Finally found an MX400 that posts on this mobo! A pretty one at that!

Unfortunately not all that stable of a combo so far. For me.

Anyone know the HDD size limit for this motherboard? Is an 8GB CF card safe to use?

8 GB is perfectly fine without using XTIDE. If you keep your boot partition(s) within the first 8 GB, you can also create a much larger extended partition/logical drive which is usable in w95c or NT4. I recently ran a test on a HDD with:

4 GB NT4-NTFS,
4 GB w95c-FAT32, and
53 GB FAT32 as extended/logical.

and without XTIDE.

Perfect! Thanks feipoa 😀

However, I personally use the XTIDE BIOS on a 3c515-tx network card and run this configuration with a 64 GB Transcend Industrial CF170 card:

partition 1: WINNT4, NTFS, 7797 MB
partition 2: WIN95C, FAT32, 53262 MB

Because I don't like having to part out my Windows partition into two. The caveat being that the IDE FIFO mode won't work in DOS if the boot drive does not reside within the 8 GB boundary, even with XTIDE loaded. I'm not sure why this is. If FIFO mode in DOS is important to you, then keep the boot drives under 8.4 GB, or whatever that "8 GB limit" may be.

feipoa wrote on 2021-10-21, 08:09:

However, I personally use the XTIDE BIOS on a 3c515-tx network card and run this configuration with a 64 GB Transcend Industria […]
Show full quote

However, I personally use the XTIDE BIOS on a 3c515-tx network card and run this configuration with a 64 GB Transcend Industrial CF170 card:

partition 1: WINNT4, NTFS, 7797 MB
partition 2: WIN95C, FAT32, 53262 MB

Because I don't like having to part out my Windows partition into two. The caveat being that the IDE FIFO mode won't work in DOS if the boot drive does not reside within the 8 GB boundary, even with XTIDE loaded. I'm not sure why this is. If FIFO mode in DOS is important to you, then keep the boot drives under 8.4 GB, or whatever that "8 GB limit" may be.

Well, I do have a 3C905 with an empty BIOS socket in mine so I guess I may as well do this? I never set up XTIDE before, do I just download the BIOS and flash it onto something, then stick it onto the NIC?

The XTIDE directions are here: https://xtideuniversalbios.org/

But in brief, you'll want to download ide_386l.bin and xtidecfg.com from here: https://www.xtideuniversalbios.org/binaries/r620/

Setup the address for the boot rom using the 3com DOS config program. Also tell the config to enable the boot rom.

Then go to the folder you saved xtidecfg.com and ide_386l.bin to and run xtidecfg.com. Load ide_386l.bin and set the configuration settings you want. Save it. Now use an external programmer to write your saved ide_386l.bin to a 32Kbyte eeprom. Put the eeprom back into the 3com NIC. Go into the MB's BIOS and enable shadowing for the address you setup in the 3com config utility.

If all went well, you should be set. The reason for programming the eeprom in the external programmer is because I don't think it can be programmed while in the 3com card using xtidecfg.com.

Some of my struggles with this are recorded in this dedicated thread: IDE support for more than 8 GB HDD storage on Biostar MB-8433UUD . Note that the main point of struggle was different notation for setting the EEPROM's address, in particular an extra digit.

Thanks again feipoa. As a final note, how do you handle Win9x/NT caching on CF cards? It's not only slow but also rather destructive, do you completely disable virtual memory?

I haven't done anything with caching. For the frequency in which we use these systems, would it make much difference?

I'm trying to get this board to work with a PCI Riva 128, and after installing Win95 drivers I get protection errors at boot. Apparently this is usually the case when an IRQ is not assigned for the card in BIOS, but for me this option is Enabled (v2014). Do I also need to enable Plug n Play OS option, I wonder?

EDIT: Nope, Plug n Play OS option did nothing either..

EDIT2: I managed to get the drivers loaded using the ASUS V3000 CD instead of vanilla nVidia drivers.

If the Riva128 is anything like the original TNT, you'll want to use Detonator v6.50, assuming you are using a CPU with only 486 instructions. Which NVIDIA driver version did you try?

feipoa wrote on 2021-12-07, 02:13:

If the Riva128 is anything like the original TNT, you'll want to use Detonator v6.50, assuming you are using a CPU with only 486 instructions. Which NVIDIA driver version did you try?

Well, I first tried the final Ria 128 PCI driver from nVidia (v3.37): https://www.touslesdrivers.com/php/constructe … php?v_code=9637

This is not a Detonator driver, I don't think those have pre-TNT bundled. Anyway, those gave me Windows Protection Errors.

I then tried the ASUS V3000 driver CD: https://www.vogonsdrivers.com/getfile.php?fil … &menustate=14,1

That seems to have worked. I'm not sure that the 486 CPU is up to the task for hardware 3d acceleration with this driver though, at least for OpenGL. I tried GLQuake and although it launches and I can walk around, it is VERY slow, even at 320x200 it is slower than software. It also can not complete timedemos, always crashes. Maybe the OpenGL ICD in this driver is premature or something. I haven't tried any Direc3D games.

I have another issue maybe you can help with. I tried to set up PLPBT5 boot manager to boot betweel C: (Win95C) and D: (MSDOS6.22) but when I installed it to the MBR it wouldn't boot. Any ideas why? PLPBT requires pentium instructions or something?

So the early Detonator drivers stripped Riva128 support? OK, I didn't know that.

Unfortunately, I don't have experience with PLPBT5. On my infinite to-do list is to find a suitable means to boot between w95c and MS-DOS 6.22 on different hard drives. My temp solution was to use SCSI HDDs and change which drive ID gets booted to in the SCSI controller's BIOS. Its not elegant. Other people use CF cards and just eject/insert which OS they want. Another approach is to use SCSI2SD's config program to enable/disable SCSI ID's. Ideally SCSI2SD would have a DOS app. which was able to enable/disable SCSI ID's, but it doesn't. All of these are inelegant workarounds.

If you get it working, please send me the thread link. I think something called System Commander was popular in the 90's.

Thanks. I couldn't get PLPBT5 working so far, if I get anything working I'll let you know. The best I could do so far is keep them on separate IDE devices, one on Primary Master one on Secondary Master. I disable detection for the Primary Master in BIOS to boot from Secondary Master (DOS), otherwise it boots to Win95. Again, inelegant. I normally love PLPBT5 but it just didn't work here. I'll try PBM6 as well. I'll also look into System Commander.

I'm having a bit of a strange issue with this board. My AT keyboard's membrane partially died, so a block of keys no longer work. Hence I tried to hook up a PS/2 keyboard with a PS/2 to AT adapter, but it would always give me keyboard errors at POST. Thinking the adapter is bad, I replaced it and the issue remained. Thinking the keyboard was the issue I got a new PS/2 keyboard and it exhibits the same thing. I get the LEDs flashing at boot, but it does not get recognized. The first keyboard I tried was a retro HP board that, according to what it says at the bottom, is rated for 5VDC and 500mA max (which is a bit extreme?), and the new one I bought is 5VFC 30mA (which sounds fine?). The second board is a USB board that came with a USB to PS/2 adapter included, and is advertised as PS/2 compatible on the box. Not sure if the USB to PS/2 thing is causing extra issues, but I ordered a newer HP PS/2 keybard hoping it will work fine. Anyone else had keyboard compatibility issues with this thing?

Is the battery on your RTC fully charged? Try resetting to BIOS defaults/ CMOS defaults, then put your settings back to what is optimal. What speed are you running the keyboard clock at? If I recall correctly, there's 7.x MHz, or PCI/2, PCI/3, PCI/4.

With respect to socket 3 boards, I've only had this kind of keyboard compatibility issue with a Shuttle HOT-433 v1. I have two such boards currently, and one of them doesn't like a particular Microsoft ps/2 keyboard, while the other HOT-433 v1 had no issue with it. I never investigated the issue very deep. I ultimately swapped out the keyboard for a Cherry mechanical board and the issue disappeared.

feipoa wrote on 2022-08-04, 06:04:

Is the battery on your RTC fully charged? Try resetting to BIOS defaults/ CMOS defaults, then put your settings back to what is optimal. What speed are you running the keyboard clock at? If I recall correctly, there's 7.x MHz, or PCI/2, PCI/3, PCI/4.

With respect to socket 3 boards, I've only had this kind of keyboard compatibility issue with a Shuttle HOT-433 v1. I have two such boards currently, and one of them doesn't like a particular Microsoft ps/2 keyboard, while the other HOT-433 v1 had no issue with it. I never investigated the issue very deep. I ultimately swapped out the keyboard for a Cherry mechanical board and the issue disappeared.

Recently the Dallas RTC was having issue keeping date/time but reatined settings, so I swapped it out for a newer (well, relatively - the new one is from 2013 🙁 ) working one and it keeps time quite fine. I reset to Setup defaults, so the KBC frequency is 7.14MHz but I also tried PCI/4 which did not help. The FSB is at 40MHz but I doubt that is what causes the issue, as it POSTs fine with my AT keyboard (except that one has some keys not working anymore..)

Also, it POSTed fine with the new USB/PS2 keyboard twice in a row after reboots, but when I went back to the HP keyboard the issue returned. That kind of prompted me into thinking maybe it's a contact issue, ie. the adapters I am using are not making proper contact with the keyboards or motherboard port, but I've done so much wiggling and reinserting to no avail that I doubt this is the case. Also, I tested the AT/PS2 adapter with a multimeter, all the correct pins on each side have continuity. The only thing that has no continuity is the connectors' outer edges (which should normally be ground, but are not connected). Could this be of importance?

I have a feeling that whatever circuitry supplies power to the keyboard port is not generating enough juice for the PS2 keyboards I am trying. I don't know how to test this hypothesis though. Let me try another PS/2 keyboard and report back, I should get the board in a few days. AT keyboards are hard to get these days and I'd really not be left without a keyboard for my favorite retro PC 🙁

I suppose wiring in the multi-meter's ammeter [in series] to the 5V port is a bit too much effort. I wonder if DMM's now have those clip on loop ammeters like they do for larger single AC wires?

If you have one keyboard that works fine, the issue is likely not with the motherboard.

feipoa wrote on 2022-08-04, 07:35:

I suppose wiring in the multi-meter's ammeter [in series] to the 5V port is a bit too much effort. I wonder if DMM's now have those clip on loop ammeters like they do for larger single AC wires?

If you have one keyboard that works fine, the issue is likely not with the motherboard.

Only AT keyboards I have at hand seem to work fine, I'll try more PS2 keyboards.. I also ordered an AT keyboard by Genious, a fairly late model from 96 with windows keys and a sleep button - I'll see how that behaves..

I actually tried a disgustingly fithy IBM Netvista PS2 keyboard tonight and it worked. The board is just being picky somehow but I do not know how or why.

Now I am trying to clean this terrible mess of dead insects and human filth.. 🙁