Laptop IDE drives have master/slave status set in firmware, not jumpers. This may cause problems if you have another device on the channel.

That's not it. As I said, it works with an ESS1868 on the same system. And HWINFO sees it as slave even with no master (in the working config).

Well, I just tested some things, and it seems the 1st time I use it with another device on the cable, it starts responding on the same ID (M/S) as the other device (tested with ZIP and CDROM). Apparently if I switch M/S of the other device, at 1st they both appear ok in HWINFO (at least if I query the other device before the slim), but if I requery any device, they both show corrupted info.
On ESS, seems to behave like always taking the setting of the other device (corrupted IDs and only 1 device appears). But alone it just works.
So yeah...it could be a master/slave problem.
But I never intended to use it with another device. So maybe even on Pentium it will clash with another device on the cable, but works good by itself (used to copy Win install files, so no corruption has occured).

Interesting how posting motivates me to try something else!

These drives use the 50-pin JAE-IDE connector that combines power, analog audio and data . On the full size drives these three things require separate connections.

Errius wrote on 2021-11-03, 22:41:

These drives use the 50-pin JAE-IDE connector that combines power, analog audio and data . On the full size drives these three things require separate connections.

That's why I have an adapter. Actually 2 adapters and I tested 3 laptop drives (well, only 1 extensively, but the simple "not working" on multiport+IDE was done with the other two as well).

interesting. I guess the adapters you have do not have a master slave jumper ?

No, you have to hack the drive firmware to change that.

Errius wrote on 2021-11-04, 16:22:

No, you have to hack the drive firmware to change that.

Fairly 100% certain, you do not.

Interesting, what adaptors are you using?

I haven't used any adapters yet.
I can only tell you from experience with older laptops, that the same physical cd/dvdrom works in different devices as either a master or a slave, as a single drive or a slave drive to the hard disk, without any firmware or even hardware modification.

That sounds like cable select. Not all IDE controllers/drives support that though.

Yes, cable select.

and let's just say that any controller/drive prior to 1994 doesn't support it.
then again, the controller doesn't even need to support, nor even care about it, with some cable hackery/magic.

Errius wrote on 2021-11-04, 16:35:

That sounds like cable select. Not all IDE controllers/drives support that though.

I suspect cable select is indeed the explanation. On a standard 40-pin IDE connector, CSEL is on pin 28. On the 50-pin connector, that pin is pin 47. The host is supposed to ground pin 28, and cables that support cable select are supposed to pass the ground level to one drive (master), and have a pull-up resistor pull this signal high on the other drive (slave). Possibly your old ISA IDE interface cards do not ground pin 28 as specified, and your DVD-ROM doesn't have the required pull-up resistor to pull the CSEL pin high if it is not grounded, so the drive might be in an completely inconsistent state, shifting around between being master and slave as it likes. If you have a multimeter, you should be able to check continuity between ground (e.g. on pin 26) and CSEL on pin 28 on your different IDE host variants.

I wonder a lot whether half the problems with compact flash cards and other media to IDE adapters boil down to that. But when ppl are already unfamiliar with PATA, never mind older versions of it, I feel like I may as well be yelling warnings/instructions at films of Charlie Chaplin, Laurel and Hardy etc.

mkarcher wrote on 2021-11-04, 17:18:

I suspect cable select is indeed the explanation. On a standard 40-pin IDE connector, CSEL is on pin 28. On the 50-pin connector, that pin is pin 47. The host is supposed to ground pin 28, and cables that support cable select are supposed to pass the ground level to one drive (master), and have a pull-up resistor pull this signal high on the other drive (slave). Possibly your old ISA IDE interface cards do not ground pin 28 as specified, and your DVD-ROM doesn't have the required pull-up resistor to pull the CSEL pin high if it is not grounded, so the drive might be in an completely inconsistent state, shifting around between being master and slave as it likes. If you have a multimeter, you should be able to check continuity between ground (e.g. on pin 26) and CSEL on pin 28 on your different IDE host variants.

Well, neither the working ESS sound card or non-working multi-io boards put pin 28 to ground. It seems to be floating (if it's connected, the trace is under the header plastic...so not likely).
But as I said, ESS only works when it's alone on cable (so CSEL is the corruption with 2 drives).
I have an extended weekend comming, so I'll try to measure other lines with oscilloscope, see what voltage ranges they use. Also I'm waiting for a simple 1-IDE-only ISA card.

As for the adapter, it's this one: https://www.ebay.com/itm/312587475807
It also has an audio port.