6800le 'Black Screen' - The Complete Guide: Identification, modification and reason

Quote:

---

Originally Posted by nobre

Although the speaker noise could be there before and I hadn't noticed, the noise directly coming from the card was definately not there before the resistor.
I added a 2.7kohm resistor that raised vmem from 2.56V to around 2.75V, no other problems added, my max oc bumped from 780 to 860 or so in memory (running daily at 800)
Would this noise be a "BS coming" sign? That's why I'm considering adding this resistor before my card starts crappying its pants :o

Thanks :)

---

The Vmem mod convert the card itself to noisy, interesting ! I suppose (just guess) that it has to be related to increase of current consumption. The higher the power level the larger temperature and stess effects. The noise generated by card has be be some kind of resonance of components and/or PCB (those act like tiny loudspeakers) and more power means higher amplitude/sound. I wonder the noise has been there all the time, but the amplitude was so low you did not heard it (?).

I am quite sure noisy sound is a sign of too high (ripple) current. However, I think that not all noisy card will be BS-affeted card during 'normal' life time (some years), it depends MANY things: component versions/values, ambient temperature, up time of computer, OC'ing, Vmodding, amount of 3D-gaming etc..

Just a thought: The idea is that the condensators are the main problem. But isn't the real problem in the ISL6534 short-detection circuitry? If that chip stops outputing the correct voltage, the condensators can hold the voltage for some time, but that chip shouldn't detect the short when there isn't really one, in the first place. Right?

I did a little experiment with connecting a LED between VCC and the PGOOD pin of that chip, and when I got the BS the LED did indeed lit up!

http://img468.imageshack.us/img468/9...aled0kn.th.jpg

How could we prevent the chip from thinking that there's a short? I don't know. Perhaps the feedback circuitry? Or are the FET's not big enough to deliver the amount of current for the RAM chips? Or could we just in some way bypass the short detection circuitry?

Quote:

---

Originally Posted by Hneel

Just a thought: The idea is that the condensators are the main problem. But isn't the real problem in the ISL6534 short-detection circuitry? If that chip stops outputing the correct voltage, the condensators can hold the voltage for some time, but that chip shouldn't detect the short when there isn't really one, in the first place. Right?

I did a little experiment with connecting a LED between VCC and the PGOOD pin of that chip, and when I got the BS the LED did indeed lit up!

http://img468.imageshack.us/img468/9...aled0kn.th.jpg

How could we prevent the chip from thinking that there's a short? I don't know. Perhaps the feedback circuitry? Or are the FET's not big enough to deliver the amount of current for the RAM chips? Or could we just in some way bypass the short detection circuitry?

---

Hi,

Nice thoughts and experiments :)

Yes, PGOOD signal changes state after BS and all outputs will be disabled except VREF (I tested those also myself, chapter 'Measurement results' ). The capacitor SHOULD provide sufficient amount current for PROPER time, but in reality it does not. Consider a case when current consumption increases a lot e.g. 1 A -> 6 A rapidly e.g. during some memory clock pulses. The capacitors should keep voltage high enough (=provide current) until next 'up' pulse charges them again. The error amplifier of ISL6534 compares output voltage and internal reference all the time. If the output voltage remains lower level than reference for 1-2 switching pulses (300 kHz) the circuit 'detects' condition as a short circuit. This means ISL6534 has NO normal overcurrent protection circuit - just the voltage difference is measured/analyzed. Normally this is not a problem, but if the ESR of the capacitors is much higher than specified the voltage drop over 'internal resistance' of capacitor (=ESR) reduces voltage obtained at output. Therefore, a VERY high current peek due to extreme 3D-activity on screen cause voltage drop that cannot be charged up during specified time -> 'short' -> BS

Adding extra capacitor or replacing to old capacitors with new helps because more charge is reserved (=voltage stays up longer up during high current pulse) and voltage does not drop much (=lower ESR).

Since there is no real short circuit detection circuit, it can not be disable or tweaked.
However, the feedback loop has MAJOR effect on performance as described during some posts in this thread (main obvervations documented in post #1, '3.4 Modification of Switching regulator Feedback'). I think that the different values of feedback components is main reason why e.g. MSI-card are much more harder to modify that e.g. GW-cards.

I have encountered some times quite a wierd 'BS-feature'. I am not yet sure it is a real BS, I wonder somebody has seen something similar effects to make this thing more clear (?).

'Case' is as follows:

- GW6800le-card 'version 1b' NO BS so far (HL2, SkiAlpin (demo), BF2 (demo), Cold War (demo) running just fine without BS)
- While running GTA-SA I have got so far 3-4 times RESTART without any warnings (no slowdowns etc.)
- Only one (?) has been real BS, the rest are 'normal' reboots
- Usually do not appear soon (at least half an hour) -> hard to debug !!


I have not seen any other bad problems with my current card and wondering whether it FINALLY ( after 8 months) will be BS-affected now or in near future (??).

Has anybody seen similar reboots or 'wierd' BS-features with GTA-SA ?

i had a rebooting problem with my x800 pro on GTA:SA.. and there were some other people that had the same problem.
From what i gather, it seems to be driver related (sound drivers maybe :confused: )
It was fixed after i got new drivers for everything and formatted

Sorry, but I'm afraid I don't think that information is really relative to this thread.

i was referring to the rebooting problem with his 6800, and how it was a common game bug, since he noted that his card had been free of the 'Black Screen" symptom for a while

Quote:

---

Originally Posted by oohms

i had a rebooting problem with my x800 pro on GTA:SA.. and there were some other people that had the same problem.
From what i gather, it seems to be driver related (sound drivers maybe :confused: )
It was fixed after i got new drivers for everything and formatted

---

Thank you for providing info :)

I googled again (more open eyes = not only GF6800 issues) and you are most likely correct ! I have quite new drivers (newest at least for sound = Audigy) but GTA-SA is NOT patched: maybe that could be one reason (I''ll patch it and test later..)

Anyway, I am quite sure one 'case' was a real BS (it happened ~ 2 months ago and I can not remember details: reboot or BS, because I was so surprised to get 'BS' again !!).

I have ideas to test a card 'near-to-BS-affected', and report the results afterwards..

I tested this 'GTA SA crashing'-issue with my card as follows.

In the very beginning I prepared myself to have cold spray near and opened PC chassis. After that I was gaming GTA SA for a while to 'heat-the-card-up'. Then I try to keep FPS up (=driving full speed in a high way) AND freezing C136 & C143 capacitors with the cold spray in the same time ! (It was quite funny (=*some* accidents ;) ). Anyway, I could NOT 'make' BS with my card even if the capacitors were fully frozen -> my card (=caps) is this in good condition and the 'crashing issue' is most likely a coding bug related to GTA SA.

After that demo I patched the game and now really hope no more craches..

---

In finnish forum was reported a BS-affected card similar to me i.e. having "4 7 470 6E" capacitors = version 1b = repaired-afterwards-by-GW (not sure in that case). The reason my card is still working might be because of lack of gaming :( not because the BS-problem would be solved by feedback modification (as I have been assumed).

Quote:

---

The reason my card is still working might be because of lack of gaming

---

Yes, because u r spending so much time helping everyone else fix their cards :)

Quote:

---

Originally Posted by KoR_Wraith

Yes, because u r spending so much time helping everyone else fix their cards :)

---

That is absolutely true :D, but hope was not 'completely-waste-of-time'.

--

'The BS-case' is almost fully considered. I am still going to make 1-2 updates to report rest of the measurement results (ESR & oscilloscope results) and maybe then is time to find out other 'unsolved' problems ;)

First of all thanks for finaly knowing what caused me so much frustration over the past few months.
I bought my XFX 6800LE in october 2004 [version 2] and was able to unlock 12x1,6vp without any artifacts.
So hell of a deal I thought.
First problems began with GTR after 3 or 4 months, nothing big just BS when watching a replay is tv-mode.
Last september [?] I bought GT-legends and as you guess same problem so I thought it had to do with the developer since they're both made by the same one.
Weird thing is, I never had BS problems with gta-vc. [finished playing oct]
After GT-legends it got worse by the minute till my mainboard broke down [leaking/bursting compensators] so I thought I knew what the problem was.
After that I struggled with one BS after the other, COD2, Brothers in arms2, GTR and GTL.
The only games that never had BS were Grandprix legends [old] and PES5.

For the solution, I got a site from a shop where I could pick the compensator[s] I wanted or needed and they'll order it for me.
Since I'm so bad at this stuff, could anyone pick one out for me?

http://uk.farnell.com/jsp/endecaSear...comSearch=true

Oh if you want pics or more info about my XFX 6800LE [rev2] just say so.
Allthough I have the second version I don't have the "new" capasitators/compensator.
It reads 44 470 6E

Yeah :) Today, after over 2 months since replacing old caps (no BS at stock speed), I added cap to TP9-TP10. Looks like all capacitors are in 100% healthy condition :) - memories on my card can do 875Mhz without BS :D
I also checked 900mhz but artifacts appeared (but no BS). I hope it won't break down over the next few years ;)
Again, BIG thanx to SikaRippa for his work. :toast:

Quote:

---

Originally Posted by Temp-Guest

First of all thanks for finaly knowing what caused me so much frustration over the past few months.
I bought my XFX 6800LE in october 2004 [version 2] and was able to unlock 12x1,6vp without any artifacts.
So hell of a deal I thought.
First problems began with GTR after 3 or 4 months, nothing big just BS when watching a replay is tv-mode.
Last september [?] I bought GT-legends and as you guess same problem so I thought it had to do with the developer since they're both made by the same one.
Weird thing is, I never had BS problems with gta-vc. [finished playing oct]
After GT-legends it got worse by the minute till my mainboard broke down [leaking/bursting compensators] so I thought I knew what the problem was.
After that I struggled with one BS after the other, COD2, Brothers in arms2, GTR and GTL.
The only games that never had BS were Grandprix legends [old] and PES5.

For the solution, I got a site from a shop where I could pick the compensator[s] I wanted or needed and they'll order it for me.
Since I'm so bad at this stuff, could anyone pick one out for me?

http://uk.farnell.com/jsp/endecaSear...comSearch=true

Oh if you want pics or more info about my XFX 6800LE [rev2] just say so.
Allthough I have the second version I don't have the "new" capasitators/compensator.
It reads 44 470 6E

---

Hi,

You have encountered lost of hardware problems, indeed :( The bad capacitors used in MOBO's is very annoying issue. I have myself (so far) installed new capacitors to two MOBO's, and maybe just before those broke down (MSI & Epox - thanks for that !!).

---

Yep. Your card is most likely 'BS-affected', althought no BS in GTA-SA. I wonder only few/some (?) users get easily BS with GTA-SA, BUT almost all get BS with Half-life 2 or SkiAlpin2005 (demo) or F.E.A.R. Have you tested with those ?

Have you tried to add GOOD capacitor between TP's ? It is easy to test and will work long time in SOME cases.

If you go for replacing capacitors, here is the best that Farnell can provide:

http://uk.farnell.com/jsp/endecaSear...=4310263&N=401

--

Fotos are allways welcome :) I have not investigated XFX-card in very detailed, so all extra info might be helpfull.

Quote:

---

Originally Posted by piaskoon

Yeah :) Today, after over 2 months since replacing old caps (no BS at stock speed), I added cap to TP9-TP10. Looks like all capacitors are in 100% healthy condition :) - memories on my card can do 875Mhz without BS :D
I also checked 900mhz but artifacts appeared (but no BS). I hope it won't break down over the next few years ;)
Again, BIG thanx to SikaRippa for his work. :toast:

---

Hi piaskoon :)

Nice to heard mods are working just fine, and quite good OC-capability :up:

I wonder capacitor replacement ("ultimate modification") has been 'repaired' BS-affected cards in most (almost all ?) cases. There have been reported some cases where BS has not been disappeared or something other 'issues' like artefacts. etc. However, I think those cards were not repaired due to bad soldering or ESD-shock (destroys components).

One example. I sent a pair of PXA-capacitors to one user. He changed capacitors and card worked for while, but severel issues appeared after ~some weeks e.g. BS again, artefacts... After resoldering of those capacitors the card worked better some weeks, but again similar problems. I sent a new pair of PXA-capacitors for him (this time larger capacitance, lower ESR = better). He sent back the first PXA-capacitors (=sited in card for ~2 months), because I was willing to test whether those have been ruined during that period. I noticed that the leads of the capacitors were almost clean of solder , which means that there has been never a GOOD conteact between lead and PCB. I measured the ESR (( Yes my ESR-meter is finally working and gives quite accurate reasults :D )) of those 'used' PXA-capacitors and compared results to those obtained with a new one. I found that ESR of 'used' PXA's was exactly the same as in new -> PXA's did NOT get ruined during that period, and I assume the problems were due to 'cold soldering'.

Here is a 'teaser' about ESR-measurements, that I will hopefully post in near future. Below is a foto of my ESR-meter and device-under-test is Panasonic FC-capacitor and in series a 30 milliohm resistor ;) That foto gives some idea *how* difficult is to measure milliohm range resistances: calibration 'resistor' is a copper wire having length of some centimeters :eek:

http://img74.imageshack.us/img74/596...258c8hl.th.jpg

Quote:

---

Originally Posted by SikaRippa

Hi,

You have encountered lost of hardware problems, indeed :( The bad capacitors used in MOBO's is very annoying issue. I have myself (so far) installed new capacitors to two MOBO's, and maybe just before those broke down (MSI & Epox - thanks for that !!).

---

Yep. Your card is most likely 'BS-affected', althought no BS in GTA-SA. I wonder only few/some (?) users get easily BS with GTA-SA, BUT almost all get BS with Half-life 2 or SkiAlpin2005 (demo) or F.E.A.R. Have you tested with those ?

Have you tried to add GOOD capacitor between TP's ? It is easy to test and will work long time in SOME cases.

If you go for replacing capacitors, here is the best that Farnell can provide:

http://uk.farnell.com/jsp/endecaSear...=4310263&N=401

--

Fotos are allways welcome :) I have not investigated XFX-card in very detailed, so all extra info might be helpfull.

---

I will post some high-ress photo's on friday.
Nope haven't done the tp9-tp10 mod yet since I can't seem to make a choise
in what the best choise is on that Farnel site.
The one you picked is for replacing right?
I just need one that's best to add. [tp9-tp10]
Can you help me out with that?

Oh another great BS testing game is "Action Ball deluxe".
Demo is free to download and gives me 100% sure BS.

So since I struggle BS I found out a few things. [you wouldn't wanna know how many hours I tried to ivestigate what the problem was and how to get rid of it.]
Don't know if it works for everyone and it defenetly doesn't work long.
-seems that when it's warm in the room [heating] on I get less BS?
I realy tested that a lot and stumbled on it when I thought temp might be the problem.
In my case this always seem to work [sofar].
-After installing older drivers I get less BS.
Tried that 3 times.
-Odd but with me I get less BS with v-sync off instead of on.

This realy is a weird problem since some simple games like that Ski demo or Actionball give BS whereass PES5 doesn't give me BS.
I realy don't know why GTA-VC never gave me BS, if I got time I will test it this weekend.

Quote:

---

Originally Posted by Temp-Guest

I will post some high-ress photo's on friday.
Nope haven't done the tp9-tp10 mod yet since I can't seem to make a choise
in what the best choise is on that Farnel site.
The one you picked is for replacing right?
I just need one that's best to add. [tp9-tp10]
Can you help me out with that?

Oh another great BS testing game is "Action Ball deluxe".
Demo is free to download and gives me 100% sure BS.

So since I struggle BS I found out a few things. [you wouldn't wanna know how many hours I tried to ivestigate what the problem was and how to get rid of it.]
Don't know if it works for everyone and it defenetly doesn't work long.
-seems that when it's warm in the room [heating] on I get less BS?
I realy tested that a lot and stumbled on it when I thought temp might be the problem.
In my case this always seem to work [sofar].
-After installing older drivers I get less BS.
Tried that 3 times.
-Odd but with me I get less BS with v-sync off instead of on.

This realy is a weird problem since some simple games like that Ski demo or Actionball give BS whereass PES5 doesn't give me BS.
I realy don't know why GTA-VC never gave me BS, if I got time I will test it this weekend.

---

Thank you for info :up:

I'll check out 'Action Ball deluxe', it is very good/usefull to find free small size programs with '100 % BS guaranteed'-feature.

--
Yes, Sanyo SVP's I suggested are for replacement, for adding one between TP's there are sooooo many alternatives e.g. this one:

http://uk.farnell.com/jsp/endecaSear...=9691987&N=401

.. see the document for detailed specs. You can also test the mod with old capacitor ripped off from old Mobo etc. However, check before using old one e.g. with multimeter it is not a SHORT.

--

Heat is an important issue. Did you check the 'Temperature measurements'-chapter ? I have also tested/played with heat/freeze and found that BS can be triggered by freezing 'green capacitors'.

--

Old drivers might have effect on BS-feature. Since BS is triggered by very high 3d-load, some (older) drivers might be unoptimal = can not take all power from card -> BS does not appear so easily.

--

It is unusual that vsync does not 'work' in regular manner. I have heard the same stories a few times before, maybe it is due to some combination of drivers, settings etc. (?)

Quote:

---

I'll check out 'Action Ball deluxe', it is very good/usefull to find free small size programs with '100 % BS guaranteed'-feature.

---

lol, it is funny that we find it good that programs make our computer crash...

Hello folks! Sorry for my english, but urgent help is necessary to me.
I broke the capacitor which marked C501 on my LeadtekA400TDH(geforce6800 reference design). It is near to swithing regulator ISL6534(opposite) Who can measure value of this component? Please help me (

I have noticed an interesting thing. When i broke this capacitor C501 my 6800 left in BS on windows desktop(when nvidia driver is installed) and always LOUD NOISE ! This capacitor connected to ISL6534, on the line which power a videocard chip. I install capacitor(jamicon wl 3300 10V) on TP9 and TP11(chip power), and my video work 100% stability(loop test in 3dmark05) AND NOISE HAS DISAPPEARED. When i remove capacitor from TP11, and install to TP10. But videocard left in BS on windows desktop...
I think that to solve BS problem is necessary directly on an input power. This solve more stability and quality!!!

picture of broken capacitor
http://img81.imageshack.us/img81/2168/c501ed9jx.th.jpg

Quote:

---

Originally Posted by KoR_Wraith

lol, it is funny that we find it good that programs make our computer crash...

---

Yep.. it is funny & sad - but TRUE ;) At least I think it would be great to find program(s) (games) that generate BS with 100 % 'reliability'. When time to verify, modify or RMA card nice to be sure the card really is BS-affected.

--

BTW, I tested 'Action Ball deluxe' ~15 minutes - and no BS with my card ((due to reasons already told..)

Quote:

---

Originally Posted by LLIaJLyH

Hello folks! Sorry for my english, but urgent help is necessary to me.
I broke the capacitor which marked C501 on my LeadtekA400TDH(geforce6800 reference design). It is near to swithing regulator ISL6534(opposite) Who can measure value of this component? Please help me (

I have noticed an interesting thing. When i broke this capacitor C501 my 6800 left in BS on windows desktop(when nvidia driver is installed) and always LOUD NOISE ! This capacitor connected to ISL6534, on the line which power a videocard chip. I install capacitor(jamicon wl 3300 10V) on TP9 and TP11(chip power), and my video work 100% stability(loop test in 3dmark05) AND NOISE HAS DISAPPEARED. When i remove capacitor from TP11, and install to TP10. But videocard left in BS on windows desktop...
I think that to solve BS problem is necessary directly on an input power. This solve more stability and quality!!!

picture of broken capacitor
http://img81.imageshack.us/img81/2168/c501ed9jx.th.jpg

---

Hi,

I will TRY to measure that capacitor. Problem is that obtained value is affected by other components on PCB, but let's see .. I'll report later.

Anyway, that component is between Vcore and FB1 of regulator. It is part of the feedback (filter) which determines the operation and performance of the regulator. So, by missing that component is it miracle to have ANY operation. The capacitor you added somehow 'repairs' that condition, but propably the voltage levels and variations are out of range.

Did you encounter any noise BEFORE you lost the capacitor ? What kind of noise, modulated by FPS etc. ? Is the noise coming from card or loudspeakers ? Any BS problems before ?

Quote:

---

Originally Posted by LLIaJLyH

I broke the capacitor which marked C501 on my LeadtekA400TDH(geforce6800 reference design). It is near to swithing regulator ISL6534(opposite) Who can measure value of this component?

---

I measured/obtained an APPROXIMATION for the capacitor value. I used the same method as explained in the document ( post #1), when I tried to acquire values for feedback components.

A short explanation:

- Capacitance (or resistance) measurement of onboard components is possible in some cases if several precautions have been taken into account
- All other components connected to corresponding nodes AFFECT on results
- By shorting etc. some nodes it is possible to prevent 'parasitic' effects
- In the case of capacitance two aspects affect on measurements: additional capacitances on board and resistance parallel to capacitance !!!!

I did as follows:

1. Measured capacitance over those pads (shown above in figure by 'LLIaJLyH') ~ 17 nF
2. Measured resistance over those pads ~ 500 ohm
3. Using discrete components I constructed a 'dummy' having 500 ohm resistor and several ceramic capacitors in parallel (=to 'emulate' PCB). I added (small) capacitances until I got value of 17 nF again. After that I removed resistor and measured capacitance again and got value ~277 nF - quite considerably difference when having resistor in parallel !!!
4. Analyzing results. I wonder value of 277 nF is in correct MAGNITUDE, but some error, say +/- 20 % is due to 'non-direct' method. There are some other (parasitic) capacitances in parallel in that node so obtained value is sum of all those capacitances. However, I suppose that one is the major capacitor connected to the node because here is required quite large capacitor (feedback of the Vcore regulator, this capacitor 'boost' gain in higher frequencies of feedback)

Therefore, I suggest to try first a ceramic capacitor 270 nF (a standard value) also values from 150 nF to 330 nF could be possible (??) if 270 nF are not available. Please post results, if you will try this out !!

---

I was thinking you case for a while and it is quite logical you got BS. The BS is result of regulator to shut down due to 'unstable' or 'too low level' output voltage. Usually Vmem fails (due to known reasons), but in your case the missing capacitor prevents card to operate properly while having high core activity - exactly in the same manner as memory voltage fails during a *standard* BS.

Sorry for not being able to post any pics yet, couldn't get a sharp pic yet with my crummy cam.
The good news is that my ordered compensators are on their way. :slobber:
The funny part is that I havn't seen any BS in GTL for the past week.
Then again, I'm playing in dx7 mode, 1024x768 with low/medium details/effects.

oh, you mentioned trying compensators from a mainboard but how do you get them of?

Quote:

---

Originally Posted by Temp-Guest

Sorry for not being able to post any pics yet, couldn't get a sharp pic yet with my crummy cam.
The good news is that my ordered compensators are on their way. :slobber:
The funny part is that I havn't seen any BS in GTL for the past week.
Then again, I'm playing in dx7 mode, 1024x768 with low/medium details/effects.

oh, you mentioned trying compensators from a mainboard but how do you get them of?

---

No hurry at all ;) BTW a good way to take sharp 'fotos' of card etc. flat items is scanner.

What drivers you have now, those old ? BS is highly depend on ambient temperature which also could explain varying 'performance' (e.g. long frequent gaming sessions make BS appear more easily).

--

Capacitor removing requires lot of heat and you can use e.g. solder wick to remove most of the solder beforehand. If you are going to use old capacitors be sure those are not TOTALLY ruined (=short) or bulbing etc. Take a look these fotos to get idea how a COMPLETELY RUINED capacitor looks like.

http://www.badcaps.net/ident/