To use a 28 pin eeprom, you will likely need to cut some traces and move some wires. Thankfully the JEDEC standard pretty much stays the same for the lower pins, so your mod would involve only a few wires.

You might find this useful:
https://w140.com/tekwiki/wiki/2764

You can see that you would probably need a jumper from pin 32 to pin 30 (which will be pin 28 on your 28 pin eeprom) and since this is NC typically, it may already be jumpered on the board.
Pin 1 is going to be A15 on the 32 pin socket, but it will be VPP (high voltage programming voltage) on a 28 pin 27C256 (32Kx8) chip. If you are accessing only 32K (which you would be), this will always be zero, so not an issue. But you could also cut it and tie that pin to ground. It may even have a pull down resistor on the board since many of these were designed for both 32 pin and 28 pin already.

XT-IDE is hardware as well as the bios extension. What do you plan to do for the hardware interface to the IDE drive?

That motherboard looks like it already has an integrated IDE controller. So what is it that you are wanting to do?

Retroplayer wrote on 2025-09-02, 07:08:

XT-IDE is hardware as well as the bios extension. What do you plan to do for the hardware interface to the IDE drive?

This is true. The use case of the OP is most likely to just use the BIOS extension (more precisely called the "XUB", which is an abbreviation for "XT-IDE Univerasl BIOS"), which is a modern IDE BIOS that overcomes the 524 MB and 8GB limits in any kind of system. The XUB does not only support interfacing the special 8-bit XT-IDE hardware, but it can also interface with standard AT IDE hardware if configured accordingly.

Mao wrote on 2025-09-01, 23:04:

I have read here on vogons that some people have happily used 28-pin eeproms on this card, but I am not sure how a smaller (ie, 2-pin shorter) chip must be insterted in a larger socket: the two empty holes must be left at the beginning (toward the notch on the socket) or at the end?
Anyway, before I even try to configure the bios, the card, the sowtare, etc., I ask you to suggest what EEPROM chip I need to look for my card.
Thanks

My guess is that the two empty pins would be left at the notch end, yes.

If this diagram matches your system, you would need to jump Vcc back two spots with a short wire if Vcc is not present on pin 28 of your DIP 28 eprom.

really makes no sense that the card would have a 32 pin socket and not support at least a 128kB ROM

maxtherabbit wrote on 2025-09-02, 12:36:

really makes no sense that the card would have a 32 pin socket and not support at least a 128kB ROM

This is one weirdo of a card. 100Mbps on ISA through a PCI bridge.... I'm not sure there's even a packet driver for it.

Looks like it only had netware and ndis drivers for DOS. It shares some chips with the 3C590. Looks like the rom is connected on the other side of the bridge, so the rom might need to be a pci rom instead of ani sa rom.

Makes me wonder if anyone ever tried using that bridge chip to put a usb card on on ISA bus.

Thank you very much for all your answers. You gave me very intersting pieces of information. As said, unfortunately I am not an expert and am not able to modify those sockets by soldering wires around, so unless a 28-pin chip is usable as is, I would start by looking for a 32-pin eeprom (but which one!?). It is really annoying that the manual contains so little information. I will post here some pictures of the relevant pages as soon as I can.
Thank you again

Mao wrote on 2025-09-03, 08:38:

so unless a 28-pin chip is usable as is

Usually, these cards accept both 28- and 32-pin chips. If you happen to have a continuity tester, or you can see the that trace on the PCB, check whether pins 30 and 32 of that socket are directly connected to each other. If it is the case, a 28-pin ROM chip will work out-of-the-box. This is what douglar already mentioned, and it is very likely that the red connection in that post is already present.

ok, will do! thanks

Hey,
I attach here some pictures of the card and the relevant pages from the manual. I can't see a connection between pin 30 and 32, but you tell me. Thanks

a better picture of the side

You don't see any connection to pins 30 and 32, because this card appears to be a 4-layer board with +5V and GND on two inner layers. This means just looking at the card doesn't help you to find out whether both pin 30 and pin 32 are connected to 5V (I still expect they are both). If you have any kind of 28-pin ROM (VGA BIOS?), you can try to plug it into the card (align at the side without the notch, two pins in the socket unpopulated near the notch), and check whether the ROM is seen by the system. In the case of a VGA ROM, you might get a crash during POST after the memory test, when the second VGA ROM tries to do something with the VGA card. Still, this would confirm that 28-pin ROMs do work in this card.

hey, thank you very much. I have ordered some 28-pin ROMS from my favorite chinese supplier; they are on their way and expected by next week, I suppose. I will let you know.

Mao wrote on 2025-09-03, 08:38:

Thank you very much for all your answers. You gave me very intersting pieces of information. As said, unfortunately I am not an expert and am not able to modify those sockets by soldering wires around, so unless a 28-pin chip is usable as is, I would start by looking for a 32-pin eeprom (but which one!?). It is really annoying that the manual contains so little information. I will post here some pictures of the relevant pages as soon as I can.
Thank you again

If you're just worried to cause damage to the chip or NIC socket, you can always leave them untouched and do the bridging and pin cutting to another IC socket sitting in between them.

I received some chips from Aliexpress and tried to program them with a brand new and never used before XGECU T48. Unfortunately, the programming software always shows an ID Check error when I want to start to burn the chip. I tried three different chips of two different kinds. I can unflag the ID check so that the chip gets programmed, and at the end I can see the chip contains xt-ide, but when I stick the it on the NIC and turn the pc on, it boots as usual, no trace of XT-IDE. ROM boot and memory address are correctly set on the card via its configuration software. I might have chosen the wrong chips? they are AM27C256 (tried 2) and SST 27SF256 (tried 1); is my T48 faulty; what else could it be? thanks

Looks like I did it! It worked with the ST27sf256 chip after I modified the bios on the Vectra and reduplicated the file before programming the chip. Unflagging the ID check on the T48 caused no issue in the chip programming procedure. There is stil some minor issue to solve due to a "system board error" during POST, which might be caused by some conflict with the NIC, but it won't prevent booting. I only need to press F1 by hand. I hope I will find a way to overcome this too. Thank you very much for all the information!