Could be a problem with the card, but before you start blaming it - what mobo is this? Did you try other PCI slots? To make sure it's not one of those "extra" slots without bus mastering capability.

I've gone ahead and attached the manual for the drive.

This drive indeed has an internal terminator that is enabled using a jumper, make sure it's enabled (unterminated SCSI buses can cause all sorts of funky issues, aka "SCSI voodoo" or "SCSI black magic") and that you have no other devices inbetween. The chain must be terminated at both ends (the card and the last device in the chain)

Deunan wrote on 2024-08-07, 18:54:

Could be a problem with the card, but before you start blaming it - what mobo is this? Did you try other PCI slots? To make sure it's not one of those "extra" slots without bus mastering capability.

It's a Biostar mobo specifically chosen to be able to have dual floppies and all that other junk while also having Windows 10, other ide drives and sata. I could try swapping it with the other socket, but I have a sound blaster card in it, not sure if that'll help. I might actually not have a bios on the chip and I have no idea how to get one installed either

Vynix wrote on 2024-08-07, 19:22:

I've gone ahead and attached the manual for the drive.

This drive indeed has an internal terminator that is enabled using a jumper, make sure it's enabled (unterminated SCSI buses can cause all sorts of funky issues, aka "SCSI voodoo" or "SCSI black magic") and that you have no other devices inbetween. The chain must be terminated at both ends (the card and the last device in the chain)

I don't think the manual states clearly enough how to terminate this drive with a jumper. Best I could pick up from it was that a3 is scsi #0 and b3 was GND, so my best idea was to put a jumper on A3/B3. The card terminates itself automatically though. But yeah, quite possibly if I figure out the termination on the drive without bricking it first, it might work 😁

EDIT: Never mind, it said that A11 is responsible for termination. Does that mean I should put a jumper on A11/B11? What about the one on A3/B3, do I need this mechanical indication that the drive is meant to be scsi #0 even though it is already recognized as such by the system?

Yes, try with a jumper on A11/B11, as for the SCSI ID, you can set it to ID 1 for testing, it shouldn't hurt anything.

As far as I can recall there was a DOS version of SCSISelect, although I'm not a hundred percent sure if it's gonna work with a card as new as the AHA-2940.

Vynix wrote on 2024-08-08, 00:19:

Yes, try with a jumper on A11/B11, as for the SCSI ID, you can set it to ID 1 for testing, it shouldn't hurt anything.

As far as I can recall there was a DOS version of SCSISelect, although I'm not a hundred percent sure if it's gonna work with a card as new as the AHA-2940.

Ok, so the termination of the drive solved some other problems (the small IDE MO drive acting wonky). Putting jumpers on the first three pin pairs did alter the position of the drive from scsi #0 to #1 or #2, but otherwise changed nothing, so I just decided to take off that jumper after all. But still I get no way to read the big disks, Windows just says the disk could not be found because there is a fatal hardware error or something, even though the disks are brand new and so is the drive.

I think the problem is that there is no way for me to terminate the SCSI circuit on the other end, because I actually have no way of getting into SCSISelect in the first place. The screen prompting me to press Ctrl+A to get into it, if left idle will display the connected SCSI device, but then after some time say "Bios not installed!", and pressing Ctrl+A to get into SCSISelect will just inform me that no device has been detected and to check if it's properly seated. Tried that in both PCI slots.

So I think the problem might actually be the lack of bios or the card is otherwise damaged, since I hear it's possible to disable the bios entirely from SCSISelect and it shouldn't interfere with anything. But since I can't get into SCSISelect in the first place...

Should I start seeking after some other controller? I've seen a 2940AU for peanuts, but I've also seen a 19160 for a very good price, and I hear those are even newer and more powerful. I'd be happy to hold off on buying a new one, but I don't think I have the means to determine or fix what's wrong with my current one.

Spiffles wrote on 2024-08-08, 00:42:

I think the problem is that there is no way for me to terminate the SCSI circuit on the other end, because I actually have no way of getting into SCSISelect in the first place. The screen prompting me to press Ctrl+A to get into it, if left idle will display the connected SCSI device, but then after some time say "Bios not installed!", and pressing Ctrl+A to get into SCSISelect will just inform me that no device has been detected and to check if it's properly seated. Tried that in both PCI slots.

The SCSI Bios will only be installed if there is a bootable scsi device on the chain, possibly even only with scsi id 0 & 1 though unsure of this.

You should always be able to enter the scsi bios setup with CTRL-A even if there is no scsi device attached.
Something is clearly going wonky though I do not know what exactly to help out.

have you tried entering the bios with only the scsi card installed with no device attached, ie no other pci or isa cards present (aside from a vga) ?

weedeewee wrote on 2024-08-08, 04:15:

have you tried entering the bios with only the scsi card installed with no device attached, ie no other pci or isa cards present (aside from a vga) ?

Have tried with no device installed, made no difference. There was another card in the other PCI slot though, a SB Audigy. The mobo has no ISA slots, just two PCI-E and two regular PCI. I wonder if it might be using up all the bandwidth for the graphics and sound card? But nah, that shouldn't be it. There must be something deficient about the adapter itself. I've picked up a cheap AHA-2940AU in the meantime, waiting until it arrives and I'll see if it'll make any difference. The bios is versioon 1.32, which seems like it's a huge leap over the 1.11 that I have on my current adapter, 3 years newer too. A user on vcfed has suggested that it's not normal for the card to not be recognized in SCSISelect given that the card's presence is what lets you enter that utility in the first place, and I agree that it feels a bit counterintuitive.

Spiffles wrote on 2024-08-08, 06:37:

weedeewee wrote on 2024-08-08, 04:15:

have you tried entering the bios with only the scsi card installed with no device attached, ie no other pci or isa cards present (aside from a vga) ?

Have tried with no device installed, made no difference. There was another card in the other PCI slot though, a SB Audigy. The mobo has no ISA slots, just two PCI-E and two regular PCI. I wonder if it might be using up all the bandwidth for the graphics and sound card? But nah, that shouldn't be it. There must be something deficient about the adapter itself. I've picked up a cheap AHA-2940AU in the meantime, waiting until it arrives and I'll see if it'll make any difference. The bios is versioon 1.32, which seems like it's a huge leap over the 1.11 that I have on my current adapter, 3 years newer too. A user on vcfed has suggested that it's not normal for the card to not be recognized in SCSISelect given that the card's presence is what lets you enter that utility in the first place, and I agree that it feels a bit counterintuitive.

The fact that the bios shows on boot and detects the devices does indicate the card is working.
Why the scsi bios setup and scsiselect utility isn't finding the card is indeed not normal behaviour and suggests that whatever the scsi bios setup & scsiselect util are trying to find is incorrect.
Could be a simple PCI device ID or revision that is incorrect due to eprom corruption. at the moment I'd guess that it is fixable, though no idea how 😀

The system bios I have on that motherboard doesn't let you peek at expansion cards unfortunately. I know that with PCI cards you have to set up interrupts at DMA sometimes, but there's just nowhere I can turn to to do this in this case

Spiffles wrote on 2024-08-08, 16:56:

The system bios I have on that motherboard doesn't let you peek at expansion cards unfortunately. I know that with PCI cards you have to set up interrupts at DMA sometimes, but there's just nowhere I can turn to to do this in this case

oh no, this has nothing to do with irq or dma, i think, more with device identifier, though it's only a guess.

In linux, one wouId use lspci to look at the info. I'm sure there are other tools that are usable in dos.
maybe some of the tools like astra or hwinfo16 can help in this situation.
Re: ASTRA - Advanced System Information Tool for DOS
Introducing HWiNFO16 (for 8086/88/186/188/286)
If you want to give a go at figuring out what the problem is.

Maybe a scsi bios update is possible and fixes the issue ?

Spiffles wrote on 2024-08-07, 20:06:

It's a Biostar mobo specifically chosen to be able to have dual floppies and all that other junk while also having Windows 10

What does Windows HW manager say about the card? Is it detected, drivers loaded fine? For a card that old MS should have suitable drivers already in the OS.
What about BIOS, any PCI related options to toggle? Passive release, delayed transactions, etc?

If the BIOS of the card indeed is trashed, that could explain why.

However, from what I can find there's two variations of the 2940/2940W, one that has a surface-mounted PLCC EEPROM and another (older?) with a DIP EEPROM.

Sounds like you're going to have to use an EEPROM programmer to reflash the BIOS

However from that page, they state the following:

Adaptec support wrote:

In March 1996, the internal ID of the board was changed. This required a change in the BIOS code. In order to process a request for a BIOS upgrade EPROM for the above assemblies, it is necessary to provide us with the checksum (CS) from the label on the installed EPROM chip. If requesting a BIOS upgrade, contact us via ASK Us and provide a mailing address.

Still there's probably an utility to configure the card from DOS or Windows without having to use the card's built-in BIOS config utility.

Deunan wrote on 2024-08-08, 23:50:

Spiffles wrote on 2024-08-07, 20:06:

It's a Biostar mobo specifically chosen to be able to have dual floppies and all that other junk while also having Windows 10

What does Windows HW manager say about the card? Is it detected, drivers loaded fine? For a card that old MS should have suitable drivers already in the OS.
What about BIOS, any PCI related options to toggle? Passive release, delayed transactions, etc?

The card is detected fine in the device manager and won't let me update the driver, saying there's already the newest one. Regarding PCI settings I don't have many available, only stuff like PnP system installed. There is no way to view expansion cards currently installed or anything like that.

These are not per-card options, more like PCI in general. Usually found in the PnP / PCI BIOS menu. If the mobo is new enough to support Win10 then perhaps a card this old causes some compatibility issues? I'm not sure if PCI 2.3 is fully backward compatible with 2.0, there should be keys in the slot but who knows how the mobo was made and what BIOS sets up. Though it might just be a faulty card.

Ok, I've replaced the SCSI controller with an AHA-2940AU with bios version 1.32 and it's detected fine, SCSISelect opens ok and recognizes both the host adapter and the drive. So I think the adapter definitely isn't faulty. However, the problem in Windows still persists, inserted disks cannot be read and the error message is "I: is not accessible. The request failed due to a fatal device hardware error". Again, mechanically the drive seems OK, it accepts and returns drives and the LEDs display in correct colors. Might I still be missing some driver (maybe for Windows), or could this be a SCSI setup thing?

EDIT: I was able to get the drive to open a 9.1 GB disk one time. It failed to format it, but the volume became accessible and I was even able to put some data on it and then delete it. But after resetting, I've been getting the same "hardware error" message for this disk and other ones too. Makes me think it's either some mechanical defect with the drive or just lack of proper drivers.

This is what I get when I try to check a drive in SCSISelect. I'd appreciate if you could let me know if it's a mechanical problem with the drive, so I can return it.

Or could it be that I'm missing some jumpers still? The termination one is definitely set (jumper on A11/B11). Here's the config of the jumpers, are there any that I'm still missing? Maybe the one for enabling fast SCSI?

Spiffles wrote on 2024-08-12, 09:28:

This is what I get when I try to check a drive in SCSISelect. I'd appreciate if you could let me know if it's a mechanical problem with the drive, so I can return it.

Could be improper termination or cable damage. If your cable has other free connectors - try them. Usually you'd want the two opposite sides, as to not leave any dangling splits that would cause interference, but it's not that bad that it wouldn't work. Not with an older 50-pin cable anyway.

Another option is to try and limit the SCSI speed in the card setup for that particular device. Disable sync transfers maybe.

Deunan wrote on 2024-08-12, 11:45:

Could be improper termination or cable damage. If your cable has other free connectors - try them. Usually you'd want the two opposite sides, as to not leave any dangling splits that would cause interference, but it's not that bad that it wouldn't work. Not with an older 50-pin cable anyway.

Another option is to try and limit the SCSI speed in the card setup for that particular device. Disable sync transfers maybe.

I've got both terminator and terminator power jumpers on the drive, so I think it's terminated properly. As for the cable, it's got 3 plugs ans I've tried all 3 possible configs with them - one end to the nearer plug on the other end and one end to the farther one - both sides. Still the same error. Regarding lowering the speed, scsiselect doesn't let me adjust that option. So I might try another cable, although my current one appears to be good enough for the drive to be detected by the host controller, so I really dunno. Hoping it's some bad settings since this drive cost me a fortune

Spiffles wrote on 2024-08-12, 12:10:

I've got both terminator and terminator power jumpers on the drive, so I think it's terminated properly.

Terminator power is a tricky thing. I'd try to disable it on the drive side. Then try to disable the termination on the drive completly - see if it changes anything. Try disabling Ultra on the drive as well, via A9 jumper (but leave the termination on) if the card doesn't seem to offer link downgrade options. I only have AHA-2940UW and UW2 cards, these do have plenty of link options.