Yes, that's the most probable case scenario. I'll give it a shot when I have enough time. Windows did detect some part of the card. The drivers just didn't detect it.
I replaced all the things that seemed wrong in the Voodoo 1 but nothing. Windows doesn't detect it. And the top 3DFX labelled chip gets hot in idle. I think this is another dead card.
What is wrong with Argentina and pre 1999 good cards? 😕
wrote:Yes, that's the most probable case scenario. I'll give it a shot when I have enough time. Windows did detect some part of the card. The drivers just didn't detect it.
I replaced all the things that seemed wrong in the Voodoo 1 but nothing. Windows doesn't detect it. And the top 3DFX labelled chip gets hot in idle. I think this is another dead card.What is wrong with Argentina and pre 1999 good cards? 😕
It's all about rough handling. I undestand what you feel, when I bought my first 486 board, a giant ISA-only thing with Weitek socket and clock set by replacing Quartz(!), it had the very same chip damage. I was so full of hopes, yet there was nothing I could do. Two years later I stripped the board clean and threw it away.
Yeah, I know. Most people doesn't care about old things. So this things happens.
But there is a sunshine far far away. I desoldered the warped capacitor from the Voodoo 1 and measured it with an RLC. It measures 8uF and 56,5º of phase. In other words: it's not acting like a pure capacitor. At R is measuring the pretty high amount of 13 ohms 😵
I will replace all these electrolytics and give it a shot. There is a little tiny chance that the chip isn't fried yet.
wrote:Nothing helped. I tried clearing CMOS taking the battery off, I tried every bios setting, I tried changing the psu to see if tha […]
Nothing helped. I tried clearing CMOS taking the battery off, I tried every bios setting, I tried changing the psu to see if that helps and nothing. I tried it in the three PCI slots too. It does work ok with my Trident 9680, but if I put the TNT2 and the Trident at the same time with the CRT connected to the Trident, the system wont display anything too. It boots, I hear the hdd loading windows but no display. I also found that the power_good signal cable from my original AT psu was burning. I tried again with the other AT psu that I have for spare and everything is fine. It seems that this system is starting to show aging issues. But every stress test I ran were ok. So I don't know why the hell this isn't working. The TNT2 is working like a charm in my Deskpro EN as I said.
Regarding the Voodoo 2:
I can put those pins in place. But there is a missing pin and I have no room in the chip to place a wire. One solution could be drilling something like 0.5 - 1 milimeter the chip to make place and insert a thin pin and join it with what remains of the original pin. Another solution could be replacing the entire chip, but It's 99.99% unlikely finding a spare chip or another blown card at decent price in here.
And the Voodoo 1 still needs replacing some caps. I'm not possitive at all, but besides the issues with the TNT2 and my retro rig, I bought a bunch of working stuff at really low price. Sure, It will be a pity if I can't fix them. But they were something like a gift so I can't complaing too much about it.
Ugh, that Voodoo 2 damage is bad. Really bad. Looks like it got hit very hard. Have to agree with RacoonRider, it's a goner 😵
And yes rough handling is the cause in most cases, it happens even here in the US. Here's a dead Monster 3D I found at a garage sale a while back:
Two of the pins have been ripped from the chip; the result is a Voodoo 1 that won't allow a PC to even boot. I got it for free and keep it around as a conversation piece.
The behavior of the TNT2 is really puzzling, can't think of anything else to try. I'll let someone else take over while I do some inquiries and research. And hang in there - at least the Trident still works.
Ooohh, the pain......
That chip could be repaired by someone with a sufficient skill level.
It would not be easy even for them and the prognosis isn't great.
I have pulled off such feats before but my eyesight isn't that good anymore.
Replacing the whole chip would be easier if you can find one.
.
GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.
The damage to the V2 is hard to repair, but personally I'd do the same thing as JayCeeBee64. Perhaps you could nail it to a wall or something.
But I'd definitely not throw it out, if only for the RAM chips. Sometimes dead hardware can still be used for parts, but the dead board has to have some value for that and I think a Voodoo 2 definitely has that kinda value.
Tip: Before I start testing 'new' hardware, I always always do a thorough visual check-up of the board for any signs of damage. Not only will this prevent frustration with the damaged board, but if you decide to plug it into a otherwise working system to test the damaged board, you run the risk of damaging the other known working hardware as well. I've done this several times, killing known good hardware by attaching an unknown piece of hardware to it and then powering it up and that was it! Dead 😢
I always try to visually inspect a part before I buy it, but right before I try out an untested/'new' part for the very first time, I inspect it for any kind of damage again because despite how careful one may be, it can happen that a part gets damaged after it became part of your collection, be it rough handling, battery damage, other damage or inadvertent rough handling of your own doing (could've dropped the box it was in and never realized the part inside got damaged by the impact of the fall).
I have some Diamond PCI sound card with a chip that's damaged this was. Also some AMD 5x86 (the variant that's mounted on a small PCB) and one time I got an Audigy 2 I think? It was only after I had bought it that some tiny part very close to the PCI pins had burned pretty badly. This just kinda comes with the territory.
All the wires are still there. I would try straighten them into proper place by surgical needle work and then soldering or conductive glue.
wrote:Nothing helped. I tried clearing CMOS taking the battery off, I tried every bios setting, I tried changing the psu to see if tha […]
Nothing helped. I tried clearing CMOS taking the battery off, I tried every bios setting, I tried changing the psu to see if that helps and nothing. I tried it in the three PCI slots too. It does work ok with my Trident 9680, but if I put the TNT2 and the Trident at the same time with the CRT connected to the Trident, the system wont display anything too. It boots, I hear the hdd loading windows but no display. I also found that the power_good signal cable from my original AT psu was burning. I tried again with the other AT psu that I have for spare and everything is fine. It seems that this system is starting to show aging issues. But every stress test I ran were ok. So I don't know why the hell this isn't working. The TNT2 is working like a charm in my Deskpro EN as I said.
Regarding the Voodoo 2:
I can put those pins in place. But there is a missing pin and I have no room in the chip to place a wire. One solution could be drilling something like 0.5 - 1 milimeter the chip to make place and insert a thin pin and join it with what remains of the original pin. Another solution could be replacing the entire chip, but It's 99.99% unlikely finding a spare chip or another blown card at decent price in here.
And the Voodoo 1 still needs replacing some caps. I'm not possitive at all, but besides the issues with the TNT2 and my retro rig, I bought a bunch of working stuff at really low price. Sure, It will be a pity if I can't fix them. But they were something like a gift so I can't complaing too much about it.
Are you absolutely sure a pin is gone? Doesn't look like it on the photo. [Edit: forget that, on a closer look, it does indeed] Also, the pins are broken close to the PCB, so I'm positive that be fixed pretty easily by someone with some soldering experience. Straighten and position the pins as close to how they used to be as you can manage, it's important that the broken ends are as close to each other as absolutely possible. Apply liquid flux generously, and drag solder it using a thin hollow bevel tip. If you manage to bridge some pins, just apply more flux and redo the soldering with less solder on the tip. Be careful to not bend the pins again when soldering, be as light handed as only possible.
If a pin really is gone, some pin breaks while straightening them out, or you just can't get some of them close enough to their corresponding broken ends for the surface tension to do its work while soldering, you can use a thin cable strand as a sort of a wick. Make it a few centimeters long, tape one of its ends to the PCB, position the loose end to bridge the gap between the broken pin parts, apply flux, solder, cut off the excess part of the strand.
Afterwards, be sure to wash away the flux residues carefully, even if you use no-wash flux.
Hope I made any sense, English is not my native language 😀
About the Voodoo 1: The only thing I'm concerned about is the chip getting too hot and Windows isn't detecting the card. But the card looks 100% fine. I'll replace those electrolytics and add the missing caps. There is a little tiny one that hasn't any code. It's a brown little one. How can I be sure about the value of it? I thought measuring another of the same color, as they don't look like electrolytic and there is a good chance that aging doesn't affect them.
About the Voodoo 2: Of course I will try to repair it guys. It does worth trying. It's a nice card and I really wanted to use it on my rig. I decided to drill a 0.5mm hole on the chip (on the top of missing pin) to have space to insert a thin pin and join it with solder to the chip pin. I think it will be the easy way. And that is after realigning the other five or six pins. Let's pray for it. I don't need another broken pin falling of the chip. And yes, I'm 100% sure there is a missing pin because I can see the remaining piece of pin inserted on the chip.
Is any datasheet around the web or something like that for this 3DFX chips? It will be interesting because there are always unused or not connected pins. I know, it's a long shot, but I think it does worth a try. Also I think it will be hiper useful for doing some visual tracing and understanding the circuit. The repairing process is always easier with a datasheet.
PD: kaputnik, don't worry about your english. English isn't my native language too and I'm sure I have an horrendous writing somethimes (or maybe always 🤣). I did understand you perfectly!
wrote:About the Voodoo 1: The only thing I'm concerned about is the chip getting too hot and Windows isn't detecting the card. But the card looks 100% fine.
I wouldn't say it looks 100% fine if it's missing some stuff on the PCB though. It won't even work 😜
Jajaja yes, that's reasonable. The good thing is I desoldered all the tantalum and electrolytic capacitors from the Voodoo 1. I measured all of them with an Agilent RLC. I measured Z at 10 khz and to my surprise, they were all above 12 ohms. The electrolytic ones were below 6.5uF when they has to be close to 10uF (I assume 20% of tolerance). Most of the tantalums are regular. Not too good and not too bad. I will replace them too. I have to do the math, but If I'm not wrong, a 10uF capacitor at 10khz has to be way below than 12 - 16 ohms.
wrote:Is any datasheet around the web or something like that for this 3DFX chips?
Falconfly has some stuff: http://falconfly.de/reference.htm
No pinouts that I can remember though.
"I see a little silhouette-o of a man, Scaramouche, Scaramouche, will you
do the Fandango!" - Queen
Stiletto
Nice stuff! I'll take a good look when I come home. I have brand new caps to solder 😀
And I called again the seller and he has no problem changing me back the tnt2 for other stuff. My girlfriend is cooking chocolate brownies for him jajajaja. He deserves it!
wrote:Jajaja yes, that's reasonable. The good thing is I desoldered all the tantalum and electrolytic capacitors from the Voodoo 1. I measured all of them with an Agilent RLC. I measured Z at 10 khz and to my surprise, they were all above 12 ohms. The electrolytic ones were below 6.5uF when they has to be close to 10uF (I assume 20% of tolerance). Most of the tantalums are regular. Not too good and not too bad. I will replace them too. I have to do the math, but If I'm not wrong, a 10uF capacitor at 10khz has to be way below than 12 - 16 ohms.
The first problem is you are using 10kHz. Caps datasheets usually give ESR at 100 kHz.
If you don't use the same frequency then you can't do a valid comparison to other caps.
(Unless you have the other caps in-hand and also test those at 10 kHz.)
I checked 2 or 3 low ESR (what those -should- be) 10uF 16v caps in datasheets (at 100 kHz) and the ESR was 1.5 or less ohms.
That might not mean anything because you aren't using 100 kHz and ESR at 10 kHz should be higher. - So ???
To me 12 ohms sounds way high and, yes, the uF is out of spec too.
I don't check caps at 10 kHz so I don't have a feel for what it should be at 10 kHz, just -seems- too high.
That doesn't change the obvious problem with low uF.
.
GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.
That is not correct PCBONEZ. Comparing two different capacitors is not the right procedure. Given the frequency and the capacitance you can calculate the impedance (for that frequency). So if you rely on your equipment (as I do, I use an Agilent U1732C) you can easily see if there is something wrong with you caps given the measurements and your calculations. At the end, measuring impedance is the same than measuring capacitance. If the cap goes bad, the impedance and the capacitance will be wrong because they are function of each other. As the capacitance decreases, the impedance goes higher at a given frequency (choose any frequency you want for this purposes, but if the frequency is too high there is a high chance that the change of impedance would be tiny).
Tadaaa! I did a full recaping 😀
I used standard electrolytic capacitors and tantalum. Both are through hole because we are still with stock problems in Argentina due to several years of having our customs closed by government. Getting SMD parts is difficult in here. I had to use silicone to firmly attach some of them because there was several tracks lifted up from the board (not cutted or shorted, just lifted up). So I did an hiper carefully job to not cut those tracks while desoldering and resoldering and everything went fine!
Now I have to wait for tomorrow to test the card. The silicone has to be fully dried to be non conductive. Hopefully the card will show any sign of life. If not, I will be suspecting of some smd little chips that I will try to replace. I read somewhere that if the PCI bus can't see the card, there is a high chance that one of those chips are fried. If that doesn't fix the problem, I will just stop there and keep searching for another cards. Meanwhile I think I will try some S3 cards from this vendor. I will go to his house at the weekend.
wrote:And yes, I'm 100% sure there is a missing pin because I can see the remaining piece of pin inserted on the chip.
I see all of them on the pics you shared here.
wrote:Meanwhile I think I will try some S3 cards from this vendor. I will go to his house at the weekend.
Let it be at least Savage, otherwise even 32 bit SiS305 will do better.
wrote:That is not correct PCBONEZ. Comparing two different capacitors is not the right procedure. Given the frequency and the capacitance you can calculate the impedance (for that frequency). So if you rely on your equipment (as I do, I use an Agilent U1732C) you can easily see if there is something wrong with you caps given the measurements and your calculations. At the end, measuring impedance is the same than measuring capacitance. If the cap goes bad, the impedance and the capacitance will be wrong because they are function of each other. As the capacitance decreases, the impedance goes higher at a given frequency (choose any frequency you want for this purposes, but if the frequency is too high there is a high chance that the change of impedance would be tiny).
Sorry but I'm not the one that's wrong.
You are making a common mistake that I often see even senior Electronics Techs make.
.
You are using the abbreviated 'short' version of the equation that disregards ESR entirely.
That does not work at SMPS frequencies. (Roughly 50kHz-200kHz)
In that range Xc and Xl effectively cancel each other out leaving ESR as the ONLY significant component of Z.
For all intents and purposes in the SMPS frequency range Z = ESR, and capacitance has little or no effect at all.
That is reflected in some brand's datasheets when they head the applicable column with "Impedance (ESR)"
I will provide information from Nichicon showing what I'm talking about in a later post.
Would you like to see?
Use your equation to calculate Z for a 10uF cap at 100 kHz.
Compare your answer to the datasheet values of Z for a variety of 10uF caps at 100 kHz.
0.03 Ohms WCAP-PSLP http://www.mouser.com/ds/2/445/875105359001-540612.pdf
0.045 Ohms A700V106M016ATE045 http://www.mouser.com/ds/2/212/KEM_TC112_ALPOLY-611012.pdf
3.0 Ohms EEEFC1C100R http://www.mouser.com/ds/2/315/ABA0000CE5-68492.pdf
12.0 Ohms EEEHD1C100AR http://www.mouser.com/ds/2/315/ABA0000CE100-40631.pdf
0.03 to 12.0 Ohms. - They vary by between 2 and 3 orders of magnitude with the same uf and frequency.
The equation you are using would result in the same value for all of them.
That is because it's the wrong equation for this application.
.
.
The (understanding) problem comes from the way they teach electronics.
They teach these things with radios in mind and so using radio frequencies were ESR is so small it is insignificant.
- Thus they drop the ESR factor out of the equations.
- The students forget ESR ever came up and that it has anything to do with it.
- And then at SMPS frequencies (where ESR is significant and the largest variable) they forget to put it back into the equation.
.
GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.
Impressive job on recapping! Let us know if your Voodoo works!
Btw, if you decide to throw away your V2, consider keeping the RAM, it can be used in other videocards or onboard graphics:
I've used re-claimed DRAM chips in a number of Trio64 cards before, which almost never have the slots populated.
Check out my blog and YouTube channel for thoughts, articles, system profiles, and tips.
