Have given up trying to disable the onboard ethernet on this thing - despite everything I try it always seems to be enabled according to hwinfo.

Next task is to try and get a working floppy drive (machine came without one). However, that's easier said than done! I have 3 known-good floppy drives here:

Mitsumi D353M3
Samsung SFD321B
Sony MPF920

I also have several floppy cables - a couple of straight connector, single connections (no twist), as well as a traditional twin drive cable with twist.

I am coming to the conclusion that there is something odd about the floppy arrangement on the motherboard because....

With any of the drives and either of the straight through cables (or the twist cable on the second connection), the drive works (as B:)
With any of the drives and the twist cable (on the first connection), the drive doesn't work.
With any two drives on the twist cable, only the second drive works (as B:)

I could live with that setup, but the problem of course, is that the bios doesn't allow booting from drive B and doesn't have tri-mode support, not remapping of A<->B or anything like that.

I am assuming that because these machines were designed for a single floppy drive (no space in the case for anything other than 1x floppy + 1x hdd) that the floppy drives that were fitted at the factory were altered from the default IBM style drive jumper settings to one which would allow a drive on the end of a single, straight-through, no-twist cable to be accessed as drive A.

I've opened the Mitsumi drive and it has 0-ohm jumpers across DS1 and DS3, with DS2 and DS4 open - I guess if I choose to use this drive with this board, I'll need to open DS1+DS3 and close D2+DS4?

So you're thinking that by setting the jumpers on the drive to configure it as the second drive, it will actually work as the first drive? I don't think there's any harm in trying.

megatron-uk wrote on 2023-01-01, 11:46:

Have given up trying to disable the onboard ethernet on this thing

Maybe the chip has a convenient enable pin, or single power input that you can fairly reversibly disconnect?
Hopefully the firmware doesn't care it's not there anymore, then...

megatron-uk wrote on 2023-01-01, 11:46:

[floppies]

The Olivetti PCS 286, which supports 2 drives, has 2 connectors for 2 straight through cables but clearly this isn't the case...

Might the motherboard use the Shugart pinout, meaning you'd need a modified cable - or indeed to set the drives differently?
https://www.bytedelight.com/?p=1418
http://www.seasip.info/VintagePC/floppies.html

No idea about what your drive means with ID 3 and 4, if the meaning is the obvious one you can leave both of them off since someone misconfigured it to appear as multiple drives (one of which isn't ever used in the IBM pinout), but they might be routing different pins to the logic...

It's entirely possible - I've just tried switching from the default DS1/3 jumpers to DS2/4 and I still get the "Floppy A failure" from the POST ... So my 'cunning plan' didn't entirely work.

In fact I have tried every pair of DS settings on the drive as well as each DS jumpers singlely... And it only ever works as drive B.

Searching through previous threads, I can see a few other Tulip pc owners (mainly 486 models though) that have had trouble replacing the original floppy drives... So it does seem that Tulip 'cooked' the floppy interface wiring across several models and generations of machines.

There's almost nothing online in terms of documentation though.

Okay, fixed:

So it seems Tulip did bodge their motherboards to not accept 'standard' PC floppy drives.

I tried all settings I could with the Mitsumi drive, but was limited to just drive select enable IDs (1-4). I knew the Samsung had more options as I recently configured one for use on an Amstrad CPC which needed the ready signal and other mods enabling.

So I started with trial and error; first setting the drive to DS0 (the Samsung is 0-3, rather than 1-4 like the Mitsumi) and it stopped complaining about Drive A at the post... But gave "Drive ready" errors when attempting to access in Dos ... But it did spin the motor... So it was a start.

Next I looked at what other jumpers were available and one stood out - Disk Change. So with that changed to remove the default 0 ohm resistor (defaults to the resistor fitted - I have 3 Samsung drive like this ) I tried again and still had a lack of Post errors, but this time I also got a normal response in Dos accessing A: and the PC booted from floppy for the first time since I put it together.

Tldr; Tulip PC's need floppy drives setting to DS0 (or whatever your floppy drive uses to indicate the FIRST drive signal) AND you must have the Disk Change signal line set as above
I suspect if you do not have a disk change option then you probably can't use the drive in a Tulip. My guess is the Mitsumi drives have a pad for it somewhere, but it's not labelled.

Does the floppy cable you are using have a twist in it? It's not visible on the photo.
If it doesn't, that would explain the need to switch to DS0.

I have both twist and non-twist cables. Attached as the end drive on a twist cable a standard 3.5 drive never gets detected as A or B. As a second drive or on the end of a straight cable all drives were *always* detected as B:. Changing DS jumpers on the Mitsumi drive made NO difference whatsoever.

It looks like the Tulip motherboard does the 'twist' onboard and expects drives to be jumpered as DS0 (rather than the typical IBM standard of DS1), as well as requiring the disk change signal to be set as above.

It didn't matter what cable or position on the cable the unmodified drives were connected with, none would ever respond as drive A.

Good work. I have an old (Compaq or maybe a P.Bell, I forget now) motherboard that only accepts a straight cable and drive at DS0 to make work as drive A, no other cabling worked for some reason iirc.

Finally got around to benchmarking this system. It's a good, solid performer, perhaps let down a little by the 72pin SIMM memory implementation (rather than the optional SDRAM interface of some 430VX implementations):

In particular I'm very impressed by the compact flash performance of over 10MB/sec on the primary IDE controller.

I'm fitting a Voodoo 1 (4mb Diamond Monster 3D) to this system alongside the Trio64 (2mb Diamond Stealth 64 DRAM) as it feels like it's the right type of machine to have it. It's also now got an ESS1868F + XR385/DB60XG.