snufkin wrote on 2021-04-27, 14:10:

What happens if you try to format a floppy, with no command line switches? Does it say what capacity it's going to try before failing? Had an odd case on another thread where the BIOS menus had the correct settings, but the registers ended up with the drive as a 360k 5.25".

It will format as 1.44mb (so it's seeing the CMOS setting correctly), then fail with a track 0 unusable which at this point is whatever our problem is.

snufkin wrote on 2021-04-27, 17:11:

I'm assuming the interface cards have been tried in different slots?

Yes, every slot as a sanity check.

Does the drive head ever step (e.g. when doing a step to track 0 on startup)?

Yes, it will do that head step thing. Additionally as a sanity test I dropped in a known bad floppy controller and yes, POST will give you a FLOPPY CONTROLLER FAIL error, so the BIOS is polling the controller and expecting a response with every controller I dropped in and it's doing that, just no data to/from the drive itself.

What is the full BIOS string on the initial POST screen?

Let's make sure that you have the correct BIOS.

[edit bad phrasing here]
As well as checking the BIOS, it may be worth experimenting with some jumpers. I had a look through the photo you posted of the board and compared with:
http://www.dosdays.co.uk/media/asus/ISA486SV2/jumpers.txt

One differences I spotted is the hard wired link for 1x CPU multiplier on JP1 down by the BIOS chip. It's currently wired as 2-3, which the silkscreen says is 1x and the dosdays listing shows as 2x.

Also, dosdays listing gives the factory defaults for TP13 (just beneath the 16bit block of the 4th ISA slot) as DREQ5 pulled low, rather than the current jumper set to pulled high (which matches the setting for DREQ6 & 7). But that section of the listing looks wrong anyway (it lists jumping the same pins for both pulling low and high) so may not be a good guide. Still, might be worth playing with. Perhaps also the IRQ6 jumper down near the BIOS chip.

If the head is stepping then instructions have to be getting to the FDC. So I'd guess data can be written the disk. But I think that when DOS tries to format a disk it will write and then read track 0. If the read fails it gives up since the FAT will be broken. So my guess is that the read fails.

What does IMD http://dunfield.classiccmp.org/img/index.htm do on either the RPM test or the alignment test? I'm not questioning the alignment, just whether it manages to start showing any data or just stalls trying the ID the disk.

i have the same problem but no solution... AQUARIUS SYSTEMS, INC. A10001 - Does not want to read from the floppy

a curiosity can u post a full pic of your ISA post code card ?

snufkin wrote on 2021-04-28, 10:31:

[edit bad phrasing here]

One differences I spotted is the hard wired link for 1x CPU multiplier on JP1 down by the BIOS chip. It's currently wired as 2-3, which the silkscreen says is 1x and the dosdays listing shows as 2x.

What does IMD http://dunfield.classiccmp.org/img/index.htm do on either the RPM test or the alignment test? I'm not questioning the alignment, just whether it manages to start showing any data or just stalls trying the ID the disk.

Later revisions of my board apparently replaced a crystal clock circuit with an MX8310 clock generator (like this one). It has its own cluster of jumpers up by the unpopulated QFP footprint so my guess was they just hardwired the X1/X2 jumper because it was no longer needed.
IMD is a mess. It will poke at the drive but everything it gets back are just as weird.

a curiosity can u post a full pic of your ISA post code card ?

It looks intimidating with all those GAL's but this seems to be a variant of the ForeFront Discovery Card. The IRQ and DMA lamps turn on whenever their respective lines are polled and they latch on until the reset button on the card is pressed. Much more useful than the POST cards you find now.

Edited: Oh yeah, the BIOS string is 40-0806-001292-00101111-080893-I486SI-H

CelGen wrote on 2021-04-29, 00:47:

snufkin wrote on 2021-04-28, 10:31:

https://hosting.photobucket.com/images/a166/ballsandy/IMG_8383.JPG It looks intimidating with all those GAL's but this seems to […]
Show full quote

a curiosity can u post a full pic of your ISA post code card ?

It looks intimidating with all those GAL's but this seems to be a variant of the ForeFront Discovery Card. The IRQ and DMA lamps turn on whenever their respective lines are polled and they latch on until the reset button on the card is pressed. Much more useful than the POST cards you find now.

Wow Nice ! looks like an upgrade to my original 1992 Discovery Card, bought back in late 1990's and still use it for diagnosing XT and 286 stuff:

added: Probably should save the 51 page manual to PDF and post to our library here with very clear pics of the front and back in case anyone wants to make their own. The TTL chips are common, no special PALS or GALS.
Sorry not trying to steer the topic off bad floppy but those early Diag cards are quite rare !

BIOS is correct for that motherboard according to elhvb:
https://www.elhvb.com/webhq/models/486vlb2/486sv231.htm

Horun wrote on 2021-04-29, 01:59:

added: Probably should save the 51 page manual to PDF and post to our library here with very clear pics of the front and back in case anyone wants to make their own. The TTL chips are common, no special PALS or GALS.
Sorry not trying to steer the topic off bad floppy but those early Diag cards are quite rare !

Yes i need to reproduce it and share pcb.

I have checked all the connections of the isa connector and they are OK.
When i i try to read the floppy
I have this type of signal on a working card, in the NOT working one the DAK02 / has no activity,

CelGen wrote on 2021-04-29, 00:47:

The IRQ and DMA lamps turn on whenever their respective lines are polled and they latch on until the reset button on the card is pressed.

kinda makes me wonder which dma line it listens to... dma ack or dma req or hopefully both... though that, they stay latched until the reset button is pressed doesn't make me very hopeful about both.

My two cents...
I've got the same board, working fine but one thing is not ok with it.
The original battery is removed in time, and I now run it without one. Whenever I connect an external battery the system becomes unstable and gives me all kinds of errors.

This includes random lockups, weird characters on any VLB video card and... the FDD refusing to work
Actually 5 were tested, that are known to be working fine in other setups

An ISA graphics cards and controller seems to get a little stability back, but not all. Removed the external battery and it works perfectly stable again.

dataino.it wrote on 2021-04-29, 07:01:

I have this type of signal on a working card, in the NOT working one the DAK02 / has no activity,

Inspect your cards, both the working and non-working one, and see if the ALE signal (B28) is connected to anything. It's easy to tell, missing or not connected gold contact - ALE not used on the card.

CelGen wrote on 2021-04-29, 00:47:

IMD is a mess. It will poke at the drive but everything it gets back are just as weird.

Track 246, sector 22. That's a big disk.

It's a consistent mess though. I think that means the FDC is reliably reading the start of sector ID marks, including the correct CRC. With IMD, if I move the heads out of alignment (so the data read is corrupted) then I just get a bunch of '?' showing, and it would seem unlikely for the CRC to be corrupted in just the right way to match a corrupted track number. So it's another indication that it's failing to move data correctly from the FDC to the motherboard. Do you know if those DREQ and IRQ jumpers change anything?

In IMD, does the capital 'Z' at the top of the screen (where it says fwRZd) change to lower case if you move the heads off of track 0 (try pressing '+' to go up one track and '4' to move to track 40), and also see if the reported track number changes. If it does then that suggests the motherboard is able to read the status flags properly. Also check if there's any change in the output between head 0 and 1 (press 'h'). Oh, and of course 'd' to try and read data - if the FDC is actually reading the disk fine then it should say it can read the data ok. I think it only flags an error if the data CRCs don't match.

Also, on waterbeesje point, can you check what the power supply looks like on the floppy and the controller card? I've got a floppy drive that uses a zener diode to generate a reference voltage for reading the magnetic data from the disk. If there's too much noise on the supply line then the reference voltage probably wouldn't be stable enough to get good data. On the other hand, something like that would probably cause random errors in the data, not consistent errors.

Do you know if those DREQ and IRQ jumpers change anything?

They don't seem to change anything at all when moved around.

n IMD, does the capital 'Z' at the top of the screen (where it says fwRZd) change to lower case if you move the heads off of track 0 (try pressing '+' to go up one track and '4' to move to track 40), and also see if the reported track number changes. If it does then that suggests the motherboard is able to read the status flags properly. Also check if there's any change in the output between head 0 and 1 (press 'h'). Oh, and of course 'd' to try and read data - if the FDC is actually reading the disk fine then it should say it can read the data ok.

With a freshly formatted floppy I can step between tracks and the read command seems to work but writing still fails. If the head selected is indicated by H0 or H1 then it is properly changing heads, otherwise it isn't.
Curiously while doing these tests I ran into quirks. On one occasion I got NO INTERRUPT FROM FDC that would not clear until I warm reset the machine. On another occasion the drive stopped responding entirely but IMD never reported any errors.

Also, on waterbeesje point, can you check what the power supply looks like on the floppy and the controller card? I've got a floppy drive that uses a zener diode to generate a reference voltage for reading the magnetic data from the disk. If there's too much noise on the supply line then the reference voltage probably wouldn't be stable enough to get good data. On the other hand, something like that would probably cause random errors in the data, not consistent errors.

+5 reads 5.11v and +12 reads 12.1v. I'm not seeing any major ripple on either rail with an oscilloscope. It's somewhere around 0.1v?
I did entirely recap the power supply before I started all this as preventative maintenance. This is some Taiwanese TOUCH MPS-5200S AT supply but I can dig into my storage and I might have another spare AT supply somewhere. No promises. Likewise I've been pulling my hair out on this board for nearly two weeks now. I'm getting very tempted to just swap in that test board and be done with it because I need my bench back soon. :U

Edited: I tried a different power supply with no change in symptoms, so unless two power supplies are having the same problem it's not the PSU

CelGen wrote on 2021-05-01, 03:33:

With a freshly formatted floppy I can step between tracks and the read command seems to work but writing still fails. If the head selected is indicated by H0 or H1 then it is properly changing heads, otherwise it isn't.

Ok, so I think that proves the FDC is reading data correctly from the disk. As I understand it, IMD is just checking a status register in the FDC to see if it throws a CRC error, so the data isn't actually moved to the motherboard. If the track 0 indicator is changing from 'Z' to 'z' then that suggests the FDC status registers can be read correctly. So it would probably be able to see if there had been a CRC error.

Power supplies sound fine. I think everything so far says this has to be a problem on the motherboard (you've tested controllers and drives elsewhere), and yet other cards can transmit data fine.

I'm sure this has already been covered, but on the controller card, does it have IDE ports with anything plugged in to them, and do they work correctly (this should test whether anything can get data across the bus from that particular card)? Can you post a photo of the card? I'm just wondering if there could be some issue with the FDC not being able to switch the bus lines fast enough (maybe some line capacitance issue?), but it should be buffered anyway. And you've tried multiple cards.

What sort of scope do you have? It'd be a fair bit of work, but I'm wondering if it's worth looking at a bus data line (not sure what to trigger off, maybe the index pulse on the floppy connector?), and comparing it with a different motherboard. Essentially just hunting around for when signals start looking different between a working system and this one.

Other than that though, I'm stumped.

Oh, not sure if this would tell us anything, but try formatting a DD 720k floppy and see if any of the numbers in IMD alignment test change, particularly the sector number (which I think was stuck at 22 before).

Curiously while doing these tests I ran into quirks. On one occasion I got NO INTERRUPT FROM FDC that would not clear until I warm reset the machine. On another occasion the drive stopped responding entirely but IMD never reported any errors.

I had something similar happen (drive stopped responding) when I was doing bad things to a drive and trying to measure the timing pulse to the motor controller. But that was with a faulty drive I'd tried in multiple machines, whereas you've tested the drive and it's ok.

Still, maybe there are bus lines being driven directly by the FDC and it (plus all the FDCs on the other controllers) are struggling to drive the line. Or the other way round. If a line from the motherboard is slow to rise then it could throw signal timing out? With my broken drive I could compare the shape of RDATA with a good drive and it looked ok. I eventually found some completely missing pulses on the RDATA line from my broken drive, which has pointed me toward the analogue read stage since the digital stage looked ok.

Likewise I've been pulling my hair out on this board for nearly two weeks now. I'm getting very tempted to just swap in that test board and be done with it because I need my bench back soon. :U

Understood. I've spent far too long trying to get my broken floppy drive working. But it's so nearly working, so I can't just let it go, but have put it to one side for the moment. I find it much easier if something is clearly and definitely not working. At least then it's usually easier to identify the problem and decide if it's worth trying to fix. Situations where I feel there's probably just one duff capacitor, or a corroded but not broken track, and I can't find it, are annoying.

it could also be a failed or missing decoupling cap.

weedeewee wrote on 2021-05-01, 12:31:

it could also be a failed or missing decoupling cap.

True, line bounce could be a thing. To be honest, I would have expected the wrong track/sector numbers from IMD to be random if it were a decoupling issue, but maybe not.

This is where my lack of knowledge of the ISA bus gets in my way. We know (I think) that the floppy drive can read data from the disk and get it to the FDC on the controller card, but the FDC can't get data to the CPU. Or at least not reliably given that status seems to read ok. And this is with multiple drive and controllers, and trying different slots. But we also know (I think) that other things on the same controller card can, so I don't see that it can be a general bus issue.

So, which, if any, lines on the ISA bus only go to the FDC on controller cards, and if there are only a few, do we know what a rising and falling edge should look like on them? And is there anywhere else on the motherboard for the data to get consistently corrupted?

Hang on, just re-reading the thread, I've just realised I missed the post about the missing DACK2 signal on the non-working motherboard when there's a signal on a working motherboard. That sounds important. Sorry, I should have seen that. DMA controller not being programmed correctly? Or not responding to DRQ2? Some FIFO just filling up and never realising data is being read? But why would that only affect the FDC?

I've just been skimming the DMA section of this: https://gist.github.com/PhirePhly/2209518 . As I said, I don't know the ISA bus, but it looks like pins to check on a scope and compare with a working board might be DRQ2, AEN and DACK2. Look at the rise and fall shape, and if you have a dual channel scope, check the relative timings if possible. Are the DMA lights on your bus test card triggered by the DRQ or DACK signal?

snufkin wrote on 2021-05-01, 13:46:

But why would that only affect the FDC?

Because generally nothing other than floppy and sound blaster use the DMAC on an AT-class system?

I'm sure this has already been covered, but on the controller card, does it have IDE ports with anything plugged in to them, and do they work correctly (this should test whether anything can get data across the bus from that particular card)? Can you post a photo of the card?

It does and every time so far we've run IMD we've been running through that card using a DOM, so the IDE channel it's safe to say is working fine and its sharing the same chipset as the floppy controlle so I doubt the chip is half-dead.

What sort of scope do you have? It'd be a fair bit of work, but I'm wondering if it's worth looking at a bus data line (not sure what to trigger off, maybe the index pulse on the floppy connector?), and comparing it with a different motherboard. Essentially just hunting around for when signals start looking different between a working system and this one.

Tektronix 564B. It's very old but it's the only scope I have which has any storage capabilities and it's just fast enough to handle most basic logic work. Unfortunately I don't think its fast enough to trigger and capture from the bus.

Oh, not sure if this would tell us anything, but try formatting a DD 720k floppy and see if any of the numbers in IMD alignment test change, particularly the sector number (which I think was stuck at 22 before)

Here's the alignment program with a formatted 1.44mb floppy inserted. Note that in the middle I changed heads and stepped to track 50.

And again, but with a formatted 720k disk.

Hang on, just re-reading the thread, I've just realised I missed the post about the missing DACK2 signal on the non-working motherboard when there's a signal on a working motherboard. That sounds important. Sorry, I should have seen that. DMA controller not being programmed correctly? Or not responding to DRQ2? Some FIFO just filling up and never realising data is being read?

All I said was that one of the input lines for the glue that DACK2 was connected to was not directly going to the chipset. It's a pin that is connected in a middle board layer and isn't visibly traceable from either side, so either it's broken OR it's going through some other logic and that's the reason why I don't see a direct connection to the chipset. I didn't chase this because it's going to take a while to track it down.