Yes possible. Small socket is for LS138? Difficult to read, but should be required for using the ROM.

Predator99 wrote on 2020-03-10, 11:43:

Yes possible. Small socket is for LS138? Difficult to read, but should be required for using the ROM.

How, what ROM? I read 74LS138, and what exactly is it for?

Think somebody else can explain better than me what the 74LS138 is needed for 😉

XT-IDE would fit very well on that card, you can use i.e. W27E512 EPROMS.

Predator99 wrote on 2020-03-24, 18:13:

Think somebody else can explain better than me what the 74LS138 is needed for 😉

XT-IDE would fit very well on that card, you can use i.e. W27E512 EPROMS.

What XTIDE ROM file should I use? AT 8KB or 12KB ? Will 27C512 work with it?
I read something here and worry if it doesn't work:

http://www.vcfed.org/forum/showthread.php?434 … BOOT-ROM-socket

Any reason why you want to add a ROM at all? It's not needed for the controller to work, since the BIOS can use it just the way it is.

Also, the XT-IDE Universal BIOS has no special support for VLB. You will most likely lose any way to use the controller in 32-bit mode when installing XT-IDE.

27C512 is printed on the PCB therefore it should work. I would start with a rewritable one like W27E512. You can use 8kb or 12kb.
If your ROM is larger than the image file you may need to fill it. So copy the 8kb image 8 times on the 64kb ROM-chip to fill it up with data.
You may need some tries to get this running therefore a rewirtable one is a good idea...

derSammler wrote on 2020-03-24, 19:02:

You will most likely lose any way to use the controller in 32-bit mode when installing XT-IDE.

Oh, thanks for it.

Hi! Did you try to find the original BIOS dump?

No, I could not.

Greetings everyone! I can see this thread went stagnant so I'm going to give it the closure it needs.

I too have this exact same card and I followed all of the instructions on here so I now have a nicely working card with large drive support. It seems the smaller chip is a logic inverter of sorts and yes it's needed for the BIOS chip socket. To find it just search for 74LS138 on eBay. i think mine had PC after the name.

For the ROM chip, I used a Winbond W27C512 64KB DIP28 EEPROM chip.

As for the BIOS, I configured XTIDE r625 in DOSBox using a wrapper of sorts I created from an existing GOG install and editting it to suit my needs to use the 32-bit type on both master and slave IDE controller ports on the 386L.BIN file. I then saved the settings to the BIOS file. Next, I made a 60KiB file from the 12KiB file with "copy /B 1.Bin + 1.BIn + 1.BIN + 1.BIN + 1.BIN BIOS60.BIN". This repeats the 12KiB file five times in a 60KiB file and is needed to properly boot XT-IDE on a 64KB ROM.

Then, I programmed the chip with my TL866CS programmer with zeroes at the end to properly setup a 64KB ROM.

once finished, i plugged in both the EEPROM and the other chip.

I set the jumper from Disabled to D000h and reinstalled the card.

XT-IDE r625 booted right up following the POST

Hope this helps someone.

I should also mention I removed the ground pin 20 on the card's two IDE ports to accomidate 80-wire IDE cables without pin 20 used as a key of sorts to prevent improper cable intallations.

Shouldn't it be repeated at 16k intervals? Or maybe not repeated at all?

I mean, in my old computers (8 bit), it made sense to repeat two times a 16k ROM into a 32k chip (because you don't have to worry about A14 pin, and the size of ROM is fixed).

On a PC, the internal BIOS looks for external BIOS at fixed addresses. So it makes me wonder...
- At which addresses does it look for BIOS? If it looks at 16k intervals, it would make more sense to repeat your BIOS every 16k but...
- What happens if it finds your BIOS more than one time? Does it run more than one time? Does it appears at upper memory more than once? To avoid "BIOS collisions" (more than one BIOS trying to perform the same thing) and having more UMB space available, the BIOS should not be repeated at all.
- What value should be used for "filling" (i.e.: make a 16k BIOS from a 12k one). Should 0x00 be used, or leave as 0xFF (the unused/unprogrammed value on EPROM chips)?

If you repeat your BIOS every 12k, this would happen...
- If the BIOS look at 4k, 2k or 1k intervals, it should find your BIOS at every repetition.
- If the BIOS look at 16k intervals, if should find your BIOS two times (at 0k and 48k).
- If the BIOS look at 32k intervals, it should find only one BIOS copy, so it would be fine.

I wonder how your card is working...

IDK but it works and this wasn't my idea originally. I found it on one of these VOGONS forums. I can't remember which one though.

jkamm1985 wrote on 2023-10-03, 09:28:

As for the BIOS, I configured XTIDE r625 in DOSBox using a wrapper of sorts I created from an existing GOG install and editting it to suit my needs to use the 32-bit type on both master and slave IDE controller ports on the 386L.BIN file. I then saved the settings to the BIOS file. Next, I made a 60KiB file from the 12KiB file with "copy /B 1.Bin + 1.BIn + 1.BIN + 1.BIN + 1.BIN BIOS60.BIN". This repeats the 12KiB file five times in a 60KiB file and is needed to properly boot XT-IDE on a 64KB ROM.

Are you saying you produced an XTIDE version that will support VLB? Do share please!!! Is it compatible with all VLB IDE chips or just the UMC varieties? Thanks

mockingbird wrote on 2023-10-13, 12:14:

Are you saying you produced an XTIDE version that will support VLB? Do share please!!! Is it compatible with all VLB IDE chips or just the UMC varieties? Thanks

It's there already in the large 386 image. https://www.xtideuniversalbios.org/ "Native support for QDI Vision QD6500 and QD6580 VLB IDE controllers"
There is also some experimental native support for Promise VLB controllers.

I did some benching with XUB ide_386l.bin image and an QD6580 VLB IDE controller with rom shadowing enabled for the XUB address range (<--very important!)

My notes were that XUB was a little faster than the native BIOS (no ECHS or LBA support) , pretty much on par with MRBIOS 1.41 (non standard LBA support ) , but slower than both the DTC LBA bios and EzDrive 9.09W (w/ LBA & 32bit access enabled).

I can't remember if the QDI dos driver supports large drives or how it performed. I'll double check that today if I get a chance.

douglar wrote on 2023-10-13, 13:17:

It's there already in the large 386 image. https://www.xtideuniversalbios.org/ "Native support for QDI Vision QD6500 and QD65 […]
Show full quote

mockingbird wrote on 2023-10-13, 12:14:

Are you saying you produced an XTIDE version that will support VLB? Do share please!!! Is it compatible with all VLB IDE chips or just the UMC varieties? Thanks

It's there already in the large 386 image. https://www.xtideuniversalbios.org/ "Native support for QDI Vision QD6500 and QD6580 VLB IDE controllers"
There is also some experimental native support for Promise VLB controllers.

I did some benching with XUB ide_386l.bin image and an QD6580 VLB IDE controller with rom shadowing enabled for the XUB address range (<--very important!)

My notes were that XUB was a little faster than the native BIOS (no ECHS or LBA support) , pretty much on par with MRBIOS 1.41 (non standard LBA support ) , but slower than both the DTC LBA bios and EzDrive 9.09W (w/ LBA & 32bit access enabled).

I can't remember if the QDI dos driver supports large drives or how it performed. I'll double check that today if I get a chance.

Thanks. When you talk speed, what was it like PIO 4? What was the transfer rate?

If you like this setup, then you ISA-only fans out there will surely love this 16-bit ISA-only dual card setup that seems to be equally as fast. The setup consists of a Promise Technologies EIDEMax set as secondary channel with matching hex address as XT-IDE and the default ROM replaced with the same Winbond W27C512 and XT-IDE in 16-Bit 2 channel mode (basically default settings on 386L.bin r625 rom then copied 5 times to 60KiB file and programmed with zeroes). This is teamed up with a Diamond Flower Inc. MIO-400KF Rev. F set as master. I can boot off either channel and Windows 95C OSR2.5 just magically sees the IDE CD-ROM following an add new hardware scan which installs the second channel in Windows. Now, I have 2 channels active and everything seems to fly. And yes those are brand new 80-wire IDE cables with the ground pin 20 removed as a key.

mockingbird wrote on 2023-10-13, 18:24:

Thanks. When you talk speed, what was it like PIO 4? What was the transfer rate?

I'm almost done with the testing grid. Another day or so.

I'm always surprised by the strange incompatibilities of hardware from this period and how a storage device that does quite well with a promise controller might not do well with a QDI controller.

Anyway, XUB 625 386L rom is doing pretty well in the tests, second only to the QDI dos drivers.

The read speeds are in KB/s
The write speeds are all +/- 5% so I didn't put them in the chart.
CF = Compact flash
SD = SD card in a Sinitechi

This is one of those side switching PCs but sadly, it's missing the plastic shroud that sits around the switch. If anyone has a solution to this, let me know as it attaches to the power supply via two threaded holes in the PSU shell.