exxos's Atari, Amiga & retro forum

Bother. I think I have a dodgy data line trace from the CPU "socket" on the motherboard. I've soldered my board directly onto the M/B now and I'm getting errors if I flex the motherboard slightly.

exxos wrote: 28 Nov 2020 15:46 I don't know how its designed now. But I think originally it used zeners to clamp the bus to 3.3v rather that doing it properly with bus transevers. So I'm really not surprised of that malfunction. You would have to take US apart and see what its doing.

I think it must be the same on this one. I see a whole load of what look like small MOSFETs in a line near the connector.

Yeah.. If they were MOSFETs it should have proper voltage conversion.. But suspect they are clamping zeners .. Problem with zeners is they tend to start conducting well before their rated voltage. I got caught out on my early PSU batches as I used zeners and innocently changed brand and all hell broke lose. Its possible when you had 2.2k on the bus, they could have started to conduct earlier down to 2v for example. They may well be 3.3v zeners..

I think it must be A5 which is dodgy. If I rest the 'scope probe on it the machine boots. If I don't then it crashes.

OK, it wasn't a bad trace in the end as the patch wires made no difference.

Then I was checking the signals on the MMU socket, touched A6, there was a creaking sound as the "pin" moved and suddenly the system's stable. It was a poor contact between the MMU and the socket! I can flex the board as much as I like now.

Seems to be completely stable now.

So, I think I can say that this project is a success. I've not yet got the TF536 low enough to fit fully in the standard case yet but that will require getting the SMD 68030 board working and fitted on the TF. (I should order those CPUs actually.)

If anyone wants a board I have four more of them.

Well done for finally getting there @stephen_usher :thumbup:

Here are some pictures of the "low slung" board on the machine.

As mentioned previously, the height of the tower can be reduced further by soldering the TF directly on top of the board (~3mm) and then an SMD CPU on and adapter board directly onto the TF itself (about another 3mm).

The top of the PCB is actually below the top of the MMU socket, though you can't see it here.

This is interesting. Since changing the pull-up resistors on the motherboard the TF536 can't reliably read the floppy at all, but switching to the 68000 with the same OS it has no problems. This wasn't a problem before.

Maybe I should get some 4.7K resistors and replace the 2.2K ones.

I'm not sure that is the root cause of the problem...

On my H4 + TF536 I'm still using an earlier build of EmuTOS by @agranlund as that one reliably accesses/reads the floppy after boot...but when I tried the v1.0 EmuTOS image (still booting from the same 192k EmuTOS boot ROMs to load the 256k EmuTOS image from floppy) that seemed to stop the reads from floppy working.

So, no h/w changes, just a later EmuTOS image and the floppy stops working once booted (from floppy).

Apparently I need to reflash the TF536 to get the IDE on that working, so I can only use the floppy for now, hence the move back to the earlier working EmuTOS image.