Did you try a regular ISA graphics card?

majestyk wrote on 2021-11-02, 16:04:

Did you try a regular ISA graphics card?

Yes, I did try to leave the IO/IDE card as the only VLB card in the system. No change.

Some more testing done, still no results.

I switched back to the AWARD BIOS, then attempted to use the DOS driver (attached) to enable 32bit mode on this controller.

When the driver loads, the computer stops for a while, then reports errors and corruption in the CONFIG.SYS file. If i try to launch a "dir" command once the boot completes, I simply get garbled data.
I tried to disable "turbo" mode on the controller via driver's switches, and also to set the speed of the drive to the minimum (BTW, the driver reports a jumper setting that enables Turbo mode and the fastest speed for the drive).

I redid all the above tests with 1WS enabled on the VL-bus. No change.

Had a look for any matching boards and couldn't just see one. Looks like the PDC20230 has some pin settings for HDD speed. Found a manual for a FIC 486-JAL that has one built in and that has jumper settings specifically for the PDC, for HDD speed (J16&J17) and CPU speed (J18). I'm thinking that maybe (and this is pretty unlikely) JP2&3 on your board are for HDD speed, assuming they route to pins on the PDC. On the FIC board then 1-2,1-2 is Speed 2, and 2-3,2-3 is Speed 0. Moving them probably won't break anything.

Taking a quick wild stab at jumpers:
JP1 - HDD activity LED?
JP2&3 - HDD speed?
JP4-9 - Factory hard wired. Floppy drive settings?
JP10 - HDD base address?
JP11 - Looks like if the jumpers are put horizontally it swaps round IRQ3&4 inputs to the FDC37C652
JP12-16 Factory hard wired. More floppy settings (they look to connects to the Motor/Drive select pins)
JP17 - can't see
JP18 - Factory hard wired. Might be FDC DMA related?
JP19,21 - LPT port settings?
JP20,22 - COM port settings?
JP23 -LPT IRQ select?
JP24,25 - Dunno, something to do with IRQ 1 & 3?
JP26 - dunno
W1-3 - dunno

Of course there could be a bug in the board design. There's a weird trace that just comes to a stop near the ISA connector, but runs off toward the PDC chip.

snufkin wrote on 2021-11-03, 14:16:

Had a look for any matching boards and couldn't just see one. Looks like the PDC20230 has some pin settings for HDD speed. Found a manual for a FIC 486-JAL that has one built in and that has jumper settings specifically for the PDC, for HDD speed (J16&J17) and CPU speed (J18). I'm thinking that maybe (and this is pretty unlikely) JP2&3 on your board are for HDD speed, assuming they route to pins on the PDC. On the FIC board then 1-2,1-2 is Speed 2, and 2-3,2-3 is Speed 0. Moving them probably won't break anything.
[...]

Good catch on that trace, impressive!
Tried to swap JP2/3, but made no difference. Actually tried most of the other jumpers too.
I'm starting to wonder if this board is actually partially dead, at least the VLB part...

Just read this thread and mkarcher pointed out that the GD542x family only use 16 bits: Good VLB card benchmark

So it's possible the problem isn't with this controller card. Don't suppose you've got another VLB motherboard to test it with? Or a card that definitely uses 32bits?

snufkin wrote on 2021-11-03, 16:30:

Just read this thread and mkarcher pointed out that the GD542x family only use 16 bits: Good VLB card benchmark

So it's possible the problem isn't with this controller card. Don't suppose you've got another VLB motherboard to test it with? Or a card that definitely uses 32bits?

That's a very good point, the motherboard could be faulty.
I do have other VLB motherboards, though they are in unknown shape, need to test them. Hopefully at least one will be in good enough shape.

As for other cards, I think I might have one or two SCSI controllers for VLB. Will have to search for 50 pin cables to connect something to them.

Interestingly, the PD2030 drivers for DOS have a flag to force the card in 16-bit data access mode, but even using that I get the same issue

Dug out a VLB SCSI controller: an NXT-80 and a SCSI2SD to go with it.

I've been able to format a drive connected to this card, run benchmarks, copy to and from said disk. No issue...

BTW, Maybe I should dump the BIOS on it while I have it on the desk...

Ok, so 32bit likely working on the motherboard. Not sure then. Without a datasheet for the PDC (I can't find one) then difficult to know what any signal should be doing. Might be worth lifting that sticker on the back that's covering some of the VLB data lines to make sure there aren't any breaks, and to see where that weird dangling trace goes. Beyond that I'd probably start checking if those 5 74 series logic chips are all functioning properly, but I don't know if that can be done in place or if you'd have to remove them. Have you checked to make sure there aren't any bad solder joints between the PDC and board (I think they look ok)?

not all disks for all controllers and storage cards with adapters have incompatible transfer modes therefore if theres a mode that works than a mode that doen't in BIOS best just to use what it can and be happy.
Can be as simple as the card is too big for a particular mode or too fast transfer mode built in thats not supported.

snufkin wrote on 2021-11-03, 21:43:

Ok, so 32bit likely working on the motherboard. Not sure then. Without a datasheet for the PDC (I can't find one) then difficult to know what any signal should be doing. Might be worth lifting that sticker on the back that's covering some of the VLB data lines to make sure there aren't any breaks, and to see where that weird dangling trace goes. Beyond that I'd probably start checking if those 5 74 series logic chips are all functioning properly, but I don't know if that can be done in place or if you'd have to remove them. Have you checked to make sure there aren't any bad solder joints between the PDC and board (I think they look ok)?

Yup, tomorrow will cleanup the board from stickers and check under them, and also trace that stray track. I couldn't spot any bad joint

I could scope the signals on the 74' chips, but for such a small number it's not even worth the effort of setting up the instrumentation. I already started desoldering them: 3 out of 5 socketed and tested in a TL866 (which reported them as good... Worth trying swapping them anyway, at this point).

UPDATE: All the 74' tested ok. Swapping them provided no benefit.

I got hold of another VLB controller, a VL-300.
When I load the driver for 32-bit mode, I get the same issues as with the other controller. This would point to an issue with the motherboard.

EDIT:
Well well well...
Apparently I found the issue: the drivers really hate when I configure the drive as LBA in the BIOS. The VL-300 comes with an utility that checks the drive and recommends the best configuration in the BIOS, which matches the "NORMAL" disk mode.

When I repartition and reformat the drive with that, loading the board drivers works fine and I no longer get any error. Pity FDISK does no longer see the full drive but lets me use only 504 megabytes out of it.

EDIT 2:
Trying with the adaptec driver linked by Warlord in this thread: viewtopic.php?t=73525 . Seems to work fine with LBA, and I can now reach ~5.7Mb/s!

In case someone ever needs it, attaching the BIOS of the VL300, plus the reversed TIBPAL16L8 at U11.