exxos's Atari, Amiga & retro forum

I managed to get the CPU in. It's not at all pretty as the legs have become so fragile that just the act of lightly touching them with the soldering iron during drag soldering (and I mean LIGHT) causes them to bend out of shape. In some areas I had to try to solder them one at a time under the microscope. I'd never have been able to do it otherwise.

Anyway, I have some activity. The CPU reads a few bytes from ROM and then gets stuck, possibly in a write cycle. It's not halted and there's no bus error but it's just sitting there.

I see the same data at the ROMs and the pins on the CPU.

Anyway. I think that I'll stop there for the night.

Actually, I do see an extremely short blip on the /BUSERR line. It can't be more than a few nanoseconds. Is this enough to trigger the CPU?

I see about four or five 32 bit reads from the ROM before it stops. So that would be the CPU reading the initial state and jump address wouldn't it?

D17 doesn't look great when generated by the ROM. I wonder if that's faulty.

I see the same activity on the address lines both on the CPU pins and the ROM, so they are connected OK.

I'm seeing the same data on the RAM data pins as on the CPU and the ROM.

The RAM chips have RAS and CAS.

Anyway, here's a picture of the CPU installation.

As I said, it's not pretty. There are no shorts between pins and all the connections tone out. I double checked the iffy looking legs.

I think my next course of action will be:

• Tone out all the data lines to the ROMs. This will prove that any data read from the ROMs is getting to the CPU.
• Tone out the address lines to all the chips the CPU address bus is connected to.
• Tone out the address lines the ROMs are on, which are connected to the VLSI chip.

Seeing as the CPU is only getting to the first part of the start-up before it starts to run code it's best to verify that it's getting the correct addresses read from ROM first. This is both seeing the same data that the ROMs are outputting but also the ROMs getting the correct addresses the CPU is asking for. This is the foundation I can then use to move forward.

Unfortunately, the leg spacing on the CPU is such that I can't get test clips on them without shorting the two pins either side otherwise I'd just use the logic analyser.

Things look a bit odd around the top right corner marked 83. some of the pins look like they have moved over a bit from where they should be, or is the camera angle playing tricks on me?

There was a fair amount of battery leakage around/in the ROM sockets so I wouldn't be surprised if marginal things are lurking near/under them.

rubber_jonnie wrote: 12 Mar 2024 11:45 Things look a bit odd around the top right corner marked 83. some of the pins look like they have moved over a bit from where they should be, or is the camera angle playing tricks on me?

No, they're very skew wiff, but they are connected correctly. That's the best I could get and the pads were starting to lift there so it's best left well alone otherwise the PCB will be beyond reasonable repair. The legs are extremely fragile too.

stephen_usher wrote: 12 Mar 2024 12:32

rubber_jonnie wrote: 12 Mar 2024 11:45 Things look a bit odd around the top right corner marked 83. some of the pins look like they have moved over a bit from where they should be, or is the camera angle playing tricks on me?

No, they're very skew wiff, but they are connected correctly. That's the best I could get and the pads were starting to lift there so it's best left well alone otherwise the PCB will be beyond reasonable repair. The legs are extremely fragile too.

Well If they're properly connected with no shorts it's definitely best to leave well alone!

Remember a lot of the CPU pins are also on the expansion socket.they will be a lot easier to access.

PhilC wrote: 12 Mar 2024 14:55 Remember a lot of the CPU pins are also on the expansion socket.they will be a lot easier to access.

Indeed. However, possibly a symptom of the problems, the address lines on the expansion connector don't seem to be connected directly to the CPU, or at least they don't tone out.