Issue writing data on floppy (Write protected)

[exxos]

I don't think that is noise.. I see things all the time... Noise wouldn't generally be so systematic like that.

I can't really suggest anything else to try. Other than sending your board off to someone to troubleshoot.. But it might just be cheaper / easier to just get a different motherboard .

[geogeo_28]

I finally fixed my issue with floppy write and corruption on UltraSatan by doing the following:
- Removed PIL of data and address bus of the CPU.
- Replaced P100 with a PIL 1.2K.
- Add PIL of 2.2K on the data 0-7 bus of the DMA. (For some reasons 10K does not work).
- Add PIL of 10K on the data 0-7 bus of WD1772.
- Add 30pF on ACK to GND of the CPU.

- For some reasons, pull-ups on the SN7406N don't work. Bring back floppy read issues.

This is working with Gotek and floppy drives.

[geogeo_28]

Here is a photo with all the changes:

IMG_0607.jpeg

[exxos]

So p100 wasn't 1.2k ? I can't remember if that is listed in the fixes or not..

Also the 1772 databus is wired to the DMA data bus , so are you sure you don't have like 1k on there now ?

It kinda hints to me that the 1772 needs replacing..

[geogeo_28]

So p100 wasn't 1.2k ? I can't remember if that is listed in the fixes or not..

I took as a reference this page:
https://www.exxosforum.co.uk/atari/last/DMAfix/index.htm

Changing back to 2.2K makes the ATARI STe stable at 32MHz.

Also the 1772 databus is wired to the DMA data bus , so are you sure you don't have like 1k on there now ?

You are right, the pull-ups on 1772 were not necessary.
I have two 1772 and both behave the same way.

I can claim victory on the floppy read/write at 8/32MHz but for the UltraSatan I spoke too fast.
The irony is, with the UltraSatan, at 8MHz I still experiencing data corruption on write but no issue at all at 32MHz.

[exxos]

I can claim victory on the floppy read/write at 8/32MHz but for the UltraSatan I spoke too fast.
The irony is, with the UltraSatan, at 8MHz I still experiencing data corruption on write but no issue at all at 32MHz.

If you want irony, booster runs at 8MHz during DMA cycles anyway...

[exxos]

All I can suggest is scope out the 1772 / DMA data bus (not CPU side) and access Ultrasatan and floppy and measure each data line one by one to see what each look like while doing accesses... Its one of those things where a scope probe on X1 may load a signal enough to "fix" the problem.

[geogeo_28]

Thanks for your help. I scoped each data line and indeed, D5 is suspicious.

IMG_0622.jpeg

Unfortunately, the scope probe did not help. This signal is lazy. ^^

Any idea what can cause this "shape". Looks like the signal is "force" at 1.

[exxos]

Its hard to say if that's normal or not as the signals are crazy to start with., though there is normally bad under and overshoot on signals but there's nothing at all on your scope ?

I guess you could try 1k on the 1772 bus ? Saying that is this with my 1.44 mod installed ? As there was a mod done at some point,, there's 2x 1k resistors somewhere in the middle of the board iirc ... Those got changed to 470R...

[geogeo_28]

I don't see under or overshoot of the signal. All lines look fine.
Playing with the pull-ups of 1772, with/without floppy mod I obtain this:

IMG_0624.jpeg

Does not cure UltraSatan issues.

Can you point me to the 1K resistors to change?
This is the similar to this DMA fix: https://www.exxosforum.co.uk/forum/viewt ... 5&start=10 ?