check the reliability of pins soldering of the TMU0 - U9

Thanks !
TMU0 is the right chip isn't it?

bimole wrote on 2024-10-29, 05:02:

TMU0 is the right chip isn't it?

Yes

Sometimes it works, sometimes it doesn't - I run under DOS the mojo program with one hand, and with the other I press the chips on the board and press a little, deform the board and watch the output of the mojo.

OK some news...

I have reflown all TMUs leads and checked the electrical continuity of every single one.
All pins are OK.

I thought at a moment it was solved because mojo worked on first try and I could access to the Voodoo 2 display properties.... But once again, I had the ''glide2x.dll'' error message.
It seems that the error message appears less than before, but it was a quick test.
3DFX ICs seem in good shape. I've also checked some resistors and the ferrite beads. Nothing special but I think I will check and reflow every resistors. Far less cumbersome than IC pins 😁

One last thing, the voltage regulator... I don't know how tend to age tantalum caps but I wouldn't be surprised to measure a messy 3.3V. Linear regs have sometimes the bad habit to oscillate at high frequency depending on the load (but it is more likely with low ESR capacitors).
Any feedback on this voltage reg ageing ?

I would also like to check the board with the ''Witchery'' diagnostic software. Any experience?

Cheers,
JB

I've performed some voltage and resistance/continuity measurements:
3v3 sounds good : 3.43V
as well as 5V : 5.09V
With the oscilloscope, voltages are clean, no spike nor noise.

I also checked the signal accross Y1, DAC crystal. It oscillates at the right frequency.

Something is weird regarding resistance between 5V and GND.
I get 3.3kohms with my ohmmeter (5V pin on "+" ohmmeter, GND flange on "-" ohmmeter) and only 1.2kohms if I reverse ohmmeter "+" and "-" probes.
In diode mode, it displays 0.38V (5V on +, GND on -) and 0.28V with reversed probes.
I measured the same pins on other PCI cards (sound cards, SATA cards, USB cards), I always measure >100k ohms values. No such low values as on the Voodoo 2.

I also measured resistance between 3V3 and GND (across the big 22µF 16V tantalum cap C39)
I measure 306 ohms (3V3 on +, GND on -) and the same value if probes are reversed.
R99 and R100 are the adjusting voltage resistors for the linear regulator. Both are in series and this resistor association appear in parallel with the 3V3 bus. Since they are 110R and 196R respectively, it is normal to measure 306 ohms across the 3V3 bus.

I get lost...
Could it be a defective RAM chip ? I will try to reflow RAM chips next week...

EDIT :
Since I have sometimes error messages regarding TMU reset :

1sst1InitResetTmus(): Could not reset TMUs...

I checked the pin 208 ot both TMUs.
Continuity between the two TMU 208 pins is OK. They are both connected to the pin 197 of the FBI.

Cheers,
JB

I have several Voodoo2, all of them show approximately the same readings in the diode test mode.
(5V on +, GND on -) 0.308...0.392
(5V on -, GND on +) 0.196...0.238
But this method seems dubious to me. It is better to check the FBI connections with TMUO, and check the resistors RA3...RA18. Оn CT6670, they are usually blue, rated at 22 Ohms

That's fine!
Finally tested the Voodoo2 on another platform:
Super socket 7 K6-2 450
Asus P5A rev 1.04
128MB RAM
Clean Windows 98SE

It works!
All testxx.exe are passing successfully, mojo ok, display properties ok...
I don't understand what is happening with my other platform (see above).
Anyway, the SS7 platform was the target system for this Voodoo2 so... that's fine!

maybe the PCI slot contacts are not ok on that board

Likely caused by the Creative drivers for the Voodoo2 which is based on an older version that have speed sensitivity issues on faster CPUs. I would suggest using the reference driver 3.03.00 (non-beta, dated 18 Jun 1999) instead.

Even with these drivers, MOJO won't properly on that Pentium III CPU unless you can find a way to slow it down.

Very interesting 👍 thank you for the trick!
I will give a try with these drivers.

I would like to add that the Voodoo2 cards and the ATI Radeon 9500/9600/9700/9800 cards, had major compatibility issues.

Just do a Google search and you will find (even on Vogons) various experiences in this regard.

Voodoo 2 Crashing Issue

I own two Voodoo 2 cards and wanted to set them up in SLI mode. However, while one card works perfectly, the other only operates for a few seconds before failing. I tested both cards in two different computers, and as mentioned, one card works fine.

For reference, these are I-O DATA Voodoo 2 cards, which have numerous electrolytic capacitors: 11 x 100uF and 6 x 47uF.
Steps I have already taken:

Cleaned the edge connector.
Replaced most of the electrolytic capacitors (almost all except for 6 x 100uF capacitors, which I am waiting to replace when my order arrives).
Inspected all components for loose pins, cold solder joints, or missing components — everything appears fine.
Resoldered all memory chips and the RAMDAC.

Current Status:

Unfortunately, these steps haven't resolved the issue. The card works in Glide mode for a few seconds to a minute, sometimes a bit longer, sometimes shorter. When it works, the image is perfect, and the card is correctly detected in Windows. However, when I launch a game like GLQuake, it crashes after a few seconds or minutes.
Request for Suggestions:

Do you have any ideas on what else I could try?
Would Voodoo Witchery provide any insights? Unfortunately, I don’t have access to it.

Any advice would be greatly appreciated!

@Krzysztof Zakrzewski
When you say that it crashes after a few seconds/a minute, does the Glide application just freeze, or it actually crashes with an error message?

If it just freezes after a time, while rendering correctly, it could be because the card has a layout a bit different than the reference design one, and might not work properly with regular drivers, at least without setting some registry keys/env. variables from autoexec.bat.
Can you try under Windows with the official driver?

https://www.iodata.jp/lib/software/g/156.htm

A lead worth checking.
I have two identical Voodoo 2 I-O Data cards, so if one works, why shouldn't the other? But it's worth ruling it out!

sdz wrote on 2025-01-02, 15:17:

@Krzysztof Zakrzewski When you say that it crashes after a few seconds/a minute, does the Glide application just freeze, or it a […]
Show full quote

@Krzysztof Zakrzewski
When you say that it crashes after a few seconds/a minute, does the Glide application just freeze, or it actually crashes with an error message?

If it just freezes after a time, while rendering correctly, it could be because the card has a layout a bit different than the reference design one, and might not work properly with regular drivers, at least without setting some registry keys/env. variables from autoexec.bat.
Can you try under Windows with the official driver?

https://www.iodata.jp/lib/software/g/156.htm

I attempted to install the I-O DATA drivers from the provided website.
To start, I used a functioning card to ensure the process.
Unfortunately, at the very end of the installation, an error occurred, and the card stopped working.
The issue might be related to the Japanese language; I am not certain.
The hardware is visible in the Device Manager but shows an error.

Krzysztof Zakrzewski wrote on 2025-01-02, 18:14:

I attempted to install the I-O DATA drivers from the provided website. To start, I used a functioning card to ensure the process […]
Show full quote

sdz wrote on 2025-01-02, 15:17:

@Krzysztof Zakrzewski When you say that it crashes after a few seconds/a minute, does the Glide application just freeze, or it a […]
Show full quote

@Krzysztof Zakrzewski
When you say that it crashes after a few seconds/a minute, does the Glide application just freeze, or it actually crashes with an error message?

If it just freezes after a time, while rendering correctly, it could be because the card has a layout a bit different than the reference design one, and might not work properly with regular drivers, at least without setting some registry keys/env. variables from autoexec.bat.
Can you try under Windows with the official driver?

https://www.iodata.jp/lib/software/g/156.htm

I attempted to install the I-O DATA drivers from the provided website.
To start, I used a functioning card to ensure the process.
Unfortunately, at the very end of the installation, an error occurred, and the card stopped working.
The issue might be related to the Japanese language; I am not certain.
The hardware is visible in the Device Manager but shows an error.

When I reinstall without specifying files and only selecting drivers from the list, the installation succeeds.

After the error, the hardware remains disabled. However, it can be re-enabled without any errors.
But when I try to run a game using Glide, I get an error stating that the glide2x.ovl files are missing.
If I copy the files from the drivers to the game folder (e.g., Quake), I receive an error message saying that these are invalid glide2x.dll files when attempting to launch the game.

In the display options, there are no settings related to Voodoo 2.

It’s possible that the issue is due to these drivers being intended for the NEC PC-98 computer:
https://www.iodata.jp/lib/product/g/220.htm
and they may not be compatible.
3DFX drivers for NEC PC-98

In any case, identical cards work properly and stably with reference drivers and Fast Voodoo drivers, so I suspect a hardware issue with the second card.