VOGONS


First post, by davidrg

User metadata
Rank Member
Rank
Member

Over the last week I've been having real trouble getting this EXP4045 motherboard to work. Only hardware installed is the Am486DX4-100 CPU, some RAM and a random Octek 16bit ISA video card (I haven't found my box of VLB cards yet). Here is a picture of the thing stolen from Ultimate Retro - mine looks pretty much identical (same CPU, same not-yet-leaking soon to be removed NiCd battery which surprisingly still holds a charge):
prw-fm20-81-5fd7b5c5174ad701015919.jpg

With the video card in the first VLB slot It would only very rarely beep and start POST only to cancel the memory test, complain about a memory error and also a keyboard error. I tried bunch of different RAM to no effect. Even no RAM at all did nothing - not even a series of beeps to complain about its absence. I checked the jumpers in case the board happened to be incorrectly configured and there were two missing (JP36 and JP32). Installing them made no difference.

I had basically given up on the thing and pulled the video card out when I decided to give it another go before I took the motherboard out of the case. This time it beeped to complain about the missing video card I guess. I put the card back in the second ISA slot and the thing started up just fine. No memory error, no keyboard error, I could get into setup just fine. I put the video card back in the first VLB slot and the problems returned. Fine, I thought, perhaps this video card is just a bit weird. I put it back where it was happy and proceeded to install an IDE controller.

I grabbed a new old stock Pine PT-627B VLB IDE controller, put it in the first VLB slot. No success. Tried the other two slots (one of which is marked "MASTER (W/SMM)" whatever that means) and got the same result. Regardless of which VLB slot that card is in it never gets through the memory test without an error. Sometimes I get two POST beeps. Sometimes it beeps a few times during the memory test. I tried removing the JP36 and JP32 jumpers I installed and that resulted in the CPU being detected as an 80486DX2-66, memory tests passed, and the VLB IDE controllers option ROM ran though it didn't seem to go anywhere from there but that might just be because no disks were connected.

I don't think I've ever run into this before. I'm sure whenever I've installed VLB cards in the past (admittedly probably 15-20 years ago while I was in High School) they just worked - or at least the system got through POST fine before any stability problems appeared. Is this motherboard just faulty? Or is the Am486DX4-100 CPU somehow incompatible with using VLB cards in this motherboard? Anyone have any ideas or experience with this motherboard?

If I can't get this going properly then I might swap it out with a Biostar MB8433UUD-A (assuming it works) as I think a DX4-100 should probably have something better than just 16bit ISA video and IDE. The Biostar motherboard likely needs a new RTC chip at this point though and it will be a few weeks before some DIP-24 sockets arrive to do that job.

Reply 1 of 5, by Unknown_K

User metadata
Rank Oldbie
Rank
Oldbie

Make sure the 486/100 was running at 3x33 and not 2x50 which will cause all kinds of hell on VLB cards and the ISA card speed as well.
Settings for 486/100

JP16 should be open for 3x mult.
JP31: open
JP32: closed
Jp34: open
JP36: closed

JP7/8/9: closed for 100Mhz

JP21:Open
JP22:Closed
JP23:Open
JP24:OPEN
JP25: 1&2, 3&4

What voltage is the CPU?
3.3V JP18, JP33 OPEN
5v JP18, JP33 Closed

Collector of old computers, hardware, and software

Reply 2 of 5, by davidrg

User metadata
Rank Member
Rank
Member

Thanks for the reply! All those settings match what I've got currently. It seems pretty stable with 16bit ISA cards only - I've got some weird video artifacts running games but the temporary VGA card I'm using doesn't look too great so not surprising if its faulty.

I ended up re-reading the PT-627B manual and wondered if perhaps the weird "Speed Selection for HDD" table was really talking about the bus speed. The manual doesn't seem to explain what the table is or what all the numbers in it are but I changed the jumpers from "Under 40MB HDD" to "25/33MHz Default" and that seemed to fix the problem! The system now starts up with the VLB card though the floppy drive test always fails but that's probably the fault of the card - it boots DOS from the disk fine as long as I press F5 to not use the cards option ROM. I think when I locate my box of VLB cards I'll try find a different card - one that doesn't make me press a key on startup to either run its option ROM or not.

So it looks like VLB works ok on this board and the problem was a jumper setting on the particular VLB card I'm using. Weirdly the 16bit ISA video card now seems happy in the VLB slots too where as previously it wasn't - not sure what's going on there but its working so I won't question it further.

Reply 3 of 5, by Unknown_K

User metadata
Rank Oldbie
Rank
Oldbie

All slots are 16 bit if you are not using a card that taps into the brown connector.

Collector of old computers, hardware, and software

Reply 4 of 5, by Intel486dx33

User metadata
Rank l33t++
Rank
l33t++

You need to use FPM Memory ( Fast Page Mode ) Non Parity
If you use Parity memory you need to occupy the two bottom memory banks first.
Install the Memory sticks in pairs.
Start with the two bottom memory banks
Or the banks marked ( bank-0 and bank-1 ).

Also cut off the barrel battery and use a coin battery adapter.
Attach it to the external battery connector.

Reply 5 of 5, by davidrg

User metadata
Rank Member
Rank
Member
Unknown_K wrote on 2022-03-18, 09:42:

All slots are 16 bit if you are not using a card that taps into the brown connector.

Yeah, thats what initially made me suspect there was something wrong with the motherboard. I initially had some random 16bit OTIVGA card in a VLB slot as thats where I intended to install a proper VLB card once I'd tracked one down. But with that card in any of the VLB slots the board wouldn't POST. Now I suspect it might just be a faulty video card - should have grabbed another one out of the box.

Though really I was just surprised the board didn't burst into flames when I powered it up the first time. I've had bad luck with tantalums lately - a few weeks ago I powered on an old DECpc 325sxLP for the first time in probably 10-15 years and a tantalum on the ISA riser released the magic smoke. Then an hour or two later I got out my old HP scanner to scan something and it did the same. My office needed a lot of airing out. Combined with some SPARCstations and AlphaStations that won't power on and a Prioris LX that is powercycling on its own I've got some capacitor shopping to do.

Intel486dx33 wrote on 2022-03-18, 11:38:
You need to use FPM Memory ( Fast Page Mode ) Non Parity If you use Parity memory you need to occupy the two bottom memory bank […]
Show full quote

You need to use FPM Memory ( Fast Page Mode ) Non Parity
If you use Parity memory you need to occupy the two bottom memory banks first.
Install the Memory sticks in pairs.
Start with the two bottom memory banks
Or the banks marked ( bank-0 and bank-1 ).

Also cut off the barrel battery and use a coin battery adapter.
Attach it to the external battery connector.

Yeah, until two days ago I didn't know the difference between FPM and EDO. I guess I thought most/all 72pin RAM was EDO and FPM was just something weird one or two odd machines needed. Guess that explains why I used to have such trouble finding RAM that would work in some machines. At least now my mystery box of memory is now sorted by type (FPM/EDO/EDRAM) and capacity so it shouldn't be an issue in the future.

Surprisingly that barrel battery is still working and holding a charge! But I've learned the hard way what those things will do if left to their own devices - I'm surprised the board managed to make it into storage with the battery still on it. I've ordered a bunch of those CR2032 battery holders so I'll chop the NiCd off when they arrive. I assume the motherboard will have been designed to not try and charge something plugged into the battery header but I guess I can confirm that with my multimeter when the NiCd is gone.

So I think I probably had three problems at various times while trying to get this thing up and running:

  • Probably a faulty VGA card.
  • Based on the ratio of FPM to EDO modules in my box of RAM I was probably trying to use EDO memory at times.
  • The one VLB card I had handy to test with was configured from the factory for a slower bus speed. I guess in the past when I was messing around with VLB cards they were already correctly configured for the sorts of machines I was using them in so never had any reason to change the jumpers. It didn't help that the manual for this card is not very clear at all on this particular setting - I've still got no idea what drive capacity has to do with bus speeds.

Tomorrow I'll have another search of my store room some more/better VLB cards - an I/O card with a less annoying option ROM and a Cirrus Logic GD5429 if I can find it. I've located two boxes of expansion cards so far but they mostly contain NICs, Sound cards, Modems and SCSI controllers. My database says I've got a bunch of other VLB cards so there must be another box of cards I've not discovered yet.