VOGONS


First post, by Frasco

User metadata
Rank Member
Rank
Member

Now that you said 6600GT in a previous topic a while ago,
I would love to play around with a MSI NX6600GT AGP (faulty) and take advantage to
learn a little bit about eletronic.

The card is artifacting in BIOS, POST screen, on DOS, etc... (weird characters)
I'm pretty sure this video card is gone. Oddly enough, every time I touch it, the artifacts dissapear.
Another push and artifacts say hello again.
Once it worked for 20 minutos without issues...Does it need a reflow ?

And now, what I would like to understand...
Some tools I got here (Multimeter/Capacheck) report 5 capacitors are shorted (marked in green).
What's going on here ? Cause or effect ?

6600GT AGP.JPG
Filename
6600GT AGP.JPG
File size
121.09 KiB
Views
670 views
File license
Fair use/fair dealing exception

I've been told I can't determine the health of capacitors (in circuit) for sure.
There may have a component with "low resistive value".

Should I:

  • Take them out and check one more time ? This way I can find out with certainty if capacitors are fine.
  • Let it go. Just too complicated to grasp. Take it to the doctor.
  • Forget all about those tools. You have to get a ESR meter.
  • Just recap the whole thing and leave us alone.

Capacitors can go bad without having any physical damage (Leaky/bulged) ?

Inverstigation:
PSU has 15A in 12V line. I don't think this is relevant. The card consumes only 80 Watts full load.
*** Here we go reading the topic "PSU - bust the myth" all the way.

Previous owner said the card was tested in a motherboard and since then the problems began...

So I am looking for some tips. I will accept >>> study this / read that / Frasco stinks
(rabbits stink reference here)
I am really lost! Kind of guy who can rescue some hardware through persistence, but don't have skills
with schematics, resistors, eletronics so to speak.

I have here with me two motherboards with same "issue" (shorted capacitors):

  • ASUS P5NE-SLI (smoked while I was messing with front panel connectors)
    Southbridge is shorted and getting hot as hell.
  • EPOX KL133-ML
    Running Windows 98, didn't take any stability test.
    Onboard sound is artefacting.

Reply 3 of 10, by nforce4max

User metadata
Rank l33t
Rank
l33t

I hope that you keep it and not toss it in the bin, agp cards will eventually start getting to be expensive as supplies of them dry up or greed takes over like what had happened to the retro consoles.

On a far away planet reading your posts in the year 10,191.

Reply 4 of 10, by kenrouholo

User metadata
Rank Member
Rank
Member

Sometimes testing caps in circuit works, sometimes not. It's particularly tricky when they are in parallel which at least some of those will be. The right way to test them is with an ESR meter. A standard ohmmeter or capacitance meter will not be sufficient. If using an ESR meter, you can almost always trust it to be correct if the cap reads bad. But you can't trust it much at all if the cap reads good in circuit. If the cap reads good out of circuit, it's probably good.

If the caps are dead, it's because they're right next to the heatsink. Make sure you have good airflow in your case. Do everything you can reasonably do to keep those capacitors as cool as possible.

Find a program that tests GPU RAM and run that.

Try underclocking the card with a tool like Afterburner and see if that helps.

Edit: And yes, it's possible that a reflow may fix it.

Yes, I always ramble this much.

Reply 5 of 10, by Frasco

User metadata
Rank Member
Rank
Member
nforce4max wrote:

I hope that you keep it and not toss it in the bin, agp cards will eventually start getting to be expensive as supplies of them dry up or greed takes over like what had happened to the retro consoles.

I have no compulsion to toss hardware not responding (although, as I said, I'm a terrible technician) 😠
I would love to save any hardware I can put my hands on, learn stuff and hence improve my rate.

Yeah, people these days are asking a lot of money for AGPs
I saw 20 new units of 7900GS AGP for 70US (each) and passed. What you say ? I kind regret that - The
feeling when you don't follow your intuition.
A good thing I bought a Sapphire X1950GT AGP for peanuts !
Peanuts as 3 SMD capacitors are missing on PCB. I can play games day and night, run benchmarks
on it just fine. Although this bothers me a little bit.

There is one guy here on local e-bay. He was talking to me and said something about the
missing components: must be filters, doesn't have any data function, decoupling, etc...
so the card doesn't lose data. His words.

I could send the 6600GT to him (and a 9800GTX artefacting / one missing capacitor).
This route would cost me gold (35$ each). And buying hot air station to fix 2 or 3 parts is no sense.
Is this method arguably?

Oh, yeah. Desoldering and soldering those mounted capacitors on the 6600GT would be a PITA.
So I guess the answer is take it to the doctor 🤣

I'm curious about the "shorted capacitors" though. Why this happen ? False positive ?
Nothing to worry about? Replace them now?

Reply 6 of 10, by kenrouholo

User metadata
Rank Member
Rank
Member

With missing capacitors, the card will put more electrical stress on the motherboard (and to a smaller degree the PSU itself), and more ripple may get through to the card, which could make your rig unstable and/or shorten the life of any of those components. And it will definitely shorten the life of any remaining capacitors that now see much more ripple (capacitors have a maximum ripple rating). Not saying it's going to die tomorrow, though it's certainly possible.

The prices you're giving all seem terrible... every card you mentioned should be able to be found on Ebay for US$30 or less. BTW, every "GS" card Nvidia ever made was crap.

Capacitors shouldn't read as shorted even in-circuit. Can you show a picture of that reading? Are you sure it's showing a short and not just some measurable amount of resistance? And you aren't confusing "OL" (overload i.e. too high to measure) with 0, right? Infinite or OL is what you'd want to see on a cap, 0 is not. Are you sure you didn't accidentally touch the probes together? If it really is a short then you might indeed have one of those rare cases where the cap is so bad that it really is measurable without an ESR meter. Could also be something else shorted, even a diode or transistor or something. But caps can fail with a short.

Yes, I always ramble this much.

Reply 7 of 10, by Frasco

User metadata
Rank Member
Rank
Member
kenrouholo wrote:

Sometimes testing caps in circuit works, sometimes not. It's particularly tricky when they are in parallel which at least some of those will be

I assume the mounted ones cause they are near. 😐 So they do that to increase their capacity ?
Never mind. I started reading something about capacitors in parallel and retreated in shame!
That's not easy. It's like coming back to high school.

kenrouholo wrote:

The right way to test them is with an ESR meter. A capacitance meter will not be sufficient.

I did think so. ESR meter is better than my gear.

kenrouholo wrote:

If the caps are dead, it's because they're right next to the heatsink. Make sure you have good airflow in your case. Do everything you can reasonably do to keep those capacitors as cool as possible.

Dead ? Dead how ? Shorted or presenting low capacitance ?
As I said, externally and physically they are like new ! The capacitance is correct, too.

kenrouholo wrote:

Make sure you have good airflow in your case.

I'm using all my retro stuff without a case (on a big desk). Heat is no problem, but thanks for the reminder.
Perhaps the previous owner maculated this card in any way you can think!

kenrouholo wrote:

Find a program that tests GPU RAM and run that.
Try underclocking the card with a tool like Afterburner and see if that helps.

Name ?
Underclocking can remove the artifacts? Then I have to record the new values into the video BIOS. I can try.

kenrouholo wrote:

With missing capacitors, the card will put more electrical stress on the motherboard (and to a smaller degree the PSU itself), and more ripple may get through to the card, which could make your rig unstable and/or shorten the life of any of those components. And it will definitely shorten the life of any remaining capacitors that now see much more ripple (capacitors have a maximum ripple rating). Not saying it's going to die tomorrow, though it's certainly possible.

This even applies to those tiny capacitors ? Scaring. The X1950GT will remain unused.
I tend to think they are less important. My ignorance. 😵

kenrouholo wrote:

The prices you're giving all seem terrible... every card you mentioned should be able to be found on Ebay for US$30 or less. BTW, every "GS" card Nvidia ever made was crap.

I agree on this one. A 7900GS for 70$ is so nuts. The problem is I always forget about e-bay to
compare prices and I gotta pay the shipment.
And sorry to disagree with you, but you couldn't be wronger about 8800GS. It's close to the 8800GT
in performance (-15% or so). This baby is not gonna die. 😈 3x better than a 8500GT 🤣

Reply 8 of 10, by nforce4max

User metadata
Rank l33t
Rank
l33t
Frasco wrote:
I have no compulsion to toss hardware not responding (although, as I said, I'm a terrible technician) :angry: I would love to sa […]
Show full quote
nforce4max wrote:

I hope that you keep it and not toss it in the bin, agp cards will eventually start getting to be expensive as supplies of them dry up or greed takes over like what had happened to the retro consoles.

I have no compulsion to toss hardware not responding (although, as I said, I'm a terrible technician) 😠
I would love to save any hardware I can put my hands on, learn stuff and hence improve my rate.

Yeah, people these days are asking a lot of money for AGPs
I saw 20 new units of 7900GS AGP for 70US (each) and passed. What you say ? I kind regret that - The
feeling when you don't follow your intuition.
A good thing I bought a Sapphire X1950GT AGP for peanuts !
Peanuts as 3 SMD capacitors are missing on PCB. I can play games day and night, run benchmarks
on it just fine. Although this bothers me a little bit.

There is one guy here on local e-bay. He was talking to me and said something about the
missing components: must be filters, doesn't have any data function, decoupling, etc...
so the card doesn't lose data. His words.

I could send the 6600GT to him (and a 9800GTX artefacting / one missing capacitor).
This route would cost me gold (35$ each). And buying hot air station to fix 2 or 3 parts is no sense.
Is this method arguably?

Oh, yeah. Desoldering and soldering those mounted capacitors on the 6600GT would be a PITA.
So I guess the answer is take it to the doctor 🤣

I'm curious about the "shorted capacitors" though. Why this happen ? False positive ?
Nothing to worry about? Replace them now?

This is one reason why I spend so much time looking through eBay even though some times still manage to slip by, bagged a XFX 7900 GS agp for $25 shipped and a X1950 pro agp for $8 on a as is that turned out to be working. Luck finds like that are getting to be HARD to come by lately and it is slowly starting to get on my nerves but also the volume of cards has dropped off and even modern cards like the GTX 600-700 series are drying up 😮

On a far away planet reading your posts in the year 10,191.

Reply 9 of 10, by Frasco

User metadata
Rank Member
Rank
Member
kenrouholo wrote:

Capacitors shouldn't read as shorted even in-circuit. Can you show a picture of that reading? Are you sure it's showing a short and not just some measurable amount of resistance? And you aren't confusing "OL" (overload i.e. too high to measure) with 0, right?

Yeah, I can take pictures of the device in action. Just ask.

Confusing Overload with zero ? No way. Though I saw null readings a couple of times when the capacitors
were really battered. And I'm positive. The capacitance meter shows short and even emits an alarm sound
and the red LED turns on.

kenrouholo wrote:

Infinite or OL is what you'd want to see on a cap, 0 is not. Are you sure you didn't accidentally touch the probes together? If it really is a short then you might indeed have one of those rare cases where the cap is so bad that it really is measurable without an ESR meter. Could also be something else shorted, even a diode or transistor or something. But caps can fail with a short.

Is OL good? I didn't get the point 😊
No. I took care of that, I didn't touch the probes. I only shake using the iron solder 😢

Primary quest in fulfilled. I don't know such things as diode. This 6600GT needs a pro.
Thank you for clarifications 😀

Reply 10 of 10, by Frasco

User metadata
Rank Member
Rank
Member
nforce4max wrote:

This is one reason why I spend so much time looking through eBay even though some times still manage to slip by, bagged a XFX 7900 GS agp for $25 shipped and a X1950 pro agp for $8 on a as is that turned out to be working. Luck finds like that are getting to be HARD to come by lately and it is slowly starting to get on my nerves but also the volume of cards has dropped off and even modern cards like the GTX 600-700 series are drying up 😮

A camper just like me 😈
Nothing beats buying motherboards "as-is". Little do they know you just got to read the manual,
clean them up and eliminate all the possibilities.
God only knows if I had skills on controlling electric energy in semiconductor.
***plenty of motherboards even in my kitchen***

I have a GTX650 Ti Boost 2GB laying around. No video and faint sound!
Man, that dude from my local e-bay gonna be rich.

1950 PRO AGP for $8 ? Awesome bargain. Even if it was faulty. Very well 😀