MPF F11W floppy write protect issue (STE)

[exxos]

So on my second test machine the floppy drive seems to be stuck in write protect.

To cut a long story short, I traced the WP switch right back to the large chip, where the signal switches fine when inserting and taking out the floppy itself.. However the output of the chip which goes on to pin 28 (WP to the 1772) is not switching..

Has anyone ever had this problem before ? I only swapped the drive because my MPF920 drive was intermittently coming up disk write protected.. But this F11W drive just seems to be stuck like the chip is ignoring the WP signal..

I am going to assume the output of the chip is "open collector" so I am going to run a write from the switch to the floppy header and see what happens...

IMG_3663.JPG (454.01 KiB) Viewed 8275 times

[exxos]

White wire now makes the WP line go low on the 1772.. but still the issue persists

I have had a few reports over the past years of similar issues on STE's... So maybe there is something else going on here... I guess the 1772 could be faulty, but doesn't really explain the WP switch on the drive not working

IMG_3664.JPG (506.49 KiB) Viewed 8268 times

[exxos]

I have tried the drive out of my other STE which is working fine, and its not working on the problem STE either. So likely the 1772 or LS06 buffers are somehow faulty. Oddly also, the problem STE is having issues reading from the drive as well. So I will add the resistor pull ups on the LS06 tomorrow and see if that has any effect. Though it's looking like the 1772 is faulty as its ignoring the WP signal.

[exxos]

UPDATE:

As I had a SR98 in the STE, I was running at 5.05V and downed it to 4.8V.. Now the WP issue has gone away! The drive was failing pretty much 95% of the time and now its done 4 write without failure.

My assumption is pretty much as normal, noise somewhere or simply bad grounding on the 1772. So I need to look into this more...

[exxos]

This particular STE was the same one I did another fix as listed below..

https://www.exxosforum.co.uk/forum/viewt ... =17&t=1015


Though this fix isn't causing this new problem..

I took a scope of R303..

nn1.png (4 KiB) Viewed 8153 times

nn2.png (4.07 KiB) Viewed 8153 times

Seems to have some bad voodoo going on there.. oddly when reading on normal meter, I see 0.5V DC offset on that pin, when its supposed to be low. Its hard to see in the images, but gnd level is normally 0.5V then when its switching , it drops down to 0v DC line, and undershoots down almost a volt..

Ironically, when the meter connected, the floppy drive starts behaving. It is not apparent on the scope images, but the rise time of the signals ends up being slightly longer and the voltage ends up being slightly less overall as well.

So slowing down the signal a bit by loading, it seems to be the direction for investigation..

To be continued...

[exxos]

So this is very odd indeed. It will format a floppy in GEM everytime, but right at the end of the format, it will say the disk is write protected. Forcing the WP pin on the 1772 low results in the 1772 refusing to do anything (expected), but HI should mean the disk is good to go.. which it is.. the signal never switches low on the 1772... So why does TOS think the disk is write protected when its not ?!

[exxos]

OK I figured it out..

I saw this in the 1772 PDF...

777.jpg (136.06 KiB) Viewed 8133 times

Basically going onto the databus of the 1772/DMA... So I scoped out bit 6...

h1.png (4.01 KiB) Viewed 8133 times

h2.png (3.89 KiB) Viewed 8133 times

I actually have known about this issue in relation to DMA & hard drive failures.. But now it is even more concrete that the databus needs 10K pull ups..

h22.png (3.42 KiB) Viewed 8133 times

IMG_3668.JPG (321.69 KiB) Viewed 8133 times

So we basically went from this...

IMG_3667.JPG (104.51 KiB) Viewed 8133 times

To this..

IMG_3669.JPG (93.62 KiB) Viewed 8133 times

And yes I was formatting single sided.. It was quicker...


I wasn't able to capture the worst of the waveform images.. But the ringing was seriously bad on the low side.. It was like starting from a 5V level slowly ringing down to 0V where it would jump back up to 5V again...

So obviously the problem here is all the ringing is causing corrupt data bits making TOS think the floppy was write protected when it actually wasn't!

This also explains some odd corruptions I was getting with my desktop.inf file for example was causing this...

IMG_3665.JPG (66.48 KiB) Viewed 8133 times

So I go back to something I said many years ago, in, how the hell are these machines have a work at all from day one ?

Thanks for playing Atari today

[PhilC]

Lol, starting to see a pattern to what was wrong with my STF.

[ijor]

I'm not sure I see the relation between ringing at the FDC data bus and a write protect error. Obviously it could create all sort of problems. And might be it is more critical when writing and hence format fails. But it shouldn't precisely produce a write protect error unless actually it doesn't, may be formatting just fails and TOS is issuing a misleading error message.

To cut a long story short, I traced the WP switch right back to the large chip, where the signal switches fine when inserting and taking out the floppy itself.. However the output of the chip which goes on to pin 28 (WP to the 1772) is not switching..
...
I am going to assume the output of the chip is "open collector" so I am going to run a write from the switch to the floppy header and see what happens...

It is not clear to me how you were measuring the write protect signal at the drive. Was that with the drive connected, working and active? Because otherwise what you describe is correct and it is the expected behavior. The drive is not supposed to assert write protect unless it is selected. For the same reason is not correct to bypass the chip and connect the switch directly to the output header. Although it might be harmless as long as you don't connect a second drive.

[exxos]

The WP signal is set on bit 7 of the status register. So when TOS reads it, it sees a false WP signal because of ringing on the bus.. So sometimes its sees the floppy as write protected, other times not. It is for whatever reason, seeing the wrong logic level. It actually doesn't have anything to do with the WP pin on the 1772 itself (which is what I thought originally was at fault).