VOGONS


Reply 20 of 38, by meljor

User metadata
Rank Oldbie
Rank
Oldbie

Even when it works in another computer, make sure all the legs on the 3dfx chips are straight. i have had cards where these touched or almost touched and it gave some very strange behaviour and they sometimes worked, and sometimes were missing in hardware. I saw this on v1 and v2 cards.

Also clean the contacts, you never know. On an open testbench i also had various cards causing problems when not kept straight up in the slot (mostly voodoo3), so make sure they make good contact.

Good luck.

asus tx97-e, 233mmx, voodoo1, s3 virge ,sb16
asus p5a, k6-3+ @ 550mhz, voodoo2 12mb sli, gf2 gts, awe32
asus p3b-f, p3-700, voodoo3 3500TV agp, awe64
asus tusl2-c, p3-S 1,4ghz, voodoo5 5500, live!
asus a7n8x DL, barton cpu, 6800ultra, Voodoo3 pci, audigy1

Reply 22 of 38, by feipoa

User metadata
Rank l33t++
Rank
l33t++

For the RTC CMOS clear, I have tried RESET' and RCLR'. RESET' does nothing (powered or otherwise). RCLR' clears only your CMOS settings, not your previously saved ESCD (ISA/PCI card config. values).

I have been trying to figure where exactly those ESCD values are physically saved to on this motherboard. Are they saved in SRAM as part of the chipset? Are they saved in the EEPROM FLASH chip? I thought they were saved in the RTC module, however I cannot understand why a RTC reset does not clear this information. The Dallas 12887A RTC contains 114 bytes of battery-backed SRAM, the 12887 does not have this battery-backed SRAM, however both RTC modules are able to save ESCD data somewhere. Where?

It could be a contact issue with the card, however I've already cleaned the contacts with ISO. I could also try the eraser trick.

Windows does see the card in Device Manager. Keep that in mind. Perhaps just one or a few pins are not making contact with the PCI slot?

After another swap around, I did get the Voodoo2 card working with a dozen resets and power cycles in, both, Win95c and NT4. Then I left the system overnight, touched nothing, powered up the next day, and Windows I get the "Voodoo expected, none found" error. Shuffling the PCI cards, and the error goes away. It really seems like a BIOS configuration issue to me. I wonder if I use a dead ODIN RTC, will the ESCD get saved? EDIT: Even with a battery-depleated ODIN RTC, the ESCD information is being saved, but where?

Plan your life wisely, you'll be dead before you know it.

Reply 23 of 38, by h-a-l-9000

User metadata
Rank DOSBox Author
Rank
DOSBox Author

ESCD is saved i.e. in the BIOS chip. Some defective boards say 'unknown flash type' during the boot process, at the point where they would like to update the ESCD. I did wonder about the obviously incorrect Wikipedia article, as the RTC lacks room for 2–4 KB data as they say there.

Your problem could be bad solder joints on the card. When you move the cards around the joint gets fixed temporarily. If this is the case you can often easily crash the system by twisting the card a bit while it is in use.

To exclude any software issue you could use a DOS PCI ID tool which reads the current IDs directly from hardware.

1+1=10

Reply 24 of 38, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Is it that non-EEPROM BIOS chips, such as UV EPROM and OTP ROM also have 2-4K of FLASH memory? Is there a way to disable this FLASH RAM? Or a way to clear the FLASH RAM at boot?

I still beleive the issue is related to the ESCD and PCI configuration data. I have noticed that I can never get the Voodoo2 card working when SCSI is on IRQ 11 and 2D VGA is on IRQ 9, for example. SCSI on IRQ 9 and 2D VGA on IRQ 11 has the best results, however even this requires shuffling the PCI cards.

EDIT: The reason I do not really think the issue is related to a loose contact point or joint on the Voodoo2 card is because removing and reinserting the card to the same PCI slot does not fix the problem. I need to put the Voodoo2 card into a different PCI slot, whereby a the ESCD gets rewritten. Perhaps my SST PH29EE010 FLASH chip has faulty CMOS memory where the ESCD resides?

Plan your life wisely, you'll be dead before you know it.

Reply 25 of 38, by feipoa

User metadata
Rank l33t++
Rank
l33t++

It does seem that the ESCD is stored in the EEPROM FLASH BIOS chip. When I remove the chip, reprogram it in my external programmer, and reinsert the chip into the motherboard, I get an ESCD update. If I remove the same EEPROM chip, reinsert it, I do not get an ESCD update. If I reprogram that same BIOS chip again, I get an ESCD update (and the Voodoo2 works as expected).

The SST PH29EE010 datasheet mentions, "Write Inhibit Mode: Forcing OE# low, CE# high, or WE# high will inhibit the write operation. This prevents inadvertent writes during power-up or power-down." Does this mean that if I reprogram the EEPROM chip, then tie WE# to Vcc, that the system will not be able to write the ESCD data? And if so, leaving this work up to the OS?

Or how about using an Intel P28F001 EEPROM, which is 12 V programmeable instead of 5 V programmeable and leave the 5 V jumper enabled. This way, the system cannot write the ESCD data into the boot block. EDIT: Interesting, there is also a "Disable NVRAM / ESCD Update" option in Win95c Device Manager under PNP BIOS. Perhaps this is all I need...

EDIT: No go. Actually, all one need to do is remove the 12v/5v BIOS programming jumper to invoke the inability to write ESCD to the BIOS. This did not solve my problem, though. There is more going on here than meets the eye. Trying different MB-8433 motherboards did not solve the problem.

Plan your life wisely, you'll be dead before you know it.

Reply 26 of 38, by feipoa

User metadata
Rank l33t++
Rank
l33t++

The problem has been solved. I ordered a new old stock STB Voodoo2 from Russia and it works just as the old one did when it was new. I'm not sure what happened to the old Voodoo2 during the case swap which caused intermittant operation that would resolve itself after 5+ minutes of uptime. There could be a solder point which is making intermittant contact to an inner layer trace, or some transistors in the actual ICs which are going bad. Is this a common problem with these Voodoo2 cards?

Before I got the NOS STB Voodoo2 card, I had ordered a Diamond Voodoo2 card from eBay. That card didn't even have intermittant working contitions - 3D was dead entirely and I tested it on several motherboards and OSes. Device Manager did, however, show a Voodoo2 device.

Plan your life wisely, you'll be dead before you know it.

Reply 27 of 38, by JayCeeBee64

User metadata
Rank Retired
Rank
Retired

That's good to hear. I've never seen a Voodoo 2 that only works when it wants to; perhaps a more detailed, cursory examination is needed.

Unfortunately Voodoo 2 cards do have a tendency to fail quite suddenly and without warning, even if you take the utmost precautions when storing and/or handling them. Their size makes them rather unwieldy and difficult to handle (I've lost count of how many times I bumped my Voodoo 2s when placing them in a PCI slot). Excess heat is also an issue. There are many VOGONS members that have posted about their Voodoo 2 cards not working properly (or at all) after some time. And as you found out, getting one from eBay can be a crap shoot.

At least you got this issue solved now, but if I were you I would get another Voodoo 2 as a spare; who knows how long the one you currently have will work flawlessly.

Ooohh, the pain......

Reply 28 of 38, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Prior to swapping cases, I had been running the old Voodoo2 pretty hard with GLQuake running in autoplayback/loop mode for several hours over several days. I was trying to qualify my modified IBM 5x86c-100 QFP chips at 133 MHz and 3.75 V. But this CPU is so under powered for for the Voodoo2, I didn't think it would be a problem. Maybe it was, maybe it wasn't?

Plan your life wisely, you'll be dead before you know it.

Reply 29 of 38, by JayCeeBee64

User metadata
Rank Retired
Rank
Retired

Sounds like you overclocked the IBM CPUs. It' usually not an issue unless the PCI bus was also running out of spec (higher than 33MHz); if this is the case then your old Voodoo 2 could have sustained some damage. Still, 486-class processors will not push a Voodoo 2 hard at all - when I had my 3dfx Voodoo 2 1000 in my Socket 7 PC (Pentium 166MMX) it barely got lukewarm even after playing several Glide/OpenGL games for quite a while (the CPU cooler was actually much hotter). It takes a Pentium II or III to get a Voodoo 2 very warm (and even really hot) if running a game like GLQuake for an hour or more.

I still believe that examining your old Voodoo 2 with a magnifying glass will probably give some idea of what kind of problem/issue it has; hopefully the damage is not internal - that will make it much harder to figure out.

Ooohh, the pain......

Reply 30 of 38, by feipoa

User metadata
Rank l33t++
Rank
l33t++

The PCI bus was run at 33.3 MHz the entire time.
I did spend about 10 minutes in the past looking for loose solder joints, but came up empty.
As the old Voodoo2 seems to function fine in other boards, I might use it elsewhere, like in my 430TX board w/GF4.

Plan your life wisely, you'll be dead before you know it.

Reply 31 of 38, by Dustinslab

User metadata
Rank Newbie
Rank
Newbie

Hi, I'm building a retro gaming PC and I ran into the same problem as described in this post. After first installation of Windows 98 my Voodoo 2 SLI setup was working fine. After testing a few games and restarting it a bunch of times, the cards weren't visible in Dxdiag (Display 2 tab is gone). They are however still visible in the device manager. The display properties tab "V2-1000" will also stop functioning. Games behave like there is no 3D card available.

I tried a bunch of FastVoodoo versions, I reinstalled windows. I tried Windows ME. Same thing. After a few boots, DirectX suddenly forgets about the cards. When you reinstall the drivers, sometimes they get detected, sometimes they don't. Reseating the cards in a different slot usually resolves the issue, but It keeps happening. I am absolutely clueless about what is going in. In a different system, the cards work fine. Since Voodoo cards don't use IRQ there is nothing to set in BIOS. I tried both PnP aware OS Off / On.

I think it has something to do with DirectX, but I have no clue on how to troubleshoot this.

Does anyone know how and where Win9x defines and stores the graphics cards (registry/ini/files) ?

System is :
PCCHIPS M560 w/ P233MMX 512KB Cache
80MB EDO RAM
SB AWE32
Trident 9385
VOODOO 2 SLI

Grtz Dustin

Reply 32 of 38, by Dustinslab

User metadata
Rank Newbie
Rank
Newbie

Hi, I think I may be on to something here. I suspect the Bios-battery is failing on my board. Yesterday, I reseated the boards in the machine (cleaned the pci contacts) and all was good the entire day. A few reboots, a few shutdowns, no issue whatsoever. When I came back this morning, I booted and the problem was back. 😒

Now, the PCChips M560 does not have a normal battery. Check this out :

m560p.jpg

I'll attempt the repair in the video below... Hope this is what is causing my problems...
https://www.youtube.com/watch?v=NdlSfqto_0o

Grtz Dustin

Reply 33 of 38, by Dustinslab

User metadata
Rank Newbie
Rank
Newbie

Well, I will bump up this old topic one last time, as I have discovered the reason for the problem. The issue results from a bad connection between the pci-slot and the boards themselves.

The case I used apparently places the Voodoo 2 boards at a slight angle (almost invisible) in the PCI-slot which caused the problem. None of the other boards in the system have this issue, only the voodoo cards for some reason.

At a certain point I could open dxdiag without pressure on the back of the boards and not see the "display 2" tab. A minute later I opened dxdiag again with pressure on the boards and voila. The display 2 tab was there. I repeated this a few times and the results were conclusive.

I will clean the pci-slots and try to fix the issue with the angle of the boards.

I hope this information may someday help someone else!

Grtz Dustin

Reply 34 of 38, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Your conclusion is interesting. I had tried cleaning the Voodoo2 card's contacts already but the problem continued on. I bought another NIB of the same card and I've used it countless times without issue. I felt the issue was probably due to a marginal contact of an IC on the PCB of the Voodoo2, but I have no proof of this.

Plan your life wisely, you'll be dead before you know it.

Reply 35 of 38, by kixs

User metadata
Rank l33t
Rank
l33t

This got me interested...

A while ago I tested two Creative V2 12MB and both displayed the same error - expected Voodoo^2, none detected - when starting Q3A Demo. But when I inserted Diamond V2 12MB it worked right away. Both cards were detected in Device list.

Tested in another computer. One Creative card worked fine, while the other had the same error. I did notice the cards don't seem to seat very good in the PCI slot. I will try the non-working card again with some force applied.

The usual checks were already done (cleaned the contacts, check for bent pins, missing caps...).

Requests are also possible... /msg kixs

Reply 36 of 38, by feipoa

User metadata
Rank l33t++
Rank
l33t++

I'm skeptical still. In my case, I had to clear force a clear of ESCD data for the Voodoo2 to get detected. I never touched the card in between.

Plan your life wisely, you'll be dead before you know it.

Reply 37 of 38, by doaks80

User metadata
Rank Member
Rank
Member

Put the voodoo5 out of sight, the system is obviously getting distracted.

k6-3+ 400 / s3 virge DX+voodoo1 / awe32(32mb)
via c3 866 / s3 savage4+voodoo2 sli / audigy1+awe64(8mb)
athlon xp 3200+ / voodoo5 5500 / diamond mx300
pentium4 3400 / geforce fx5950U / audigy2 ZS
core2duo E8500 / radeon HD5850 / x-fi titanium

Reply 38 of 38, by Dustinslab

User metadata
Rank Newbie
Rank
Newbie
feipoa wrote:

Your conclusion is interesting. I had tried cleaning the Voodoo2 card's contacts already but the problem continued on. I bought another NIB of the same card and I've used it countless times without issue. I felt the issue was probably due to a marginal contact of an IC on the PCB of the Voodoo2, but I have no proof of this.

Well, cleaning the cards contacts didn't help me either. Cleaning the PCI-slots themselves (with some contact cleaner and carefull insertion of stick and cloth) and making sure the cards are in straight seems to have solved my problem.

I too would sometimes get the "display 2" tab back without touching the cards. I suspect heat/cold also move the cards in the slots?

In any case, i'm happy this project is almost done now 😀