Reflowed TMUs on a Voodoo 2 - with pretty positive results Re: Voodoo2 component map

smtkr wrote on 2025-01-08, 03:16:

Which drivers are you running. I have a similar setup and, while Deus Ex doesn't run great in outdoor environments, I definitely […]
Show full quote

Which drivers are you running. I have a similar setup and, while Deus Ex doesn't run great in outdoor environments, I definitely get 30-45 FPS in the first level.

I'm running
1GHz Pentium 3 (SL4KL)
512MB PC133
Geforce 2 GTS

Windows XP SP1

version 5.32 (June 2000) under Windows 98 with 256Mb RAM. It's possible that something is wrong with the card, the motherboard or the config. 3DMark 2000 score is 6430, which I think is perhaps a little low. But certainly not half what it should be.

Tried to survive this PODP5V83 CPU, unfortunately was unable to test it right after the purchase and found that it didn't work too late 🙁
Motherboard didn't turn on with this CPU, power supply immediately went to the short circuit protection mode and I can hear the high pitch tone.
Decided to de-solder capacitors and voltage regulator from the CPU and try to check them separately. No luck 🙁 capacitors looks good, do not know how to test voltage regulator, but even without them motherboard didn't run.

Is there a short across any of the pads where there shouldn't be?

Kahenraz wrote on 2025-01-08, 22:25:

Is there a short across any of the pads where there shouldn't be?

No, all ground pads, marked as 1, 3, 6, 9, 11 are connected together, it is strange that pad 5 and 7 are not connected together as they both are should be 3.3 output if I am not mistaken
Here is the resistance between 1 and 2 - 365 k ohms; 3 and 4 - 3.12 M ohms ; 9 and 10 - 367 k ohms; 11 and 12 - 365 k ohms

Created a custom "slotket" mounting solution that holds it rock solid

Made from 3d printed parts and some solid copper bar that I drilled and tapped at either end on my friends lathe, thought the copper made sense because the CPU is a "coppermine" haha!

3d printed parts probably not so retro looking but I like it!

Eked out a space in by boxroom to sit a laptop/tvtray type table to set up my Sharp PC-4641 on to get my Turbo XT fix, coz it's a bit heavy to use much on the lap and the power plug is easily snagged out. Ran LM60 on it and woohoo, told me it was as fast as a 7mhz AT, speed demon, and got a massive score of 10 in TOPbench. Though sometime previous I tried it, I seemed to be seeing lower scores and some things were detecting an 8087... which is meant to mean system clock can only be 8Mhz with that enabled. Saw nothing on the setup screen (unusual for an XT but controls screen and other stuff) about FPU.. IDK if I knocked or reset the switches on the bottom at some point. I can't recall seeing an 8087 inside it and maybe had it misconfigured, or maybe didn't. Literally while my back was turned the other machine in the room decided to die (see modern) otherwise I would have had a relaxing evening just futzing around in DOS, getting the harddrive more "civilised".

In Voodoo3 2000 PCI I replaced EZ1585 with APW7120+2SK3918. Now this unit does not heat up.

Postman5 wrote on 2025-01-09, 06:07:

In Voodoo3 2000 PCI I replaced EZ1585 with APW7120+2SK3918. Now this unit does not heat up.

Beautiful mod!

Man, I wish there were drop-in replacements that ran cooler without having to make a custom PCB with multiple components.

Susanin79 wrote on 2025-01-08, 23:10:

Kahenraz wrote on 2025-01-08, 22:25:

Is there a short across any of the pads where there shouldn't be?

No, all ground pads, marked as 1, 3, 6, 9, 11 are connected together, it is strange that pad 5 and 7 are not connected together as they both are should be 3.3 output if I am not mistaken
Here is the resistance between 1 and 2 - 365 k ohms; 3 and 4 - 3.12 M ohms ; 9 and 10 - 367 k ohms; 11 and 12 - 365 k ohms

The larger pad of voltage regulators (pad 5 in your pic) is mostly intended for heat sinking purposes. It is connected to the output pin (pad 7) internally, so no urgent need to connect them externally.
Resistances look ok to me. No shorts that would prevent the board from booting.
Does your board still boot with a 486? Does is "start" without any CPU?
Can you measure resistance in the 5V rail with / without the POD in the board?
Maybe some other chip on the board is gone...

Postman5 wrote on 2025-01-09, 06:07:

In Voodoo3 2000 PCI I replaced EZ1585 with APW7120+2SK3918. Now this unit does not heat up.

Nice!
Did something similar with a Riva TNT:
Switching regulator module as replacement for hot linear regulators (mainly used on PCI VGA cards)

Still haven't seen any good drop-in replacements for not having to do custom PCBs (but also haven't looked around for a while).
To know what current you actually need is the crucial point.

ChrisK wrote on 2025-01-09, 10:53:

The larger pad of voltage regulators (pad 5 in your pic) is mostly intended for heat sinking purposes. It is connected to the ou […]
Show full quote

The larger pad of voltage regulators (pad 5 in your pic) is mostly intended for heat sinking purposes. It is connected to the output pin (pad 7) internally, so no urgent need to connect them externally.
Resistances look ok to me. No shorts that would prevent the board from booting.
Does your board still boot with a 486? Does is "start" without any CPU?
Can you measure resistance in the 5V rail with / without the POD in the board?
Maybe some other chip on the board is gone...

Yes, this board works fine with the normal DX4 CPU, It also can power on without the CPU.
I did the measurement with and without the CPU installed; the resistance between 5v rail and the ground on molex connector is 87 ohms, and it shorted when CPU is installed.
Did the measurement on my second not working 486 motherboard and there is no shorts, resistance 80 ohms with and without the CPU. MB can power on. Unfortunately I need to fix that board first to boot it normally. But it is already some progress.
Will inspect first motherboard more carefully, as I already found that some pins of cache sockets were badly damaged, may be there is something else been damage.
Thank you for you advice, may be this CPU is not dead

I think this CPU is supposed to have an active cooler on it. Maybe the fan died and it overheated. Or it was used without the cooler at some point.

Kahenraz wrote on 2025-01-09, 20:36:

I think this CPU is supposed to have an active cooler on it. Maybe the fan died and it overheated. Or it was used without the cooler at some point.

CPU came with the heat sink and cooler, I just remove them to have a better access to the mounted elements. Unfortunately I have only 3 486 motherboards, one of them didn’t work and the second old EISA without overdrive socket. So will continue to search what may cause this short circuit on Jetway J-437 as now it is the main suspect.

Against my better judgement, I tried putting a VIA-based USB 2.0 card into a Celeron system with the 440ZX chipset. Predictably, both WinME and Win2000 froze during startup. Tried all PCI slots, didn't make a difference.

Then, I took out the USB 2.0 card and put in a 3Com 3C905B-TX 10/100 network card. It was instantly recognized by both operating systems and drivers were installed automatically with no conflicts whatsoever. I did have to manually enable 100Base-T speed and Full Duplex, but that was like two clicks. Getting a respectable 8 MB/s transfer speed with a Coppermine Celeron 600, which perfectly suits my needs for that rig.

I replaced the fan in my DTK PTP-153 power supply. It wasn't worn out, just noisy. Talking about service engineer friendly:

PD2JK wrote on 2025-01-10, 14:38:

I replaced the fan in my DTK PTP-153 power supply. It wasn't worn out, just noisy. Talking about service engineer friendly:

The attachment DSC_2897.JPG is no longer available

You can also use a buck converter to lower the voltage.

I will often clip the wires on these fans in power supplies, if they're soldered in, and add my own resistors to adjust the fan speed.

Kahenraz wrote on 2025-01-10, 18:01:

I will often clip the wires on these fans in power supplies, if they're soldered in, and add my own resistors to adjust the fan speed.

Interesting solution. What size\type of resistor do you use and where do you install it exactly? I've never just used a DIY resistor to slow a fan down before, but I've always been curious about doing so.

Old power supply fans are definitely prime candidates for such a thing though. Most of the fans seem to be extremely long-lived and decent quality but they are just so excessively loud and annoying.

I just resurrected a motherboard, QDI Advance 10T, with Tualatin CPU support, currently to see if it works, I used a Celeron Coppermine 766 MHz, after changing the RAM, it started, for the record I had changed all nine capacitors 1000uF 6.3V, some of these were swollen, but changing them all, at least these will be fine for quite a while.