exxos's Atari, Amiga & retro forum

So a couple weeks ago I picked up the STF I mentioned in this thread. That PSU is long gone, and I've jury-rigged a Meanwell RD50A for power. I bought this as untested, and I'm poking it to determine if it'll be a salvage, or a donor. Waaay on the backburner, I have my H5 board that I'd like donor parts for, and I don't want to tear up a working machine.

Things are rarely so binary though, and this machine might be close to working.

At the moment, I occasionally get many bombs (20+) on boot, OR, the system boots to desktop but it doesn't read the floppy correctly. It reads some filenames but with garbage appended, and says 0 bytes are used on the disk. The disk itself is good, and the floppy drive is good: I can drop both into another machine and they're healthy. So I've eliminated those as the likely problem.


So I guess maybe the floppy controller is at issue? It looks like somebody's played with it before -- I don't imagine these were originally socketed -- but that might just mean they tried to replace and the problem persisted.


But from there... I'm not really sure how to debug this issue. This is a TOS 1.0 machine (which I'm enjoying, because I've never had one before), but one of the TOS chips has been replaced.

That suggests there was a (suspected) ROM issue in the past, too?

So, I'm curious what folks reckon. Is it a bad ROM which I'd fix by dropping in 1.4 anyway, or a bad WD1772, or a bad joint on the socket, or something else? I haven't touched any mandatory fixes on this yet, fwiw, so those are all up for grabs too.

I like to verify circuit paths are good by continuity testing point to point based on the schematics.

The FDD issue could be the WD1772, but if it's been replaced then it may be fine. It could however be the DMA, as this sounds similar to one of my Megas that was sporting a socketed and clearly replaced 1772, but still had FDD problems. Ultimately it was the DMA that was dead. By that I mean not working at all with the floppy, and not just a so called 'bad' DMA chip, which we all know isn't really a thing :)

Of course there are also some logic ICs around the FDD circuit, so you should probably check that the inputs and outputs of these ICs are correct.

One of the other issues is the crappy original cable. If you have any suspicions you can fit a normal 34pin floppy socket to the board and make your own straight through cable as well.

The corrupt filenames etc. does reflect a DMA chip issue.

However, check the bus pull-up resistors and make sure none have broken. If the some of the data lines are not being pulled up correctly then this could cause such problems especially as the DMA is at the far end of the data bus. I was getting similar corruption with incorrect pull-up values (too high in my case).

stephen_usher wrote: 22 Feb 2022 10:54 The corrupt filenames etc. does reflect a DMA chip issue.

However, check the bus pull-up resistors and make sure none have broken. If the some of the data lines are not being pulled up correctly then this could cause such problems especially as the DMA is at the far end of the data bus. I was getting similar corruption with incorrect pull-up values (too high in my case).

Continuity on the data lines looks good between the DMA and the CPU, and also between the DMA and the floppy controller.

Looking at the pull-ups, that might be something. RP2 and RP3 sit on the data lines and should be 10k resistors, if I'm eyeballing the schematics correctly? Both are marked on the board as 10k.

RP2 measures at ~3.1k, apart from the last pin which reads 10k.
RP3 measures at ~3.1k

I can't read/find the markings on some of the other pull ups, but I decided to look at those as well. I measure the following:

RP1: 3.1k, but last pin reads at 4.7k (schematic says 1k?)
RP4: 3.1k (schematic says 10k?)
RP5: 4.7k (schematic says 4.7k ✔️)
RP6: 4.7k (schematic says 10k?)
RP7: 10k (schematic says 4.7k?)
RP8: 10k, but last pin reads 3.1k (schematic says 10k)
RP9: 10k (schematic says ??, is 10k on board)

I've never looked at these resistors before; I don't know what's normal, or what varies between the actual production boards. This is a C070523-001 REV.D, though mine has a mix of red and black network resistors compared to the boards pictured in that thread.

In any case: before I even get to mandatory fixes, at the very least maybe both RP2 and RP3 ought to be replaced? With 10k ohm resistors, or other? RP1 and RP8 look weird to me also. Thoughts welcome on these, or the others. It looks like only RP5 behaves well and matches the schematic, but maybe the actual values aren't super important!

Answering my own late-night thoughts: based on this post it's probably a good idea to switch RP1, 2, 3, 4, 7, 8, and 9 for 2.2k resistors. I don't have these spare, but I've got some on the way.

That'd bring the 3.1k(?) resistors and the 10k resistors down to 2.2k, and leave RP5 and 6 on 4.7k. Since some of the resistors above seem to read weird anyway, maybe this is for the best.

So here's where I am with this machine:

Ignoring my own post above, I did a better job of mapping which RP* resistors to pins on the 68k, after which I found this post which happily corroborated what I'd found with my multimeter. So I've now swapped out RP1, 2, 3, 4, 7, 8, and 9 with 2.2k resistor packs. RPs 1-2 and 3-4 are rigged in parallel so they actually provide 1.1kohm resistance, but from what I interpret, that's likely still fine. Desoldering some of these was a real pain and I'm not the best solderer, but the resistances now all read as they should.

The result? I still get the same garbled filesystem info as before. It's at least very consistent garbled info. It's the same every time.


In addition to the RP resistor pack replacements, I recently bit the bullet and bought a bunch of parts for my eventual H5 build. This includes a WD1772 and a DMA chip. My results with those:

• Swapping out the WD1772 for the new one leads to the same result as above, so that seems to eliminate the WD1772. But it might not eliminate the socket.
• Swapping out the DMA chip for the new one gives me a black screen on boot. Leaving the DMA chip out completely at least gives me a white screen, even if the system never manages to reach the desktop. (Bearing in mind this is an ancient STF with TOS1.0, do different DMA chips work on only some models of ST?)

Right now I'm not sure what's more likely: a bad socket connection for the WD1772, or a bad DMA, or a bad connection for the DMA socket, or a bad floppy cable, or something else entirely!

I see you've done the RPs, have you followed the list of mandatory fixes? The RESET circuit could be playing up so definitely worth a look.

Also, a white screen often means a ROM issue, so that may be worth a look. Do you have a diag cart? They can be helpful, though do need a partly working ROM set to start.

Have tou looked at the sockets and tweaked the PLCC sockets and cleaned the IC legs. I had an issue with a machine I was repairing for somebody else and both the CPU socket and the Blitter socket had problems.

On the mandatory fixes: the machine's running off a meanwell PSU (fix 1), and I'm running original drives so the 1772 pull up fix isn't necessary (fix 4). Video has been fine so I haven't touched fix 9.

One fix that I haven't touched is fix 3, DMA pull-up on the databus. My DMA is socketed, so could I even just wedge a resistor array in there?

Reset fix I also haven't touched; the reset circuit seems to be working ok. Would that be likely to manifest as apparent disk corruption? (I know the drive and the disk are both good.)

I probably missed a step in my own description when testing the DMA chips and boot behaviour. I tried three things, third for fun:

• chip C025913-38: the machine boots a-ok, but I can't properly read disks from the drive. This is the chip the machine arrived with.
• chip C025914-38A: this is the DMA chip I bought from exxos's store: black screen, nothing more. (This chip will go in my H5 eventually, but I had it here to try it.)
• no chip at all: white screen, nothing more.

I've muddled around the 1772 an DMA sockets and wiped clean the legs etc; by now I'd hoped that even seating and reseating a few times would maybe have woken something up. No luck.

I haven't done anything invasive on the sockets, like resoldering their connections, yet. I'm not sure how best to check those before pulling out the soldering iron.

stween wrote: 11 Apr 2022 14:22
One fix that I haven't touched is fix 3, DMA pull-up on the databus. My DMA is socketed, so could I even just wedge a resistor array in there?

I haven't done anything invasive on the sockets, like resoldering their connections, yet. I'm not sure how best to check those before pulling out the soldering iron.

On the first item, I wouldn't wedge it in. You have no guarantee that you will have a solid connection and may introduce more problems.

I didn't mean reflowing the sockets, I meant this: https://www.exxosforum.co.uk/forum/viewt ... plcc+tweak

And only for the PLCC sockets, the DIP sockets are usually OK, worst case they need a spot of DeOxit or something similar.

Oh right! I haven't played with any of the PLCC sockets; that's a good idea. I'll take a look at them.