Trying to fix Mikro's wrecked TT board

[dml]

Hi all,

So recently Mikro offered up a TT mainboard with mysterious damage. The PCB has a corner missing :)

The original post was here: viewtopic.php?p=126340#p126340

We decided it would be fun if I wrote up progress - if there is any progress - on a repair attempt. I had a read through some other TT revivals on here including Jookie's journey so I'm prepared for endless problems :p


I have taken a lot of photos of the board in its arrival state and spent a bit of time inspecting for defects. I did notice a few odd things already.

Aside from the impact-style damage to different parts of the board, there are some clear signs of unfinished repair effort - one ceramic capacitor is missing but two nearby have one leg partially lifted. The lifted cap legs is the one that makes me think someone had a difficult time trying to debug this board.

Screenshot 2025-04-06 at 09.33.39.png

After inspecting the board, I lifted out all the socketed ICs, resistor networks, jumpers etc - anything not soldered down - and inspected the sockets. Most of the PLCCs look ok-ish but a few have sketchy contacts and one just crumbled while I was using the extractor.

There is some corrosion and minor breakage at the ROM sockets:

Screenshot 2025-04-06 at 09.34.18.png

There are minor defects all over which need a bit of attention before I try to do anything involving power. e.g. a few places where longer pins are flattened over and have maybe contacted or might do - some bent header pins, PLCC sockets with misaligned/flattened contacts.

Screenshot 2025-04-06 at 09.35.05.png

After that, I took note of all the bodge wires and spots where it looked like someone had been visiting with solder/flux and then gave the board a bath in flux cleaner.

Screenshot 2025-04-06 at 10.02.22.png

Screenshot 2025-04-06 at 10.01.58.png

I have a question though for anyone out there knowledgeable on TT hardware. Will this machine accept a PGA 68030 directly on the processor socket without the 32MHz daughter board? Is this really a CPU socket or is it some special Atari pinout which just happens to use the PGA128 because they had them in stock? I could maybe simplify things a bit by leaving the daughter board until later.

Screenshot 2025-04-06 at 09.53.12.png

[CiH]

Subscribed!

[dml]

Begin with deleting the smashed socket (oops, the smashed bit is off the top of the pic - nvm)

Screenshot 2025-04-06 at 12.15.30.png

Looking around here for replacement sockets for the ROMs and a couple of sockets for the ACIAs. I'll check the vias at the missing ACIA first, then remove the second one and fit sockets for both. Pretty sure I have two good ACIAs in a box somewhere. I can test the extracted one later.

[rubber_jonnie]

Good luck, it'll be great to see this running again.

[dml]

Good luck, it'll be great to see this running again.

:dualthumbup: optimism is good :D

I got slightly confused at the custom IC number in the TT Shifter socket but it's apparently just an older part code. Found some of the sockets here but DIP32 are on order.

The MFPs seem to be the same type as Falcon - I can probably test those. Same for the ACIAs. Will also test the RAM since it is socketed. Not sure there if there are other chips I can test but that's a head start on excluding some problems.

[dml]

Interesting...

I made an effort to clean up the lifted caps around the power connector. I used flux, low-melt solder and 300'c, which should have made it easy.

Every one of those caps disintegrated as soon as I applied heat.

Screenshot 2025-04-07 at 11.03.50.png

Screenshot 2025-04-07 at 11.03.40.png

Mikro wrote that he cleared a short somewhere in this area - and one of the caps was already missing. I don't see anything else missing or changed in that area, other than the caps.

If removing C104 cleared a short - it suggests the cap itself was the short. Blob caps which go short and disintegrate with heat suggest 'tantalum'. But these are 104s/100nF and look like typical axial ceramics.

Now I don't trust these caps at all and there are tons of them on the board. yay.

For now I'm going to assume:
- these are just a very self-destructive form of axial ceramics which can also short
- so long as the others are not short, they are probably 'okay'

Will check the area for shorts and install some new caps, moving along....

[dml]

Reworked the area next to the power supply - replacing with higher value caps but they are at least axial.

Some pitting (?) next to C101 - don't think there was a trace here but will check the schematic.

Screenshot 2025-04-07 at 13.00.08.png

Removed the IKBD ACIA, ready for sockets.

Screenshot 2025-04-07 at 12.59.55.png

[Badwolf]

I have a question though for anyone out there knowledgeable on TT hardware. Will this machine accept a PGA 68030 directly on the processor socket without the 32MHz daughter board? Is this really a CPU socket or is it some special Atari pinout which just happens to use the PGA128 because they had them in stock? I could maybe simplify things a bit by leaving the daughter board until later.

With the caveat that I'm not knowledgable on TT hardware (or exactly the purpose of this daughterboard) the schematic on dev-docs.org seems to suggset the pins are in the same layout at least. The ones visible in your picture look like they are too, but I suppose you could buzz 'em out.

BW

[dml]

With the caveat that I'm not knowledgable on TT hardware (or exactly the purpose of this daughterboard) the schematic on dev-docs.org seems to suggset the pins are in the same layout at least. The ones visible in your picture look like they are too, but I suppose you could buzz 'em out.

Yes it seems like it should be 1:1.

I had a look at the schematic last night (and the one that includes the daughterboard) and it shows the 030 mapping directly to the PGA socket and then almost directly to the 030 on the board.

I'm not sure yet exactly what is going on within that board - is it all just for clock doubling or is there more to it? Seems like a lot of logic there.

If I can't find a good reason not to - after more poking around - I think I'll just plug a known good 030+882 in and see what happens :)

There are some other chips which look like they are minumum-required for a test (GLUE, FUNNEL/DCUs, probably TTVIDEO, IKBD probably at least one MFP, maybe others..). Ideally I'd start with the minimum of chips and check the data/address lines as each one is added back to maybe catch a bad IC early.

It's a complicated circuit though, could be that it needs most of the ICs before the CPU will move.

I can begin with populating any chips I know are good by other means. I'll read the service manual first though, in case there are clues to the minimum set.

[Badwolf]

I'm not sure yet exactly what is going on within that board - is it all just for clock doubling or is there more to it? Seems like a lot of logic there.

Yeah, I tried to infer it too and mostly it seems to be resynchronising bus control signals to the old clock, I think.

It takes an external 32MHz clock input so my guess is the onboard one is 16MHz and the control signal timings are perhaps still transitioning when the 'middle' edge of the 32MHz clock might sample them, so make sure they all occur on a common clock edge instead?

Difficult to know with PALs, though.

BW