It's a really bad plan to replace a cap that handles 3700 mA ripple with one that can only take 450 mA.
Particularly when ripple is that caps main job.
.

PCBONEZ wrote:

It's a really bad plan to replace a cap that handles 3700 mA ripple with one that can only take 450 mA.
Particularly when ripple is that caps main job.
.

The cap that I used to replace the old one with was purchased a while ago, and while it does work I'll try to find another one. Finding this one was kind of hard in the first place. I did check the new one's voltage and the writing on top. I did find it odd that the new one was smaller, but I couldn't find a capacitor of the same size. I'll not power it on again without outfitting it with the right cap. This will take some time, so aside from the cap maybe someone has more advice?

Is it possible that the old cap was broken, and the new one isn't good enough to make it functional again? Did you see anything else wrong on the card? What about CR1 and R112? Please note the only thing I changed on the card is the cap, everything else was like this when I received it from an eBay seller years ago.

Much thanks PCBONEZ!

I've been working with someone else in another thread for a day or two on the same card.
He's having similar problems to you. You prolly should read it.
We just worked out what the last two caps (the two big ones) need to be today.
(After I had a brain fart for a whole day. - damit!)
I will take another look and see if I see anything.
Voodoo 5 5500 AGP Compatibility Problem
.

There is a spec of something on the pins of U11. (RAM chip)
Make sure that's not a solder flake shorting them together.

CR1 looks fine. - Not unusual for parts to be off kilter.
So long as they are soldered in securely and not shorting something it's not a problem.

R112 is probably fine too but just to be sure can you read and tell what is written on it?
I can't make it out.

I didn't see anything else to check.

If you do not have an ESR Meter I would replace all the small caps.
Those are not the best brand, they are ~18 years old and replacements for those are cheap.

Most folks on your side of the pond go to RS Components or Farnell for parts.
I don't know what they stock.

How are you set for test equipment?
.

R112 has nothing written on it, I don't have the card with me right now but I'm 99% sure U11 has a dust particle on the picture.

Unfortunately I don't have an ESR meter, I do have both an analog and a digital multimeter.
Those two meters don't suffice for checking caps right? I would need to take them off the board and measure resistance then?

If I can find the right caps I might as well replace all of them, I'll check out those two shops, thanks.

To replace the cap that I put on the card, I would need this one (took it from the other forum topic) right?

470uF 6.3v ESR=0.025 Ripple=3700
http://nl.farnell.com/panasonic-electronic-co … md/dp/2354724RL

This one is the right one, right?

Do you think I should replace that cap first or is it better to just replace all of them? They're nearly 20 years old and I have no idea if they still work correctly.

Yes that's actually the same cap from the same series. Can't get a more exact match than that.
At the time the card was built Sanyo owned OS-CON, now Panasonic owns it.
That particular series has stayed in production.
The markings have changed a little and the can style may be modernized, but that's all.
The specs are still the same.
.
Will answer your other questions in a bit. Need to stretch my legs.
.

.
With those meters you can't fully test caps but there are a few tests you can do.
They won't tell you for certain the cap is good, but for some failure modes they can tell you for certain it's bad.
The reason is they only check the DC characteristics.
ESR is an AC characteristic so they are not complete tests.
==================
The ESR in a bad cap can be HUGE and it won't affect the DC characteristics at all.

==================
Testing a Capacitor with an Analog Multimeter.
= Fully discharge the cap. (If high volts are suspected short through a 1k resistor until cap voltage is zero.)
= Set meter up for resistance checks. Select a high range - at least 20k Ohms - more is better..
= Connect the Meter leads to the Cap.
... Note The needle movement and reading. Small uF caps are done fast. Larger ones take longer.
=== Shorted Caps will show very low or no resistance and the needle will stop there.
=== Open Caps will not show any meter deflection.
=== Good Caps will start low and increase towards infinite. (Note the reading where it stops.)
(On the last one... You are watching the cap charge up from the battery in your meter. As the cap voltage goes up the current through the meter goes down so the resistance reads higher.)
== Now reverse the leads and do it again. Do not discharge in between.
... Note The needle movement and reading.
=== Good caps will peg the needle low (discharging through meter) and then do as before as they charge up in the other direction. The needle should stop roughly where it stopped with the other polarity.
==================
I have seen people try that with digital meters and it doesn't work well.
The numbers on a digital jump around too much and it's hard to tell what is going on.
It's best to have the needle so you can watch the movement.

Digitals do work fine if you are only checking for shorted caps.

==================
.... I am disregarding Leakage tests because I don't do them.
If a cap has significant leakage issues it will have other issues and I will catch it that way.
.... There are other tests but they are more complicated to set up (one is freakin'n dangerous!) and they don't tell you any more than the easy test I just detailed.
.

Justin1091 wrote:

I would need to take them off the board and measure resistance then?

For most tests the caps need to be off the board to test them. Especially in computer equipment.
Parallel paths or capacitance will give inaccurate readings otherwise.
Many ESR meter ads tout 'in circuit testing' but that only works if there are no parallel paths or caps.
In PC equipment most of the caps have parallel paths.

Justin1091 wrote:

Unfortunately I don't have an ESR meter, I do have both an analog and a digital multimeter.
Those two meters don't suffice for checking caps right?

At a minimum you need an ohm meter, DC voltage meter, an ESR meter and a capacitance meter.
There are meters that combined those into one meter. Truly good combined meters are not cheap.
Standalone capacitance meters are not expensive but make sure they handle the uF ranges you need.

GOOD standalone ESR meters range from around $90 for DIY kit forms to up to $150 for fully assembled.
(That is ignoring premium brands which can be up to $1300 for a handheld or $5000 for a lab benchtop.)
**ESR meters are not a product where higher cost means better.
**ESR meters designed for things like TV/Radio repair usually can't cut it for PC repair.
--- The ESR meter should have a 100kHz test frequency. At 100kHz you can compare the readings as read to datasheets. If it's not a 100kHz test signal you have to do a conversion to parse with the datasheet values and you'll be spending a lot of time with a calculator.
--- The ESR meter should be good down to 0.01 ohms at 100kHz. There are a number of expensive ones (even name brand) that can not do that.
--- You need one with a digital display. It's too hard to see the difference between 0.01 and 0.04 on an analog meter.

===
There are ways you can find ESR and capacitance using an O'Scope.
Although I own 3 O'Scopes I don't do it that way. It's too much trouble to set them up.
At this point it's been so long since I did it that I don't even remember how.

===
If you are only going to do caps occasionally then I don't recommend getting an ESR meter.
If you suspect bad caps then just replace them all. Many people do that.
You will have to decide where the cost of replacing a few caps unnecessarily vs the cost of an ESR meter tips the need/want scale for yourself.
.

I'm going to replace all the caps and see what happens. Thanks for the explanation. An ESR meter would be used once, not worth the money. Below links are what I think are the right items:

https://sinuss.nl/componenten/passieve-compon … onic-components

https://sinuss.nl/componenten/passieve-compon … onic-components

https://sinuss.nl/componenten/passieve-compon … onic-components

Are these the right ones?

Justin1091 wrote:

I'm going to replace all the caps and see what happens. Thanks for the explanation. An ESR meter would be used once, not worth t […]
Show full quote

I'm going to replace all the caps and see what happens. Thanks for the explanation. An ESR meter would be used once, not worth the money. Below links are what I think are the right items:

https://sinuss.nl/componenten/passieve-compon … onic-components

https://sinuss.nl/componenten/passieve-compon … onic-components

https://sinuss.nl/componenten/passieve-compon … onic-components

Are these the right ones?

Yes those selections are good.
.

New update:

So I've replaced every cap on the board, still same problem. Hmm, no idea how to proceed. One would think one of the vsa chips is bad, but how can they function perfectly with 'guardband clipping' disabled...

Edit: I still find it odd that R112 has nothing written on it (and this one is colored), unlike every picture of Voodoo 5's I see online and a working one I have. Could one resistor cause problems? Maybe the two coils need looking into?

I'm not giving up so more suggestions are appreciated!

I am not a video card expert so I don't inherently know what to look for.
Also I don't have a schematic or pinout for these chips so anything here forward is just easter-egging and guessing.

When you go into single chip mode does it let you select which chip is active?
If not do you know which one is active?

Complex IC's usually have multiple small supply voltages (sub 3.3v) that are provided by local POL (Point of Load) regulators.
Some of those can be very small SMD MOSFET based circuits.
Some have multiple MOSFETs in one package so more legs than you would expect looking for MOSFETs.

If I understand correctly you are only having trouble in dual-chip 3D modes. ( Is that correct? )
The objective becomes to find a defective POL regulator (and/or circuit) that only affects 3D functions.
To determine which chip (or if both) triggers the problem might help narrow down which regulators to look at.

If both chips have a same dedicated regulator (each it's own but the same) you can do comparison checks between the two.

All of that is looking for power to chip problems which is not the only possible problem, but a good place to start.
Not a bad idea to draw the schematic for anything (sections) you trace out.
.

PCBONEZ wrote:

When you go into single chip mode does it let you select which chip is active?
If not do you know which one is active?

No, it doesn't let me choose. If I understand correctly it has a master and a slave chip.

Complex IC's usually have multiple small supply voltages (sub 3.3v) that are provided by local POL (Point of Load) regulators.
Some of those can be very small SMD MOSFET based circuits.
Some have multiple MOSFETs in one package so more legs than you would expect looking for MOSFETs.

I've no knowledge or experience other than measuring continuity and a little with resistance. Do you mean Q5? How would one find a POL regulator on a card?

If I understand correctly you are only having trouble in dual-chip 3D modes. ( Is that correct? )
The objective becomes to find a defective POL regulator (and/or circuit) that only affects 3D functions.
To determine which chip (or if both) triggers the problem might help narrow down which regulators to look at.

Yes, that is correct. Maybe I can let it run in single chip mode for a while and feel which chip is warmer (in Windows 98, they both feel warm to the touch). The new capacitors at c90 and c106 also get a little warm (nothing like the chips though).

If both chips have a same dedicated regulator (each it's own but the same) you can do comparison checks between the two.
All of that is looking for power to chip problems which is not the only possible problem, but a good place to start.

Can you maybe point me in the right direction where to start? I have another voodoo5 which I can use to measure things with the multimeter and compare the results of the two cards. For instance R112 has 828 written on the good voodoo5, while the one we're talking about has only a blue color on it. Could it be useful to compare the two resistance values between the two? I did get two different readings, but I probably do it wrong. Other resistors gave almost the same values (when compared with the other voodoo card).

Edit: found a sheet http://www.voodooalert.de/board/index.php?pag … 9046&highlight= maybe helpful for others

According to this sheet, R112 should be 6.98. The spare voodoo5 reads 6.94. While the voodoo5 that doesn't work reads 5.63 (earlier 4.30). Could this be something worth looking into?

That list will help. Nice find.
I've been spending some of my breaks looking for a VSA100 pinout. No luck so far.
(Bad back. I take back breaks. Helps that I am my own boss.)

I don't want to get your hopes up.
I could do this in my own shop with my tools but through the Internet with someone else's tools I'm not so confident.
Also my availability will be spotty. I've only been active on Vogons this time because I was convalescing from major surgery. At this point I'm convalesced and I'm way behind on everything. I was planning on bailing out of here for a while. Additionally I just found out one of my in-laws is going in for surgery for cancer so I may have to head out of state soon.

I will try to pop into this thread every day or two to see how you are doing.
Hopefully an actual video card guru will show up because, that I am not.

Caps getting a little warm is normal. Just means they are busy removing ripple.
If the heat seems too high check the PSU for excessive ripple.

R112 is +/-5% so should be 6.593 to 7.287 ohms.
Yes, it's the wrong part or bad and should be replaced in the course of things
It's early in the main power circuit so probably not the cause of your 3D only issues.

I looked at your photos again earlier and made some notes..
I didn't see as many POLs as I expected.
I think it's using resistor dividers to get the voltages they want from the card's main power.

MOSFETs and such would probably be labeled Q___ on that board.
More than one in a package might be Q___ or U___.
Q is a generic designation for transistor. (Which includes MOSFETs and others.)
U is a generic designation for IC chips. (Which are usually collections of transistors and other things.)

There is something that looks like it says P59
No idea about that one. Can you post the numbers on it?

There are several RN___
I think those are resistor networks. Please check one to verify that.

I have been tracking down the datasheets that might be needed on my breaks.
Whenever possible I find them for the exact original part instead of just the family.
IR (International Rectifier) is now Infineon.
Fairchild is now ON Semiconductor

I ran out of time finding the sheets. Haven't looked at them yet.
Please list the part numbers for those I missed and verify the ones I already found are correct.
I'm looking at photos, not a card. Your photos are good. My eyes, not so much.

Updates in yellow
U1 = CMD (California Micro Devices) PACVGA105Q = VGA Port Companion Circuit
http://pdf.datasheetcatalog.com/datasheet2/c/ … wzl4ku058py.pdf
U4 = NJA = ??
U6 = Fairchild VCX08 (74VCX08) = Logic Gates Quad 2-Input AND Gate
https://www.fairchildsemi.com/datasheets/74/74VCX08.pdf
U7 = IDT QS3861Q = CMOS 10-BIT BUS Switch
http://pdf.datasheetcatalog.com/datasheet2/f/ … 9wk1q4590cy.pdf
U8 = Atmel AT49BV512-12JC = NOR Flash 512k (64kx8)
https://www.mouser.com/datasheet/2/268/atmel_ … -16-1180544.pdf
U12 = LTC1929CG = 2-Phase Synchronous Step-Down Switching Regulator (MOSFET Controller)
http://www.analog.com/media/en/technical-docu … heets/1929f.pdf
U13 = BMS = ??
AR1 = National Semiconductor LM358M = Dual Op Amp
http://elcodis.com/parts/2211669/LM358M.html
Q1, Q2, Q3, Q4= IR IRF7811A = MOSFET N-CH 28V 11.4A 8-SOIC
http://www.irf.com/product-info/datasheets/data/irf7809a.pdf
Q5 = IR LR2703 = MOSFET 30V 1 N-CH HEXFET
https://www.mouser.com/datasheet/2/196/irlr27 … pbf-1228149.pdf
Please add any Q_ I missed.

You can look in the datasheets. Figure out which pins are in or out.
And track down where they go with continuity checks.
For DC in/out you should also check the voltages against the good card.
Make sure they are in the same condition (3D on/off) when you check.

There are TP___ all over the board. Those are "Test Points".
Take the voltages at the TP's and compare to the functional card.
Also compare in conditions with and without the 3D problem active.
That might give some clues where to look for problems.
== NOTE. Some of the TP's are probably for PWM or video signals. For those you can't check the intended TP parameter without an O'Scope but what DC is doing still might give clues. Some may not have a DC voltage at all.
Also with the cards powered down resistance (to ground) might give a clue.

---
After looking closer at the PCB I suspect the 12v and 5v from the cable conn. is being converted in parallel to 3.3v.
Q5 might be doing the same thing from slot (5v or 12v) power but I can't tell where it's pins go.
.

That list will help. Nice find.
I've been spending some of my breaks looking for a VSA100 pinout. No luck so far.
(Bad back. I take back breaks. Helps that I am my own boss.)

I believe I've found it, see attachment at this post.

I don't want to get your hopes up.
I could do this in my own shop with my tools but through the Internet with someone else's tools I'm not so confident.
Also my availability will be spotty. I've only been active on Vogons this time because I was convalescing from major surgery. At this point I'm convalesced and I'm way behind on everything. I was planning on bailing out of here for a while. Additionally I just found out one of my in-laws is going in for surgery for cancer so I may have to head out of state soon.

Damn, that's just horrible. I hope he/she will be okay. My sympathies.

R112 is +/-5% so should be 6.593 to 7.287 ohms.
Yes, it's the wrong part or bad and should be replaced in the course of things
It's early in the main power circuit so probably not the cause of your 3D only issues.

Ok! I will replace it in the future.

I looked at your photos again earlier and made some notes.. I didn't see as many POLs as I expected. I think it's using resistor […]
Show full quote

I looked at your photos again earlier and made some notes..
I didn't see as many POLs as I expected.
I think it's using resistor dividers to get the voltages they want from the card's main power.

MOSFETs and such would probably be labeled Q___ on that board.
More than one in a package might be Q___ or U___.
Q is a generic designation for transistor. (Which includes MOSFETs and others.)
U is a generic designation for IC chips. (Which are usually collections of transistors and other things.)

There is something that looks like it says P59
No idea about that one. Can you post the numbers on it?

There appears to be a piece of tobacco on it, hah. I'm not able to take a look now but I will tomorrow.

There are several RN___
I think those are resistor networks. Please check one to verify that.

I will tomorrow.

I have been tracking down the datasheets that might be needed on my breaks. Whenever possible I find them for the exact original […]
Show full quote

I have been tracking down the datasheets that might be needed on my breaks.
Whenever possible I find them for the exact original part instead of just the family.
IR (International Rectifier) is now Infineon.
Fairchild is now ON Semiconductor

I ran out of time finding the sheets. Haven't looked at them yet.
Please list the part numbers for those I missed and verify the ones I already found are correct.
I'm looking at photos, not a card. Your photos are good. My eyes, not so much.

U1 = CMD (California Micro Devices) PACVGA105Q = VGA Port Companion Circuit
http://pdf.datasheetcatalog.com/datasheet2/c/ … wzl4ku058py.pdf
U4 = Can't read
U6 = Fairchild VCX08 (74VCX08) = Logic Gates Quad 2-Input AND Gate
https://www.fairchildsemi.com/datasheets/74/74VCX08.pdf
U7 = IDT QS3861Q = CMOS 10-BIT BUS Switch
http://pdf.datasheetcatalog.com/datasheet2/f/ … 9wk1q4590cy.pdf
U8 = Atmel AT49BV512-12JC = NOR Flash 512k (64kx8)
https://www.mouser.com/datasheet/2/268/atmel_ … -16-1180544.pdf
U12 = LTC1929CG = 2-Phase Synchronous Step-Down Switching Regulator (MOSFET Controller)
http://www.analog.com/media/en/technical-docu … heets/1929f.pdf
AR1 = National Semiconductor LM358M = Dual Op Amp
http://elcodis.com/parts/2211669/LM358M.html
Q1 = IR IRF7811A = MOSFET N-CH 28V 11.4A 8-SOIC
http://www.irf.com/product-info/datasheets/data/irf7809a.pdf
Q2 = Can't read
Q3 = Can't read
Q4 = Can't read
Q5 = IR LR2703 = MOSFET 30V 1 N-CH HEXFET
https://www.mouser.com/datasheet/2/196/irlr27 … pbf-1228149.pdf
Please add any Q_ I missed.

You can look in the datasheets. Figure out which pins are in or out.
And track down where they go with continuity checks.
For DC in/out you should also check the voltages against the good card.
Make sure they are in the same condition (3D on/off) when you check.

Will do!

There are TP___ all over the board. Those are "Test Points". Take the voltages at the TP's and compare to the functional card. A […]
Show full quote

There are TP___ all over the board. Those are "Test Points".
Take the voltages at the TP's and compare to the functional card.
Also compare in conditions with and without the 3D problem active.
That might give some clues where to look for problems.
== NOTE. Some of the TP's are probably for PWM or video signals. For those you can't check the intended TP parameter without an O'Scope but what DC is doing still might give clues. Some may not have a DC voltage at all.
Also with the cards powered down resistance (to ground) might give a clue.

Interesting! I didn't know about those TP's. For testing I guess I'll power it on and attach both multimeter wires to one and see how much voltage it reads, right?

I want to say that I really appreciate your help, and I doubt someone else on this board has this much knowledge about this kind of stuff. Thank you, I think I will post an update tomorrow. I just thought about reducing the clock speed of the chips, maybe it won't lock up then.

So I've had a little time today. Think I have plenty of time tomorrow.

P59 says '472'. Just like the other RN's. I think it's a resistor or resistor network. Multimeter told me 4.47 ohms. Just like RN9 next to it.

About the resistor networks, yes I think they are just that. Multimeter read about the same numbers in ohms as written on them.

U1 = correct
U4 = NJA (did a quick search, couldn't really find something other than a place to buy it (alibaba.com or something).
U6 = correct
U7 = correct
U8 = correct
U12 = correct
U13 = BMS (found this one near the power connector, found the following;
http://www.alldatasheet.com/datasheet-pdf/pdf … 07C11BMS+H.html
http://www.alldatasheet.com/datasheet-pdf/pdf … 07C15BMS+H.html
http://www.alldatasheet.com/datasheet-pdf/pdf … 07C55BMS+H.html
http://www.alldatasheet.com/datasheet-pdf/pdf … E1357BMS+H.html
http://www.alldatasheet.com/datasheet-pdf/pdf … 7701ABMS+H.html
No idea which one is right at the moment.
Edit: think neither of them, U13 has 5 pins, not 3.

AR1 = correct
Q1 = correct
Q2 = same as Q1
Q2 = same
Q3 = same
Q4 = same
Q5 = correct

I did not find any more Q's.

Also checked resistance on the TP's

TP4 and TP7 are connected: same on both cards
TP8 and TP10: bad card 01.4 - good card 01.5
TP9 and TP11: bad card 01.0 - good card 00.9
TP2 and TP5: bad card 176.0 - good card 178.5
TP1 and TP3: bad card 01.4 - good card 01.5

I'm guessing that doesn't tell us anything, I did try to measure DC voltage from them, but got nothing. I couldn't reach TP1 and 3 for voltage measuring, I'll need to take out the PC from the case and put it on desk to reach them safely.
I'd expect something from the TP's when it's powered on, maybe I measured them the wrong way (i measured them the same way as i did measuring the resistance from them, with the correct setting on the multimeter of course).

I had a quick look at some of the PDF's in your previous post, but couldn't find a pin IN or pin OUT for the chips. I'll do some more reading soon.

Downclocking the card didn't work, Windows 98 started but the screen was grey with lines through it. Rebooting and reverting to 166 MHz fixed this.

This is just to let you know I didn't go *poof* on you. I'm rather swamped ITRW yesterday and today.
I will be on the weekend too but I should have time on Friday.

Try the TP's this way...

[DC Volts Checks - with it setup so the problem is in effect.]
TP (+ lead) -to- Ground (- lead) -- Compare to other card.
By ground I mean the ground plane for the card. The ground pins on the molex should work as TP's.
The negative side of those caps you just replaced should work too. Check continuity to the molex ground.

[Resistance Checks - card out of machine is okay.]
TP (+ lead) -to- Ground (- lead) -- Compare to other card.
+3.3v (+ lead) -to- TP (- lead) -- Compare to other card.
- The 3.3v will be all over the card similar to the ground plane. It will be where the outputs of the 12v and 5v DC-to-DC converters come together and then distributed through the card traces. There may even be a TP devoted to it.
Only need to check to 5v and 12v if 3.3v shows a complete open indicating it's not on 3.3v power.
+12v (+ lead) -to- TP (- lead) -- Compare to other card.
+5v (+ lead) -to- TP (- lead) -- Compare to other card.

Write everything down so you don't have to repeat checks.

That is all just Easter-egging to localize where to look for an actual problem.
Anyplace that doesn't match between the cards is where to look closer.
.

Before we go much further have you tried different firmwares?
I dunno which are good or bad. Just know there are more than one.
.

I found another TP, TP6. It's connected to TP5 and TP2. What do you mean with ground plane? Is the card's pci bracket sufficient for ground? Or Q6? If I understand correctly there should be a place on the card that is the ground plane, but I can't find it.

I did measure voltage of the connected tp's between the card's, can that tell us anything or do I need to do it again with 1 connected to ground?

Anyway, tp2 and tp5 1.65v both 2d and 3d for bad card, 1.70 for good card
Tp 5 and 6 1.70 volt both 2d and 3d for bad card, 1.65-1.68 good card
Tp 2 and 6 0.12 volt both cards

Near the molex connector on the pcb, there is written +5v and +12v. How can I measure either 3.3 5.5 or 12? You said all over the card, just like the ground plane, but what is a good place to attach the meter to? What about the 4 molex pins?