are high

XRES signal seems to be wrong - 4.9V (not active) and 2.3V (active) - something broken is connected to this line. I've made a list of chips connected to XRES and measured XRES against VCC on them (should be 10 k):
- U100 (68030) pin 64 - 48.4 R
- U409 (WD1772 / C302434) pin 13 - 48.7 R
- RP101 - pin 3 (pull up) - 48.8 R
- U204 (TTVIDEO) pin 41 - 48.8 R
- U111 (PAL) pin 2 - 48.9 R
- U206 (MPU) pin 105 - 48.9 R
- U403 (LS244) pin 15 (ACSI) - 48.9 R
- UA02 (5380 - SCSI) pin 31 - 49.0 R
- U303 (68901) pin 24 (ma CLK4) - 49.1 R
- U607 (68901) pin 24 - serial port D - 49.2 R
- UA00 (DMAC - C301223) pin 80 - 49.2 R
- U302 (YM2149) pin 23 === 1.6 R - 49.2 R
- U800 (SNDSHIFTER - C300579-001) pin 31 - 49.3 R
- U900 (DMAC - C301223) pin 80 === 1.2 R - 49.3 R
- U102 (LS05) pins 4 and 8 - 49.5 R

The list is sorted by the resistance, lowest first, as mentioned here on forum in the 'finding the short circuit using multimeter' thread. The lowest resistance is on U100, so my first guess would be that possibly that one is broken (even though I would personally blame some peripheral chip, which I consider as more probable, but I might be wrong).

TTVIDEO chip (U204) should be generating 16 MHz, 4 MHz, 2 MHz clocks (from 32 MHz on input), and even though it's receiving 32 MHz on input, the generated clocks are not there (it's in H or L level, but not toggling). This chip could be also one of the faulty ones, but maybe the clock generation just needs to be enabled from CPU, which is completely halted, so who knows.

I've measured the resistance of address and data bus signals against VCC, they all should be 10 k, but some of them are not:
- D9 - 560 R
- D11 - 4 R
- A4 - 331 R
- A11 - 51R
- A12 - 329 R
- A13 - 329 R
- A21 - 863 R
- A25 - 327 R
- A26 - 327 R
- A27 - 327 R
- A29 - 327 R
- A31 - 45 R
Some of them have same resistance, but are not completely shorted together (A12 + A13, A25 + A26 + A27 + A29) - they measure around 7 ohms between them. So obviously some chip is broken and holding the bus in some invalid state.

So I've made a list of chips, which are connected to the broken address and data lines (because some chips are connected to A and D lines, but as they are connected to the A and D lines which are fine, I assume they are not causing any other trouble):
- U100 (68030) - whole A + D +XRES
- U800 (SNDSHIFTER - C300579-001) - A4 + D9, D11 +XRES
- U303 and U607 - A4 +XRES
- U206 - whole A +XRES
- U209 (funnel) - D9, D11
- U900 (DMAC - C301223) - whole A +XRES
- UA00 (DMAC - C301223) - whole A +XRES
- U605 (TTSCU) - whole A
- U703 (PAL) - A24-A31
- U708 (LS244) - A4
- U709 (LS244) - A11-A13
- U710 (LS244) - A21
- U711 (LS30) - A21

I've removed EPROMs from sockets and also TTVIDEO from PLCC socket, then checked the problematic A and D signals, but the resistance didn't change, so I assume here that they are not the cause of the issues.

I've made another list of chips which are heating up a bit (or more):

A) they are heating up a bit and have XRES connected:
- U303 (68901)
- U607 (68901)
- U111 (TIBPAL16R4)

B) they are heating up a bit, but they don't have a connection to XRES, DATA or ADDR bus (so even if they would be faulty, they shouldn't disturb the buses):
- U908 (MC3486 = DS34F88) quad differential LAN receiver
- UD00 (DP8530)
- U910 (GD75188 = MC1488)
- U112 (C301837-002)

C) they are heating up and they are connected to address bus::
- U702 (C398721-001)
- U703 (C398722-001)
- U100 (68030)

The most heating up chip is the U100 (68030), which heats to ~50 degree Celsius in 30 seconds, other mentioned chips get only to ~35 degrees Celsius in those 30 seconds (and then I turn off the PSU to not stress the board any further).

The heating up and lowest XRES value lead me to U100 (68030) to being the most probable cause of the issues. The not generated clocks from TTVIDEO are suspicious, too, but I would leave replacing TTVIDEO chip to time when the data bus, address bus and XRES signals are all fine and the clocks still wouldn't be generated.

If anyone has any hints what else should I measure, or what other cause of the issues could be, let me know, please. If no other possible measurements could point to the source of issues, then I'll start by removing the U100 (68030) and then measuring the buses again (and if it won't improve, I will remove other chips, one after another, until I find the impostor ).

Jookie

[Steve]

I would suggest e-mailing Matthias from the Thunder/storm team, I don't know if he hangs out in this forum. But as far as I know, he is probably one of the most knowledgable TT experts that I know of. Matthias/Tuxie sent me some improved/reprogrammed pals/gals for my VME bus once when I had issues.

[jookie]

I would suggest e-mailing Matthias from the Thunder/storm team, I don't know if he hangs out in this forum.

Thank you for suggestion, I've written to thunderstorm[at]tuxie[dot]de, let's see if I get a reply.

[jookie]

While waiting for any hints I got impatient and I've removed U705 and U709, as they are connected to some of the problematic signals (FC0, A11, A12, A13), they are easy to get (74LSxx chips), they are not needed for basic TT boot (they are VME bus drivers, which I won't use). Removing them had ho effect on the state of the problematic signals. I was hoping that it would, as the VME PCB showed signs of resoldering pins, so I assumed the issue came from VME part, but it seems it didn't.

But on the other hand, the FC0 signal is connected to less chips then the Ax and Dx bus signals, so that at least narrows the issue to these chips (again, with resistance of FC0 against VCC next to them):
- U100 (12) - 4.1 R
- U206 (45) - 4.1 R
- U705 (8) [already extracted this morning] - 4.5 R
- UA00 (53) - 4.6 R
- U900 (53) - 4.7 R
- U605 (50) - 4.8 R

...and the lowest resistance pointing against towards U100 (68030). To test if I should replace it I decided to unsolder and lift just the FC0 pin of U100, and yes, while the pin was 4.1 R agains VCC, now the rest of the FC0 on the board was showing 10 k as it should. So I've removed the U100 and now I'm waiting for new one to arrive. All the Ax, Dx, XRES and FC0 signals are now fine - they show the value of their pull up resistors.

I've checked if the TTVIDEO now generates 16 MHz, 4 MHz and 2 MHz clocks, but they are still always H - either TTVIDEO is broken, too, or needs some init / enable from CPU to start generating. I've ordered a new TTVIDEO chip.

Some chips are still heating up a bit (~35 degree Celsius): U908, U303, U607.
Others are heating up more (~45 degree Celsius, like before): U111, U112, UD00
The warmer ones might need a replacement later (let's see).

[Steve]

Wow that's awesome, sounds like you're on the right track

[jookie]

Wow that's awesome, sounds like you're on the right track

Yes, let's see how it will continue... In some part of the journey I might encounter a chip I can't get a replacement for, or a state where everything seems to be fine, but it still won't boot. (I hope that won't happen, but it's still a possibility).

[czietz]

I didn't read the entire thread, so I'll leave it at two remarks:

- It's highly unlikely that the CPU would have suffered such a massive fault (multiple pins affected) without an external cause. Maybe the PSU failed and sent a massive overvoltage into all the chips? In that case, expect many, many faulty ICs.

- The TT shifter (TTVIDEO) generates the clocks without needing SW initialization.

Unfortunately, it's next to impossible to debug such a computer remotely.

[jookie]

- It's highly unlikely that the CPU would have suffered such a massive fault (multiple pins affected) without an external cause. Maybe the PSU failed and sent a massive overvoltage into all the chips? In that case, expect many, many faulty ICs.

Okay, noted

- The TT shifter (TTVIDEO) generates the clocks without needing SW initialization.

That's good info, new TTVIDEO chip is already ordered and to be replaced.

Unfortunately, it's next to impossible to debug such a computer remotely.

Sure, I don't expect such a thing. Getting tips like those two is good enough for me. Thank you!

[DoG]

35-45°C doesn't seem that hot. Could be working temp for some of the chips. It's when you can't hold your finger on them you should be worried.

[jookie]

35-45°C doesn't seem that hot. Could be working temp for some of the chips. It's when you can't hold your finger on them you should be worried.

Noted, thanks!