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

SikaRippa
i discovered that this capacitor case size is 0402. 0402 is smaller than 0603(0603 is next from 0402). Then i found the table of chip capacitor where is a dependence of capacity on the size. The max capacity of 0603 is approximately 0,22uF. Then i buy capacitor 47nF(i think approximately max for the 0402) and soldered it.
After that my VGA(unblocked 16x1,6vp) works fine with 100% stability and no BS. C136, C143: 49 470 6E. To videocard a year. But it is necessary even more test...
I have asked my friend to measure value of C501(soldered) and his MM show 25nF.
Maybe this solve to remove BS?!

Quote:

---

Originally Posted by LLIaJLyH

SikaRippa
i discovered that this capacitor case size is 0402. 0402 is smaller than 0603(0603 is next from 0402). Then i found the table of chip capacitor where is a dependence of capacity on the size. The max capacity of 0603 is approximately 0,22uF. Then i buy capacitor 47nF(i think approximately max for the 0402) and soldered it.
After that my VGA(unblocked 16x1,6vp) works fine with 100% stability and no BS. C136, C143: 49 470 6E. To videocard a year. But it is necessary even more test...
I have asked my friend to measure value of C501(soldered) and his MM show 25nF.
Maybe this solve to remove BS?!

---

Good investigations & mod ! Yes it is '0402' size and the maximum value is propably quite small. In fact there exist capacitors up to 1uF available in 0402 package but the price is extremely high (=manufacturers do NOT use such components if ANY space left on PCB).

As I explained earlier the capacitance value I got was 17 nF, which is quite near to 25 nF (difference propably due to different meter). BUT due to parasitic other components (see carefully my previous post) the value in the meter is NOT capacitance value of C501.

Anyway, when having good performance with 47 nF I think no need to test other values again :) Do you have still extra capacitor ??

First i think need to solder out original C501 and measure value. To compare with 47nF. Then to solder 47nF or it is more value on BS affected card to test. I can`t, because i don`t know my VGA BS affected or not...

Quote:

---

Originally Posted by LLIaJLyH

First i think need to solder out original C501 and measure value. To compare with 47nF. Then to solder 47nF or it is more value on BS affected card to test. I can`t, because i don`t know my VGA BS affected or not...

---

Sorry took little time to answer.. (forum was down few days?)

--

Yep, removing C501 and measuring would give accurate value. I am hoping to get/obtain some day a burn (or almost burn) 6800le-card to rip-off/measure ALL relevant (and even unrelevant :D) components.

I am quite sure the BS your got was very similar to 'standard' BS. The more I thought your explanation the more clear the case showed out: BS is 'feature' caused by faulty over current (=low level voltage) protection of the regulator. Vcore part of the regulator is much better designed and therefore BS is not occured under NORMAL conditions. Having C501 ripped off (=worse performance at high power level) the case is different and also Vcore circuit is vulnerable to BS..

If I change the 2 green capacitors with the capacitors from the attached photo
(23 1500 6A) from an old gf4 ti, will the 6800LE overclock higher than with the green caps ?
(I have the AGP albatron 6800LE 128MB ram)

Quote:

---

Originally Posted by santong7

If I change the 2 green capacitors with the capacitors from the attached photo
(23 1500 6A) from an old gf4 ti, will the 6800LE overclock higher than with the green caps ?
(I have the AGP albatron 6800LE 128MB ram)

---

Hi santong7,

Sorry took little bit to answer..

Short answer is: most likely no, propably performance would be worse than before !

The reason for this comment is that I suppose those capacitors are not suitable/better than original 'green' capacitors. I do not know what series those are (I suppose 'CV-AX' by Sanyo), but allmost all capacitors are not BETTER that original 'green' capacitors (as far they are not yet ruined). There are only a few recommended replacement capacitors available, here comes a short list:

- Chemi-con "PXA", PXH", "PXE" ( "PXE" is the best !!)
- Sanyo "SVP"
- Panasonic "WA" (not manufactured any more)
- NIC COMPONENTS CORP "NSP"
- Vishay "94SVP"
- Hitano "EVS"

.. and capacitor in the foto is not any of those (AFAIK).

--

Some questions:

- Do you have any (black screen) problems now ?
- What is your maximum OC'ing frequency for memories and is it limited by 'black screen' ?
- What labels you have top of the 'green' capacitors ?

Recently I purchased some capacitors to test them in С501. I found optimal range 100-220nF. If value less than 100nF, VGA gives audio noise. If value more than 220nF, VGA gives low freq audio noise(crash audio). Unfortunately i purchased only 100,150,220nF capacitors. So I soldered 150nF, and my VGA does not give out sound noise even when scroll in IE and NO BS. But may be 150nF not optimal value... It is necessary to test more values from 100-220nF range. Who can test it on BS affected VGA? :)

Quote:

---

Originally Posted by LLIaJLyH

Recently I purchased some capacitors to test them in С501. I found optimal range 100-220nF. If value less than 100nF, VGA gives audio noise. If value more than 220nF, VGA gives low freq audio noise(crash audio). Unfortunately i purchased only 100,150,220nF capacitors. So I soldered 150nF, and my VGA does not give out sound noise even when scroll in IE and NO BS. But may be 150nF not optimal value... It is necessary to test more values from 100-220nF range. Who can test it on BS affected VGA? :)

---

Very nice experiments :up:

The VGA did not crash/shutdown/BS (=audio was only problem) with all those values, correct ? Since, all (?) 6800le cards have some audio *issues* the stock value might be wrong/unoptimal e.g. 270 uF as I assumed earlier. Therefore, noisy audio might be removed/minimised considerably by changing C501 no matter BS-affected card or not.

I agree, it would be very usefull to make that mod also to BS-affected card..

---

Anybody have/know were to find a 6800le-card with following specs:

- (Completely) BS-affected at stock frequencies
- Still working in some extent (expect BS)
- Moderately cheap

.. I might be interesting to buy (or possibly repair IF I am allowed to test some little risky modifications before hand).

Please, send PM if interested :)

Quote:

---

Originally Posted by SikaRippa

Hi santong7,

Sorry took little bit to answer..

Short answer is: most likely no, propably performance would be worse than before !

The reason for this comment is that I suppose those capacitors are not suitable/better than original 'green' capacitors. I do not know what series those are (I suppose 'CV-AX' by Sanyo), but allmost all capacitors are not BETTER that original 'green' capacitors (as far they are not yet ruined). There are only a few recommended replacement capacitors available, here comes a short list:

- Chemi-con "PXA", PXH", "PXE" ( "PXE" is the best !!)
- Sanyo "SVP"
- Panasonic "WA" (not manufactured any more)
- NIC COMPONENTS CORP "NSP"
- Vishay "94SVP"
- Hitano "EVS"

.. and capacitor in the foto is not any of those (AFAIK).

--

Some questions:

- Do you have any (black screen) problems now ?
- What is your maximum OC'ing frequency for memories and is it limited by 'black screen' ?
- What labels you have top of the 'green' capacitors ?

---

I haven't changed the capacitors.
I have not found any BS probs.
I haven't overclocked it yet just unlocked it

Quote:

---

Originally Posted by santong7

I haven't changed the capacitors.
I have not found any BS probs.
I haven't overclocked it yet just unlocked it

---

OK.. in that case I do not recommend replacing capacitors. However, adding a capacitor between TP9-TP10 COULD give some boost to OC-capability (and/or Vmodding :) )

Capacitor labels of my card :
41/470/6E
45/470/6E

I don't have audio noises on the soundcard output.

However, the vga card itself has some mechanical resonance in some (rare) occasions. I can reproduce this by opening a dos box and do a "dir /s". Perhaps some componentents should be fixed with some silicone-paste or something. I haven't found out which components make the noises.

I have a Gainward 6800 GS 128MB AGP. It recently crashed while running and now the whole computer will not start with the card in the machine. I tried other power supplies. I tried replacing the green caps and a few of the purple caps with replacements but I'm still getting the same problem. Do you think it could be something to do with the ISL6534 chip or would it be something else??

i've installed aditional capacitor with 3300uf, 10v, and i don't have black screen in 3dmark06 at 750mhz, but i have it at 775... my core is 440mhz at 1.5v with zalman fs-v7 fatality cooler, but i 3dmark05 has passed all tests with 440/860 without artifacts and i've score 5135 (drivers set to perfomance)... before extra capacitor, i have black screen at stock speed 300/700...i have gainward 6800le 128mb agp made in taiwan with green caps 43 470 6e.... i've noticed that quake3arena is the most efective to black screen when i run timedemo1 with demo001... do you have any advice how to remove black screen at all? what else can i do with my card? thank you

Quote:

---

Originally Posted by jlovrek

i've installed aditional capacitor with 3300uf, 10v, and i don't have black screen in 3dmark06 at 750mhz, but i have it at 775... my core is 440mhz at 1.5v with zalman fs-v7 fatality cooler, but i 3dmark05 has passed all tests with 440/860 without artifacts and i've score 5135 (drivers set to perfomance)... before extra capacitor, i have black screen at stock speed 300/700...i have gainward 6800le 128mb agp made in taiwan with green caps 43 470 6e.... i've noticed that quake3arena is the most efective to black screen when i run timedemo1 with demo001... do you have any advice how to remove black screen at all? what else can i do with my card? thank you

---

You should be expecting increase of BS very soon. :)
Adding that cap helps just for a short time, after which BS appear even with 300/500. Read precisely first post of this thread and do Ultimate Capacitor Mod!! It will remove BS permanently*.

* - hmm at least few months :cool:

Quote:

---

Originally Posted by santong7

Capacitor labels of my card :
41/470/6E
45/470/6E

---

Thanks for info ! Card with those models have not (yet) reported to be BS-affected (those codes are also quite rare). So, your case also proofed this.

You have different codes top of C136 & C143 capacitors ?

Quote:

---

Originally Posted by Hneel

I don't have audio noises on the soundcard output.

However, the vga card itself has some mechanical resonance in some (rare) occasions. I can reproduce this by opening a dos box and do a "dir /s". Perhaps some componentents should be fixed with some silicone-paste or something. I haven't found out which components make the noises.

---

I wonder the noise coming directly from card is due to heating/cooling-> vibration of some components (coils?) as a function of current consumption variation (=the same fundamental origin that causes noisy audio to speakers)

Quote:

---

Originally Posted by kramer_117

I have a Gainward 6800 GS 128MB AGP. It recently crashed while running and now the whole computer will not start with the card in the machine. I tried other power supplies. I tried replacing the green caps and a few of the purple caps with replacements but I'm still getting the same problem. Do you think it could be something to do with the ISL6534 chip or would it be something else??

---

So, it crashed without any warnings ? And after that (single case) never booted up again ?

- Any (visible) burn components, bad smell?
- Fan is spinning ?
- Computer can be booted with another VGA without problems ?

Quote:

---

Originally Posted by jlovrek

i've installed aditional capacitor with 3300uf, 10v, and i don't have black screen in 3dmark06 at 750mhz, but i have it at 775... my core is 440mhz at 1.5v with zalman fs-v7 fatality cooler, but i 3dmark05 has passed all tests with 440/860 without artifacts and i've score 5135 (drivers set to perfomance)... before extra capacitor, i have black screen at stock speed 300/700...i have gainward 6800le 128mb agp made in taiwan with green caps 43 470 6e.... i've noticed that quake3arena is the most efective to black screen when i run timedemo1 with demo001... do you have any advice how to remove black screen at all? what else can i do with my card? thank you

---

Hi,

Are you 'jlovrek' from 'MVKTech Forums' ;) ?

As 'piaskoon' already said the replacing capacitors might be the only way (easiest) in your case to increase OC-capability. Especially in the case of "4 3 470 6E" -capacitors ((and with MSI)) extra capacitor do not help too long.

In general there are following ways to remove BS (or increase 'BS-free-frequency')

1. Enhancement capacitor modification with GOOD capacitor (extra cap)
2. Ultimate capacitor modification with GOOD capacitors (replacing caps)
3. Feedback modification (changing feedback components for regulator, this one is quite tricky)

--

If you find it hard to get GOOD capacitors, send a PM for me :)

Quote:

---

Originally Posted by piaskoon

* - hmm at least few months :cool:

---

Hoping it will be BS-free for loooooooooooooooooog time:toast:

--

Are you using stock frequencies 24/7 ?

Quote:

---

Originally Posted by SikaRippa

So, it crashed without any warnings ? And after that (single case) never booted up again ?

- Any (visible) burn components, bad smell?
- Fan is spinning ?
- Computer can be booted with another VGA without problems ?

---

Right. No visible damage. No smell. Computer boots fine with other AGP cards. It's Like the card draws so much power that the computer can't turn on the fans won't even turn. Do you think that one of the IC chips got too much current but stopped itself before burning itself? The card will turn it's alarm on if you don't plug in the 4 pin molex connector so I figure it's not dead. My next step would be the other caps that I have not replaced then the tricky part of replacing the small SMD ISL6534 or whatever it is. Do you have any other ideas?? Thanks for the help!

Quote:

---

Originally Posted by SikaRippa

Hi,

Are you 'jlovrek' from 'MVKTech Forums' ;) ?

As 'piaskoon' already said the replacing capacitors might be the only way (easiest) in your case to increase OC-capability. Especially in the case of "4 3 470 6E" -capacitors ((and with MSI)) extra capacitor do not help too long.

In general there are following ways to remove BS (or increase 'BS-free-frequency')

1. Enhancement capacitor modification with GOOD capacitor (extra cap)
2. Ultimate capacitor modification with GOOD capacitors (replacing caps)
3. Feedback modification (changing feedback components for regulator, this one is quite tricky)

--

If you find it hard to get GOOD capacitors, send a PM for me :)

---

yes, i am jlovrek form www.mvktech.net..why? hm, i'm from croatia and i almost didn't find this extra capacitor, but others told me that there is no way to find capacitors like 4 3 470 in my country... so i can't mod my card with ultimate mod... hm, i wonder how someone can have memory at 940mhz? this black screen is really big problem... where can i buy capacitors for ultimate mod? is this solution that will for sure remove bs?

Quote:

---

Originally Posted by kramer_117

Right. No visible damage. No smell. Computer boots fine with other AGP cards. It's Like the card draws so much power that the computer can't turn on the fans won't even turn. Do you think that one of the IC chips got too much current but stopped itself before burning itself? The card will turn it's alarm on if you don't plug in the 4 pin molex connector so I figure it's not dead. My next step would be the other caps that I have not replaced then the tricky part of replacing the small SMD ISL6534 or whatever it is. Do you have any other ideas?? Thanks for the help!

---

Seems that the card really has severe problems. The test 'molex removed -beeps' is intersting since it shows the is not COMPLETELY dead. My guess is some of the IC's have burn and I think you should check all those carefully before replacing any other capacitors.

Take a VERY close look all 'power' IC's (backside of card). You should locate those easily by using the temperature measurement figure below.

http://img401.imageshack.us/img401/7...emp27nx.th.jpg

Any visible damage ? Any shorts over MOSFET pins (note that there can be some nodes internally connected together, check pins using datasheets) ?

Quote:

---

Originally Posted by jlovrek

yes, i am jlovrek form www.mvktech.net..why?

---

Just curious ;)

Quote:

---

Originally Posted by jlovrek

i'm from croatia and i almost didn't find this extra capacitor, but others told me that there is no way to find capacitors like 4 3 470 in my country... so i can't mod my card with ultimate mod... hm, i wonder how someone can have memory at 940mhz? this black screen is really big problem... where can i buy capacitors for ultimate mod? is this solution that will for sure remove bs?

---

Yep. That kind of capacitors are very hard to find. I have not seen a single 'electronic's shop' which have those on their stock - allways they will order capacitors somewhere (main storage e.g. UK, Germany..).

Having clock speed over ~ 900 MHz is a result of following issues:

- Speed of the Hynix-memory chips: 2.2 ns or 2.8 ns (also some other versions rarely appeared). Having 2.2 ns gives better change. You should check label above memory chips to get this info.
- Condition of 'green' capacitors and/or added extra capacitors. Best condition is replaced capacitors to better ones and a good added extra capacitor
- Model of the card (manufacturer/model). There are differences in feedback components (see document for details), and this has big effect when trying to find very maximum OC-frequency. Little bit tricky and more risky mod.

--

Is it risky to give 100 % reliablity (in any case/any problem :) ), but when PROPERLY (!!!) done the cards have been in better condition afterwards. I have heard 1-2 cases when there have been some problems after modding, but it is impossible to say WHY (bad soldering, static electricity, was the card already broken...)

--

Please, sent private message to get some hints about replacement capacitors :)

Quote:

---

Originally Posted by SikaRippa

Hoping it will be BS-free for loooooooooooooooooog time:toast:

--

Are you using stock frequencies 24/7 ?

---

Im using 425/875 for all the time :D

Quote:

---

Originally Posted by piaskoon

Im using 425/875 for all the time :D

---

Very nice values, indeed :up:

Quote:

---

Originally Posted by SikaRippa

Very nice values, indeed :up:

---

the best stable i had was 440/860... now i again wave bs, even on stock speeds... it seems that extra capacitor isn't right solution... damn.... i wanna do ultimate mod!!!

Hello BS-Community,
nice to see that there are more than just few people to have the same problem as me (:p: sounds a bit ironic, I know!)
My card is a MSI 6800le with the second fan design. I bought that card from German Ebay, so the card was giving Blackscreens from the beginning at stock speed. After a large internet research (last year March) I finally found a page which described the capacitor softmod (3300uF, 16V)! But with TP 1 and TP 9!
I bought a cap and after the mod the bs was away. But overclocking the ram over 750Mhz it reappeared. Now having read this awesome forum thread I want to try the ultimate mod.
But the problem: Actually I`m living in Sapporo (Japan) doing the Civil Service (german: "Zivildienst") instead of military. So my Computer is working at the other side of the world in Germany.But at the moment it`s only used for 2d-applications by my brother! :fact: (e.g. Word)

My questions:
-Do you think this side will exist until I`m flying back in June-July this year? I want to try the ultimate mod; but have I understood it right, that the U.m. is perhaps not forever????
-I can`t give any details which green capacitors my card has, but I don`t know where to buy good ones! How about using some from a defect Ati 9600xt?
-Could you post a picture of an soldering iron? Never seen one before! But I know somebody being good at electricial things

Bye,
Nonnenteufel

Quote:

---

Originally Posted by jlovrek

the best stable i had was 440/860... now i again wave bs, even on stock speeds... it seems that extra capacitor isn't right solution... damn.... i wanna do ultimate mod!!!

---

Yep, having BS at stock frequencies or lower = 'the card is ruined'. In that case 'ultimate modification' seems to the best (and maybe only) way to repair the card.

How long it took the card turn from 860 MHz to present condition ?

Hi,

Quote:

---

Originally Posted by Nonnenteufel

Hello BS-Community,
nice to see that there are more than just few people to have the same problem as me (:p: sounds a bit ironic, I know!)

---

Ironic, not at all: I was also somehow quite happy when I find out in the first place that in am not the only one. How many defected cards there are still 'out there', 1000 ,10000, ... ???

Quote:

---

My card is a MSI 6800le with the second fan design. I bought that card from German Ebay, so the card was giving Blackscreens from the beginning at stock speed.

---

This model has been somehow better in respect of BS as the first model (AFAIK). Since, you was not the first user it is not known whether it was BS-affected even when brand new, but since you were looking for a solution in March there is no way card was not older than half year (?). I suppose it has "4 8 470 6E" or "4 9 470 6E" capacitors, but that is not so important since these models are known to be BS-affected (and hard to modify).

Quote:

---

I finally found a page which described the capacitor softmod (3300uF, 16V)! But with TP 1 and TP 9!

---

Yep, that is the first invented mod. Having GOOD capacitor between TP's 10 & 9 gives better electrical performance, BUT many/most cases with MSI-cards that mod does not work too long.

Quote:

---

My questions:
-Do you think this side will exist until I`m flying back in June-July this year? I want to try the ultimate mod; but have I understood it right, that the U.m. is perhaps not forever????
-I can`t give any details which green capacitors my card has, but I don`t know where to buy good ones! How about using some from a defect Ati 9600xt?
-Could you post a picture of an soldering iron? Never seen one before! But I know somebody being good at electricial things

Bye,
Nonnenteufel

---

1. I am sure it will (at least I will be here :toast:)
2&3. Nobody (?) can guarantee any mod will survive for ever. However, many users have made mod several months ago and are still running without problems. The key issue is to use VERY GOOD capacitors. The specifications are listed detailed in post #1. So, in theory you could use old from ATI, but in practice it is extremely unlike you will find proper (and still at good condition !!!)
4. Never see soldering iron ;) I suggest to test soldering **little bit** before trying out 'ultimate mod' or ask friend of your to do it for you. Here comes some fotos:

http://images.google.com/images?clie...-8&sa=N&tab=wi

Quote:

---

Originally Posted by SikaRippa

Yep, having BS at stock frequencies or lower = 'the card is ruined'. In that case 'ultimate modification' seems to the best (and maybe only) way to repair the card.

How long it took the card turn from 860 MHz to present condition ?

---

hm, i think a couple of months... i don't play lot of games, just sometimes run 3dmark to see my score...

btw, now i'm trying to overclock cpu as much as i can.... for now prescott 3.0 is at 3.6 (15*240) with 1.425V, and memory at 480mhz with 2.5-3-3-8... i have titan vanessa s-type cooler...

Quote:

---

Originally Posted by jlovrek

hm, i think a couple of months... i don't play lot of games, just sometimes run 3dmark to see my score...

btw, now i'm trying to overclock cpu as much as i can.... for now prescott 3.0 is at 3.6 (15*240) with 1.425V, and memory at 480mhz with 2.5-3-3-8... i have titan vanessa s-type cooler...

---

This is quite interesting info: some (MSI) cards get ruined in some months even when NOT used for hard 3D-gaming.

hi there,

Just some info on my card, a XFX6800XT marked 15 sept 2005, so it should be one of the latest revisions. I found another posting about this card with simular findings, but two is always better then one right? :D

The BSS started almost right away with quake4, but I can't remeber if that was @ stock speed & settings, probably not.

The frustrating part is it's easy to reproduce BSS, only not @ stock speed, so RMA should be hard if not pointless :(

Ok on to my findings:

325x700 8px4v no BSS
325x800 " BSS in UT2k4 linux amd64 and Ski Alpine Demo 2005
380x700 12px6v occasional BSS in UT2k4 linux amd64
380x740 " ditto
380x800 12px6v BSS in UT2k4 linux amd64 and Ski Alpine Demo 2005

I contacted XFX and they told me to disable AGP FW and SBS and set AGP to 4x. Bastards I thought, just ignore the problem and give standard answers, but:

all AGP FW off (SBA off or AGP 4x was not possible)
325x700 8px4v no BSS
380x700 12px6v no BSS
380x740 " occasional BSS in UT2k4 linux amd64
380x800 12px6v BSS in UT2k4 linux amd64 and Ski Alpine Demo 2005

Interesting ain't it?

Now I also have problems with my HD's (1 IDE and 2 SATA) so I bought a better PSU. I already had a 550W PSU only very very cheap. The new one still is a cheap SWEEX PSU, but it has 6 Molex connectors so at least I could provide my GPU with a dedicated powerline:

AGP FW on again
325x700 8px4v no BSS
380x700 12px6v no BSS
380x740 " no BSS
380x800 12px6v BSS in UT2k4 linux amd64

So I'm not sure if I'm just pushing my card too hard or have the same problem that so many other people with 6800LE cards have. It seems though it has more to do with mem speed then with core speed.

PS I could provide a ss of the caps if that would be helpfull, they're the purple ones.

Quote:

---

Originally Posted by scorpion77

hi there,

Just some info on my card, a XFX6800XT marked 15 sept 2005, so it should be one of the latest revisions. I found another posting about this card with simular findings, but two is always better then one right? :D

The BSS started almost right away with quake4, but I can't remeber if that was @ stock speed & settings, probably not.

The frustrating part is it's easy to reproduce BSS, only not @ stock speed, so RMA should be hard if not pointless :(

Ok on to my findings:

325x700 8px4v no BSS
325x800 " BSS in UT2k4 linux amd64 and Ski Alpine Demo 2005
380x700 12px6v occasional BSS in UT2k4 linux amd64
380x740 " ditto
380x800 12px6v BSS in UT2k4 linux amd64 and Ski Alpine Demo 2005

I contacted XFX and they told me to disable AGP FW and SBS and set AGP to 4x. Bastards I thought, just ignore the problem and give standard answers, but:

all AGP FW off (SBA off or AGP 4x was not possible)
325x700 8px4v no BSS
380x700 12px6v no BSS
380x740 " occasional BSS in UT2k4 linux amd64
380x800 12px6v BSS in UT2k4 linux amd64 and Ski Alpine Demo 2005

Interesting ain't it?

Now I also have problems with my HD's (1 IDE and 2 SATA) so I bought a better PSU. I already had a 550W PSU only very very cheap. The new one still is a cheap SWEEX PSU, but it has 6 Molex connectors so at least I could provide my GPU with a dedicated powerline:

AGP FW on again
325x700 8px4v no BSS
380x700 12px6v no BSS
380x740 " no BSS
380x800 12px6v BSS in UT2k4 linux amd64

So I'm not sure if I'm just pushing my card too hard or have the same problem that so many other people with 6800LE cards have. It seems though it has more to do with mem speed then with core speed.

PS I could provide a ss of the caps if that would be helpfull, they're the purple ones.

---

Hi,

Thank you for detailed report of your experiments, many interesting 'features' :)

--

Yep, too bad not to get EASILY BS at stock frequencies - no way to get it RMA'd :(

--

About XFX's hints about AGP FW etc. That makes sense: those settings lowers memory usage/performance -> lower power consumption -> BS propability decreases. So, that 'trick' has the similar effect as 'Vsync=on' AFAIK. Anyway, you STILL got BS under some conditions (low OC'ing)

They considered that as a workaround ?? They did not give any REASON WHY you should be usÃ:banana:ng those setting ? I think card should be working with ALL standard setting at any normal condition..

--

About power supply. I am quite sure about why BS propability varies between supplies. The current sourcing capability is of course main issue, but the voltage levels varies a little bit and BS propability increases rapidly when molex 5 V level decreases. I confirmed that myself when I was measuring current consumption of a BS affected card (a small shunt resistor in 5 V decreased voltage only ~ 100 mV, but BS propability increased A LOT).

--

XFX 6800xt is rare (still), but seems to have exactly the same features/problems as XFX 6800le (v1 & v2). Your card (and XFX 6800le v2) utilizes 'purple' (Sanyo SVP) capacitors AND still have BS-problems, indicates that XFX has severe problems to manufacture cards ((MSI has also manufactured similar cards 'purple capacitors' & BS-problems as reported earlier in this thread)). I think that this is due to following reasons (and/or):

- Too high temperature is used uring manufacturing (due to RoHs), which deteriorates capacitors (and other components?)
- Feedback components of the feedback are far from optimal values -> 'BS-free-frequency' for memories is very low

At least one user has replaced purple capacitors from BS-affected XFX 6800le-card, and the BS-free frequency increased considerably. Since, he utilized EXACTLY the same models (!!!) the theory about too high temperature during manufacturing could be correct.

--

The codes top of purple capacitors might be interesting. Any change to get foto of your card :) ?

Quote:

---

Originally Posted by SikaRippa

Seems that the card really has severe problems. The test 'molex removed -beeps' is intersting since it shows the is not COMPLETELY dead. My guess is some of the IC's have burn and I think you should check all those carefully before replacing any other capacitors.

Take a VERY close look all 'power' IC's (backside of card).

Any visible damage ? Any shorts over MOSFET pins (note that there can be some nodes internally connected together, check pins using datasheets) ?

---

Ok, after some hard soldering I replaced all the ICs on the back (except the 5 pin mosfet on the bottom, couldn't identify it) Boy, did the ISL6534 suck:mad: . It has a ground on the underside of the chip that is soldered on. Anyway after all that... same problem. I can only try a few more components. I'll try the other caps. Do you have any idea the ic on the front? I couldn't identify that one either. Anybody else heard of/had this problem?? any input is greatly appreciated!:)

Quote:

---

Originally Posted by SikaRippa

Hi,

Thank you for detailed report of your experiments, many interesting 'features' :)

--

Yep, too bad not to get EASILY BS at stock frequencies - no way to get it RMA'd :(

--

About XFX's hints about AGP FW etc. That makes sense: those settings lowers memory usage/performance -> lower power consumption -> BS propability decreases. So, that 'trick' has the similar effect as 'Vsync=on' AFAIK. Anyway, you STILL got BS under some conditions (low OC'ing)

They considered that as a workaround ?? They did not give any REASON WHY you should be usÃ:banana:ng those setting ? I think card should be working with ALL standard setting at any normal condition..

--

About power supply. I am quite sure about why BS propability varies between supplies. The current sourcing capability is of course main issue, but the voltage levels varies a little bit and BS propability increases rapidly when molex 5 V level decreases. I confirmed that myself when I was measuring current consumption of a BS affected card (a small shunt resistor in 5 V decreased voltage only ~ 100 mV, but BS propability increased A LOT).

--

XFX 6800xt is rare (still), but seems to have exactly the same features/problems as XFX 6800le (v1 & v2). Your card (and XFX 6800le v2) utilizes 'purple' (Sanyo SVP) capacitors AND still have BS-problems, indicates that XFX has severe problems to manufacture cards ((MSI has also manufactured similar cards 'purple capacitors' & BS-problems as reported earlier in this thread)). I think that this is due to following reasons (and/or):

- Too high temperature is used uring manufacturing (due to RoHs), which deteriorates capacitors (and other components?)
- Feedback components of the feedback are far from optimal values -> 'BS-free-frequency' for memories is very low

At least one user has replaced purple capacitors from BS-affected XFX 6800le-card, and the BS-free frequency increased considerably. Since, he utilized EXACTLY the same models (!!!) the theory about too high temperature during manufacturing could be correct.

--

The codes top of purple capacitors might be interesting. Any change to get foto of your card :) ?

---

Thank you for your reply, very helpfull. About my new PSU. The funny thing is voltage output is actually worse then with my old one: 12v->11.5/6 5v->+/-4.85, although it seems more stable.

I talked to somebody at the shop yesterday and we agreed I could return it for a better one anytime, this is the best they have:

HIPER Type-R 580

That ought to be stable and enough I think :D

Also I tried to relax my main mem timings a bit (3-3-3-6 @ 333Mhz) and this time it took about 45 mins before I got BSS with UT2k4 linux amd64, which is kind of a record. All it takes btw is some heavy rendering in front of you like a rocket exploding in your face :D

I'll try to relax my cards timings abit to see if that makes any difference. Although after all I read I should probably just RMA the damn thing, especialy after XFX replies. You see I did asked them about BSS and problems with certain cards under high load. They ignored my questions and not only gave the standard answer about AGP FW, SBA and AGP 4x but also told me to buy a new PSU AND MB!!!!!! Now a new PSU makes sense, but simply replacing your MB to see if it makes any difference?? What do they think that I'm from Saudi Arabia or something !??

Anyway things could be worse, at the moment there is only one BSS affected game and its performance sucks anyway (compared to the default windows i386 version: 30fps versus 100+!).

PS attached some foto's, hope they're of some help to you.

h, i have psu everpower ph-500w and my voltages are (when idle):
+3.3v: 3.30v (28A)
+5v: 5.05v (45A)
12v: 11.55 ??? (25A)
+5v standy: 5.20
when my system is load, my cpu voltage drops from 1.425 to 1.32, average is 1.35... why is that? should i buy better psu?

If it's an AMD64 you shouldn't worry about that. It's just AMD Powernow/cool&quiet kicking in and that really is a good thing ;)

no, i have prescott 3.0 which is really hot...but i have titan vanessa s-type and it keep it at idle at 37c, full load 48c at 3.6 with 1.425v....

Quote:

---

Originally Posted by kramer_117

Ok, after some hard soldering I replaced all the ICs on the back (except the 5 pin mosfet on the bottom, couldn't identify it) Boy, did the ISL6534 suck:mad: . It has a ground on the underside of the chip that is soldered on. Anyway after all that... same problem. I can only try a few more components. I'll try the other caps. Do you have any idea the ic on the front? I couldn't identify that one either. Anybody else heard of/had this problem?? any input is greatly appreciated!:)

---

You have been working hard with soldering iron ! I believe ISL6534 was tough one ;). I see, you have quite good supply (shop) for components: which kind of capacitors used for replacement to 'green' capacitors ?

In the front side of the card is quite few IC's for regulator circuitry, maybe only is regulator for VTT/VREF-generation (near purple 330/16 and 1500/2.5 caps).

Have you measured voltages of TP's ? It would be usefull to measure those and compare with values listed in document.

Quote:

---

Originally Posted by scorpion77

Thank you for your reply, very helpfull. About my new PSU. The funny thing is voltage output is actually worse then with my old one: 12v->11.5/6 5v->+/-4.85, although it seems more stable.

I talked to somebody at the shop yesterday and we agreed I could return it for a better one anytime, this is the best they have:

HIPER Type-R 580

That ought to be stable and enough I think :D

Also I tried to relax my main mem timings a bit (3-3-3-6 @ 333Mhz) and this time it took about 45 mins before I got BSS with UT2k4 linux amd64, which is kind of a record. All it takes btw is some heavy rendering in front of you like a rocket exploding in your face :D

I'll try to relax my cards timings abit to see if that makes any difference. Although after all I read I should probably just RMA the damn thing, especialy after XFX replies. You see I did asked them about BSS and problems with certain cards under high load. They ignored my questions and not only gave the standard answer about AGP FW, SBA and AGP 4x but also told me to buy a new PSU AND MB!!!!!! Now a new PSU makes sense, but simply replacing your MB to see if it makes any difference?? What do they think that I'm from Saudi Arabia or something !??

Anyway things could be worse, at the moment there is only one BSS affected game and its performance sucks anyway (compared to the default windows i386 version: 30fps versus 100+!).

PS attached some foto's, hope they're of some help to you.

---

Thanks for fotos :D Some details can be seen e.g. some different capacitor models (black label capacitors), but hard to say whether this has any significant effect..

---

Nice to heard they will RMA card ! I suggest to go for that before they change mind ;) Too bad they had no idea (or just acting like have no idea) about 'black screen'.

I think you should NOT by a new MB in the first place. I consider that just marketing...

---

Quote:

---

12v->11.5/6 5v->+/-4.85, although it seems more stable.

---

Sorry, I did not get ;) What are 5 V values old/new (5/4.85) ?? Are those values got with 'software' or by actual measuring (software values tend to be erroneous sometimes..) Stability is VERY important, in your case that is propably the main reason.

--

About that XFX-card with 'purple' capacitors replaced with 'purple' I told in my last reply. I wonder those new were not EXACTLY the same model, but had capacitance value of 1200 uF instead of 560 uF (???). Maybe that had also significant effect..

Quote:

---

Originally Posted by SikaRippa

Nice to heard they will RMA card ! I suggest to go for that before they change mind ;) Too bad they had no idea (or just acting like have no idea) about 'black screen'.

I think you should NOT by a new MB in the first place. I consider that just marketing...

---

Lol you got me wrong, they did not said I could RMA the card and probably never will. Why would they, so many other parts they can blame it on... :(

The thing is I paid good money for this card, tried really hard too fix it myself and when that didn't work out asked them very polity about it. All they did was give me some standard BS (not BSS ;)) answers.

So at the moment I just want to return the card. The problem is I put a Zalman cooler on the card so I have to carefully remove it and get my hands on some termo conducting tape afterwards. The good thing is the shop I bought it from was very interrested in this BSS 'thing'.

Do you know if the Asus 6800XT (or LE) has this problem? If not that ought to be very interesting.

Quote:

---

Originally Posted by SikaRippa

Sorry, I did not get ;) What are 5 V values old/new (5/4.85) ?? Are those values got with 'software' or by actual measuring (software values tend to be erroneous sometimes..) Stability is VERY important, in your case that is propably the main reason.

---

Not sure about the old 5V values, if IIRC they were the same or worse, but 12V was about 11.9/8 stressed or not. I understood 12V was the main thing for these GPU's, that is incorrect? Everyting is measured in software, I just don't got the balls to attach a voltmeter to a running system ;)

Quote:

---

Originally Posted by SikaRippa

About that XFX-card with 'purple' capacitors replaced with 'purple' I told in my last reply. I wonder those new were not EXACTLY the same model, but had capacitance value of 1200 uF instead of 560 uF (???). Maybe that had also significant effect..

---

I hope you're wrong, because if it's not a manufactioring issue, RMA would be rather pointless wouldn't it.

Hi,
i got a better soldering iron, so i tried to fix the card again, but it doesnt seem to work, call of duty 2 is no problem, but bf vietnam crashes after a few seconds in the menu.
I removed the old capacitors and as much solder as possible and tried new ones (good ones), but they dont seem to work well.

I'll try to solder a big capacitor between tp9 and tp10 later, maybe this helps.

Any suggestions?

Thanks,
Sid

Quote:

---

Originally Posted by scorpion77

Lol you got me wrong, they did not said I could RMA the card and probably never will. Why would they, so many other parts they can blame it on... :(

---

Sorry, I did not read so carefully ;) I think they SHOULD RMA the card and if they refuse you have allways right to RMA card to manufacturer !

Quote:

---

The thing is I paid good money for this card, tried really hard too fix it myself and when that didn't work out asked them very polity about it. All they did was give me some standard BS (not BSS ;)) answers.

---

'Nice' service, but the same kind of 'service' has been seen also e.g. in Finland

Quote:

---

So at the moment I just want to return the card. The problem is I put a Zalman cooler on the card so I have to carefully remove it and get my hands on some termo conducting tape afterwards. The good thing is the shop I bought it from was very interrested in this BSS 'thing'.

---

Since they are even a little bit interesting, I think you might have change to RMA card (if they will ask XFX about 'black screen')

Quote:

---

Do you know if the Asus 6800XT (or LE) has this problem? If not that ought to be very interesting.

---

ASUS 6800 LE is almost the same card as ASUS 6800 XT in respect of card design/layout, only the core differs (AFAIK). Therefore I am 99.99 % sure ASUS 6800 XT is BS-free.

Quote:

---

Not sure about the old 5V values, if IIRC they were the same or worse, but 12V was about 11.9/8 stressed or not. I understood 12V was the main thing for these GPU's, that is incorrect? Everyting is measured in software, I just don't got the balls to attach a voltmeter to a running system ;)

---

OK.. So it can not be verified if the REAL 5 V is higher with the new one.
About measuring: it demands balls (:toast:) only at the FIRST time your start playing with multimeter...

Quote:

---

I hope you're wrong, because if it's not a manufactioring issue, RMA would be rather pointless wouldn't it.

---

I think the 'BS-feature' is manufacturing issue (some kind of) for 99.999+ % sure. In fact the moderator of GW-forum somehow approved that (after I have asked it many times :D ). Some manufacturers (like GW and MSI) have different versions and some of those are working MUCH better (BS do not appear at least during ~ half/one year). The 'XFX-case' is much more mysterious: your case shows that there are QUITE NEW BS-affected XFX-cards !!!

Quote:

---

Originally Posted by Sidolin

Hi,
i got a better soldering iron, so i tried to fix the card again, but it doesnt seem to work, call of duty 2 is no problem, but bf vietnam crashes after a few seconds in the menu.
I removed the old capacitors and as much solder as possible and tried new ones (good ones), but they dont seem to work well.

I'll try to solder a big capacitor between tp9 and tp10 later, maybe this helps.

Any suggestions?

Thanks,
Sid

---

I could not find your foto about soldering (using 56k right now - too slow to check all posts :slobber:). I remember you were using old capacitors and the lenght of the leads was quite long (???).

Do you have now brand new capacitors ? Which series/model ? 'Not working very well now' = only those crashes (=BS??) or some other problems ??

I think adding extra capacitor does not help too much if the 'c143' and 'c136' capacitors are not properly connected (=some severe problems appearing even having good/new 'c143' and 'c136' capacitors)

FYI: I've seen a photo of a Asus 6800XT card with a LE chip and I know my XFX has a LE core too.

But tell me what is more important to these GPU's 5V or 12V? The new PSU I ordered yesterday can be considered highend and at least has seperated 12V rails for the MB and ' the rest'. I have lot' s of hardware (tvcard, IDE drive, 2x SATA, 2X DVD Writer, fancontroller, external USB IDE and lot' s of USB equipment) so that might explain why I have BSS and other not(?).

Quote:

---

Originally Posted by SikaRippa

I could not find your foto about soldering (using 56k right now - too slow to check all posts :slobber:). I remember you were using old capacitors and the lenght of the leads was quite long (???).

Do you have now brand new capacitors ? Which series/model ? 'Not working very well now' = only those crashes (=BS??) or some other problems ??

I think adding extra capacitor does not help too much if the 'c143' and 'c136' capacitors are not properly connected (=some severe problems appearing even having good/new 'c143' and 'c136' capacitors)

---

Yes :)

I didnt use brand new capacitors but some others i ripped off the radeon a few months/weeks ago i haven't used till now.

But today, after resoldering the two capacitors (i removed them, tried to remove some more solder, and then soldered them again checking all of the solder was melted and that their contacts are very near to the contacts on the board) and soldering a bigger capacitor to tp9/tp10 it seems to work, at least in Quake4/Cod2/Serious Sam.
I dont really want to test Battlefield Vietnam or too much overclocking since i'm satisfied with the stock frequencies and somehow i've got the feeling every blackscreen that happens is bad for the condition of the card.

Quote:

---

scorpion77]FYI: I've seen a photo of a Asus 6800XT card with a LE chip and I know my XFX has a LE core too.

---

Thanks for info, something like that I have already assumed: XT is only a new name for LE

Quote:

---

But tell me what is more important to these GPU's 5V or 12V? The new PSU I ordered yesterday can be considered highend and at least has seperated 12V rails for the MB and ' the rest'. I have lot' s of hardware (tvcard, IDE drive, 2x SATA, 2X DVD Writer, fancontroller, external USB IDE and lot' s of USB equipment) so that might explain why I have BSS and other not(?).

---

I forget to answer that one ;)

Regulator circuitry of 6800le/xt cards (GW/MSI/XFX/Leadtek/etc.) utilizes molex 5 V and 12 V lines for memory and GPU (core) voltage geneneration, respectively. Therefore, PSU you (already) ordered should be just fine for GPU, but the most problems (like BS) with this card are due to memory voltage generation (5 V line should be very stabile).

Quote:

---

Originally Posted by Sidolin

Yes :)

I didnt use brand new capacitors but some others i ripped off the radeon a few months/weeks ago i haven't used till now.

But today, after resoldering the two capacitors (i removed them, tried to remove some more solder, and then soldered them again checking all of the solder was melted and that their contacts are very near to the contacts on the board) and soldering a bigger capacitor to tp9/tp10 it seems to work, at least in Quake4/Cod2/Serious Sam.
I dont really want to test Battlefield Vietnam or too much overclocking since i'm satisfied with the stock frequencies and somehow i've got the feeling every blackscreen that happens is bad for the condition of the card.

---

OK.. the problems you have been encountered are most likely related to soldering issues. Also the condition of the old/used capacitors might be poor, because all capacitors have limited life (not too long in the case of electrolytics).

Your card is now working well enough ?? BTW, I still have some extra replacement capacitors (PXA-series = very good), if interested just sent a PM :)

Hi,
i got the capacitors from SikaRippa a few minutes ago :)

Soldering was no big problem this time, everything worked pretty fast, here are some pictures:
http://img210.imageshack.us/img210/9...00706gg.th.jpg
(old capacitors unsoldered)
http://img205.imageshack.us/img205/8...00710mt.th.jpg
http://img205.imageshack.us/img205/9...00728lk.th.jpg

I haven't tested it much by now, but it seems to work fine since i was able to play bf:vietnam with 300/740 for some minutes. Maybe even some more overclocking is possible, but somehow 3dmark doesnt install so i have to try this later. I'm also going to test ut2004 with linux later, but i think i will be no problem.

:banana:

Sid

Whoa, it seems that bs is completely removed! I wasnt able to get bs at all, even 350/785 worked without any problems, maybe i need a vmod because with higher frequencies the overclocking tests fail.
Great!

Before (with original 4 6 470 6E caps): bs even at 300/500
A bit better (caps of a damaged 9800pro): bs sometimes at 300/700
Now (SikaRippas caps): no bs at 350/785!

Jeez this is a long thread lol my eyes look like :zombie:

The thing is I got an le ( duh! ) and got a few questions for you experienced modders/benchers! I been running with 1.5v for a couple off months now without problems but I was wondering how many off you have actually seen a card die because off this? I have a fan blowing over the pcb, I remember reading in this thread about those 3 black chips that control core voltage, well without the fan those get to melting temps. Really no kidding there, I forgot to put the fan back once and within an hour the room smelled real funny. Kind off melted plastic combined with eletrical shortcircuiting smell if you know what I mean. But with the fan blowing over it no smell, and they don't even get warm to the touch. So how quick you people recon it will die? Since everyone seems to think it will kill the card?

Secondly, about the black screens, I can play fear 450/880 without problems but when I bench 06 the sm3/hdr tests black out on me even on stock clocks. Now I read about the capacitors and I don't have any off the bad one's described in this thread and every game I play is just fine besides 3dmark06 sm3/hdr tests :confused:

So what do you guys think? Is it a sign my card is dying? And if so why can I play every game still on my for an le insane core clock without problems since I would expect games to behave the same as 06?

I think I can get even higher clocks on the ram with doing the vmod for it, would :banana: if I could get 1ghz ddr :D

Also, I did read the thread but it was a long read and my brains aches almost as much as my eyes do ;) ... so I'll just ask, is there a way to give my core more/(+stable) volts so I might increase bench clocks?

Anyway this thread rocks :toast:

Quote:

---

Originally Posted by MartianInvader

Jeez this is a long thread lol my eyes look like :zombie:

The thing is I got an le ( duh! ) and got a few questions for you experienced modders/benchers! I been running with 1.5v for a couple off months now without problems but I was wondering how many off you have actually seen a card die because off this? I have a fan blowing over the pcb, I remember reading in this thread about those 3 black chips that control core voltage, well without the fan those get to melting temps. Really no kidding there, I forgot to put the fan back once and within an hour the room smelled real funny. Kind off melted plastic combined with eletrical shortcircuiting smell if you know what I mean. But with the fan blowing over it no smell, and they don't even get warm to the touch. So how quick you people recon it will die? Since everyone seems to think it will kill the card?

Secondly, about the black screens, I can play fear 450/880 without problems but when I bench 06 the sm3/hdr tests black out on me even on stock clocks. Now I read about the capacitors and I don't have any off the bad one's described in this thread and every game I play is just fine besides 3dmark06 sm3/hdr tests :confused:

So what do you guys think? Is it a sign my card is dying? And if so why can I play every game still on my for an le insane core clock without problems since I would expect games to behave the same as 06?

I think I can get even higher clocks on the ram with doing the vmod for it, would :banana: if I could get 1ghz ddr :D

Also, I did read the thread but it was a long read and my brains aches almost as much as my eyes do ;) ... so I'll just ask, is there a way to give my core more/(+stable) volts so I might increase bench clocks?

Anyway this thread rocks :toast:

---

hm, i have zalman fatality and i have core at 445 at 1.5v... on 450 perfomance are droping... the highest known stable frequencie for core on air cooling is 450mhz so that's the highest you can go... about memory.... i have 445/860 for a while, but when i've installed 3dma06 and run tests, black screen has appered also on hdr tests... now the bs frequencie is near stock speeds.... i have extra capacitor, without it, i only could reach stock speads on memory... tomorrow i'm going to do ultimate mod... hm, you must know what type of memory chips you have.... 2.2ns, 2.4 or 2.8ns.... if you have 2.2, the highest you frequencie is 950... but that is on 2.8v, so you should do vmod.... what score have you got on 3dmark05? the highest i had was 5150....

My orb name is me_Westmaas

http://service.futuremark.com/compare?3dm05=1616091

And looks like I got a wr core then since I can bench 460 ;)

I did notice that the sweet spot appears to be between 450-460 since I got lower scores with 460 then on some lower clocks. What do you think about my blackscreens in 06? Since they only appear there and not while playing fear,hl2,hl2 lc, DoD ect it might not be a normal bs problem?

Quote:

---

Originally Posted by MartianInvader

My orb name is me_Westmaas

http://service.futuremark.com/compare?3dm05=1616091

And looks like I got a wr core then since I can bench 460 ;)

I did notice that the sweet spot appears to be between 450-460 since I got lower scores with 460 then on some lower clocks. What do you think about my blackscreens in 06? Since they only appear there and not while playing fear,hl2,hl2 lc, DoD ect it might not be a normal bs problem?

---

ABOUT the core, you went behind it's limits.... you have lower score because of artifacts that you can't see, so, my advice is that you set frequencie back to 445 and than go +1 and see on what frequencie you have the best score... about 3dmark06... i don't know, cause my card was good at 860 in 3dmark05 and other games, but after i have run 3dmark06 and failed few times, my bs frequencie has appered also in 3dmark05 even on 800mhz... so, it taKES TIME for bs to ''grow up''.... now i must do ultimate mod... hm, but maybe the problem is in 3dmark06 cause our card don't support shader 3.0...i think... first and second test are ok... right? we should ask sikarippa for his oppinion...

Quote:

---

Originally Posted by Sidolin

Whoa, it seems that bs is completely removed! I wasnt able to get bs at all, even 350/785 worked without any problems, maybe i need a vmod because with higher frequencies the overclocking tests fail.
Great!

Before (with original 4 6 470 6E caps): bs even at 300/500
A bit better (caps of a damaged 9800pro): bs sometimes at 300/700
Now (SikaRippas caps): no bs at 350/785!

---

Nice you succeeded quite well with modding :up:

So, the soldering quality/old (ruined?) capacitors has to be to the reason your previous mod was not working properly (?).

Hi MartianInvader,

Quote:

---

Originally Posted by MartianInvader

Jeez this is a long thread lol my eyes look like :zombie:

---

Yep, it is quite long ;) However, almost all relevant information is collected to post #1 (except some unposted measurement results & new info obtained from newest posts)

Quote:

---

The thing is I got an le ( duh! ) and got a few questions for you experienced modders/benchers! I been running with 1.5v for a couple off months now without problems but I was wondering how many off you have actually seen a card die because off this? I have a fan blowing over the pcb, I remember reading in this thread about those 3 black chips that control core voltage, well without the fan those get to melting temps. Really no kidding there, I forgot to put the fan back once and within an hour the room smelled real funny. Kind off melted plastic combined with eletrical shortcircuiting smell if you know what I mean. But with the fan blowing over it no smell, and they don't even get warm to the touch. So how quick you people recon it will die? Since everyone seems to think it will kill the card?

---

There have been some (few?) reports in this forum that a high Vcore value (1.4 V ->) has killed card (i.e. those black chips=MOSFET's have burn). Since, many have reported using e.g 1.5 V Core 24/7 without problems, the risk to get chips burn HAS to be related cooling (or are they are using better MOSFET's in some cards ??). If you chips are now cool enough I think 1.5 is OK. BUT higher voltage and current (temperature INSIDE chips) allways make the lifetime of chips shorter. How much - very hard to give any approximations.

Quote:

---

Secondly, about the black screens, I can play fear 450/880 without problems but when I bench 06 the sm3/hdr tests black out on me even on stock clocks. Now I read about the capacitors and I don't have any off the bad one's described in this thread and every game I play is just fine besides 3dmark06 sm3/hdr tests :confused:


So what do you guys think? Is it a sign my card is dying? And if so why can I play every game still on my for an le insane core clock without problems since I would expect games to behave the same as 06?

---

'BS-propability' varies A LOT between programs/games. E.g. almost all BS-affected cards fail with HL2 BUT usually can run Doom 3, although D3 provides quite heavy 3D-load. The difference has to be related to mechanism how to program utilizes memory i.e. the program should be able to write/read a HUGE amount data to video memory in a very short period. More precisely I assume BS happens in the case when memory usage increases extremely rapid manner from idle state to ~100 % load.

Have you tested 'SkiAlpine'-demo, how easily you get BS with that one ?


I think I can get even higher clocks on the ram with doing the vmod for it, would :banana: if I could get 1ghz ddr :D
[/QUOTE]

Some users have been reported quite high values after Vmod e.g. here:
http://www.xtremesystems.org/forums/...ad.php?t=62651

.. the value of course depends on which the memory chips you have as 'jlovrek' already mentioned.

Quote:

---

Also, I did read the thread but it was a long read and my brains aches almost as much as my eyes do ;) ... so I'll just ask, is there a way to give my core more/(+stable) volts so I might increase bench clocks?
Anyway this thread rocks :toast:

---

Quote:

---

I did notice that the sweet spot appears to be between 450-460 since I got lower scores with 460 then on some lower clocks. What do you think about my blackscreens in 06? Since they only appear there and not while playing fear,hl2,hl2 lc, DoD ect it might not be a normal bs problem?

---

I think having core frequency ~450 MHz is very near limit you can ever get, at least it is much higher than a mean value. So, having BS etc. problems within that frequency is quite 'normal'. What memory frequencies you were using in that case ? A high memory frequency is usually dominating factor to get BS.


What brand you card is, GW ? What codes you have top of 'green' capacitors (e.g. "4 3 470 6e") ?

SikaRippa

I'm sure that without the extra fan on the pcb my card would have deid already since the smell I described earlier. And probably it will shorten lifespan, but I didn't plan on keeping the card that long anyway, looking forward to the new dx10 cards so people with X1800/X1900 will sell me them cheap ;)

Ski alpine does not trigger a bs, no matter what clocks. I can do 930 on the ram with that demo but slight artifacts.

Idk ns of ram, got ram hs on them and don't want to remove them atm, I wrote it down somewhere but can't remember where I left that. I think I had 2.6ns but not 100% sure.

Brand I got is Leadtek A400LE tdh ( idk what tdh means? )... BS on 3dmark06 happens with every clock, stock also blacks out that's the strange part! HL2, FEAR, DoD ect all have no problems, but sm3/hdr tests in 06 bs all the time. I can run the HL2 Lost Coast demo with 420-880 ( lower core due to enourmous heat builtup with hdr, also in NFSMW with HDR on temp goes up way faster then without HDR.

I will check back today with the cap numbers, I'm on laptop now ;)

ok guys, i have really good news! i i've replaced caps and installed the one that gave me sikarippa, and everything is OK... i don't have BS at all, even in quake3 when i run /timedemo1, /demo demo001 ... my frequencies are 445/880, after that i've got artifacts in 3dmark05... i have passed all tests in 3dmark05, 3dmark06, quake3arena demos with timedemo function.... usually, i've got black screen in quake3 after 2 seconds on every speed that is higher then stock (i'm talking about memory...) juhhhhuuuu, my card is BS FREEEE! Thank you very much sikarippa.... i'm going to buy some stamps this week and send it to you with old caps... once again, thank you!!!!

Quote:

---

Originally Posted by MartianInvader

SikaRippa
I'm sure that without the extra fan on the pcb my card would have deid already since the smell I described earlier. And probably it will shorten lifespan, but I didn't plan on keeping the card that long anyway, looking forward to the new dx10 cards so people with X1800/X1900 will sell me them cheap ;)

---

Yep, that is true: these cards do not need to work too long - nobody will be interesting about 6800le cards after some years (or even months...). That might be the most important reason why *they* have designed such poor devices.

Quote:

---

Ski alpine does not trigger a bs, no matter what clocks. I can do 930 on the ram with that demo but slight artifacts.

Idk ns of ram, got ram hs on them and don't want to remove them atm, I wrote it down somewhere but can't remember where I left that. I think I had 2.6ns but not 100% sure.

Brand I got is Leadtek A400LE tdh ( idk what tdh means? )... BS on 3dmark06 happens with every clock, stock also blacks out that's the strange part! HL2, FEAR, DoD ect all have no problems, but sm3/hdr tests in 06 bs all the time. I can run the HL2 Lost Coast demo with 420-880 ( lower core due to enourmous heat builtup with hdr, also in NFSMW with HDR on temp goes up way faster then without HDR.

---

It is quite interesting you have problems only with a few cases. Your experiment indicates that 3dmark06/sm3/hdr test could be the HARDEST hardware test so far, thank you for info !!

Quote:

---

I will check back today with the cap numbers, I'm on laptop now ;)

---

Thank you beforehand :) Leadtek cards have been working so far quite well, at least when compared to GW/MSI/.. I have been proposed (all?) these cards will be ruined/BS-affected before end of normal life time of electronic device. The reason why different cards (manufacturers& versions) get ruined more quicky is due to different component characteristics ('green' C136 & C143 capacitors AND feedback component values)

Quote:

---

ok guys, i have really good news! i i've replaced caps and installed the one that gave me sikarippa, and everything is OK... i don't have BS at all, even in quake3 when i run /timedemo1, /demo demo001 ... my frequencies are 445/880, after that i've got artifacts in 3dmark05... i have passed all tests in 3dmark05, 3dmark06, quake3arena demos with timedemo function.... usually, i've got black screen in quake3 after 2 seconds on every speed that is higher then stock (i'm talking about memory...) juhhhhuuuu, my card is BS FREEEE! Thank you very much sikarippa.... i'm going to buy some stamps this week and send it to you with old caps... once again, thank you!!!!

---

Have a nice BS-free gaming :toast:

And thanks beforehand to getting old caps. I have soon quite impressive collecting of ruined CV-EX -capacitors :D

Late update but here goes:

Cap = 47 470 6e
Ram = 2.4ns ( found the piece off paper I wrote this down on :) )

So caps are those which can be affected after time. Must have misread it the first time because I posted I didn't have bs affected caps. Think it explains 06 sm3/hdr tests giving blackout. Still wondering why HL2, DoD, Doom3, CSsource, FEAR ect all run strong? Maybe it's a slow sign of the card dying ( and forcing me to upgrade which on its own isn't bad ;) )

Quote:

---

Originally Posted by MartianInvader

Late update but here goes:

Cap = 47 470 6e
Ram = 2.4ns ( found the piece off paper I wrote this down on :) )

So caps are those which can be affected after time. Must have misread it the first time because I posted I didn't have bs affected caps. Think it explains 06 sm3/hdr tests giving blackout. Still wondering why HL2, DoD, Doom3, CSsource, FEAR ect all run strong? Maybe it's a slow sign of the card dying ( and forcing me to upgrade which on its own isn't bad ;) )

---

i'm wondering how you can have so high clocks on memory? do you have vmod? hm, i think that sikarippa found out that with higher memory frequencies the chance for bs is higher...so, maybe just set down your memory clocks and then try 3dmark06... but fore sure, bs will be more often with the time.... the final result will be: ULTIMATE MOD!!!

Quote:

---

Originally Posted by jlovrek

i'm wondering how you can have so high clocks on memory? do you have vmod? hm, i think that sikarippa found out that with higher memory frequencies the chance for bs is higher...so, maybe just set down your memory clocks and then try 3dmark06... but fore sure, bs will be more often with the time.... the final result will be: ULTIMATE MOD!!!

---

Well it does give slight artefacts above 900 when gaming, but no I did not do the vmod ( yet! am asking for S_A_V to repost his pics off how he did the mod, would be easier for me when I could see them. I got the fan wire and some 20K vr's which I intended to use for mods on my mobo but I can spare them ;) )

And about backing down... I posted earlier that even with 300/700 8/4 the sm3/hdr tests black out :(

Quote:

---

Originally Posted by MartianInvader

Late update but here goes:

Cap = 47 470 6e
Ram = 2.4ns ( found the piece off paper I wrote this down on :) )

So caps are those which can be affected after time. Must have misread it the first time because I posted I didn't have bs affected caps. Think it explains 06 sm3/hdr tests giving blackout. Still wondering why HL2, DoD, Doom3, CSsource, FEAR ect all run strong? Maybe it's a slow sign of the card dying ( and forcing me to upgrade which on its own isn't bad ;) )

---

Thanks for info !

2.4 ns memories, sure ? There are 2, 2.2, 2.5, 2.8 ns memories, but I have not heard 2.4 ns bofore (maybe new version?).

Something is wrong indeed with that Leadtek version you have ("4 7 470 6E"). In the case of having artefacts at high OC-level the performance is memory speed limited (Vmem mod might help) not the regulator of the card (=causes BS). So, getting artefacts more easily than BS indicates card is not badly BS-affected. That could explain why you can run HL2 etc. without problems. Maybe 3dmark06 sm3/hdr test is the most hardest test so far - and therefore gives BS also in your case. I'll try out 3dmark06 myself later ((My CPU is soooo slow that propably I am not able to run the mark at all :( ))

Quote:

---

Well it does give slight artefacts above 900 when gaming, but no I did not do the vmod ( yet! am asking for S_A_V to repost his pics off how he did the mod, would be easier for me when I could see them. I got the fan wire and some 20K vr's which I intended to use for mods on my mobo but I can spare them )

And about backing down... I posted earlier that even with 300/700 8/4 the sm3/hdr tests black out

---

I replied already to that thread ;)

Probably would be 2.5ns then can't read my own handwriting :S

Still benchable 950 with arti's but hoping for more tomorrow after vmod :D

Quote:

---

Originally Posted by MartianInvader

Probably would be 2.5ns then can't read my own handwriting :S

Still benchable 950 with arti's but hoping for more tomorrow after vmod :D

---

can you post your bios? sikarippa, do you have some bios with relaxed timings cause the highest clock i can achieve is 870 withous artifacts...

Quote:

---

Originally Posted by jlovrek

can you post your bios? sikarippa, do you have some bios with relaxed timings cause the highest clock i can achieve is 870 withous artifacts...

---

Do you want the original bios or the modded one I am using atm? I am running 1.5vcore with modded bios, also bios clocks are set to 400/800, removing hassle to oc when transfering mobo's ( rt used for higher oc's )?

Must notice that with relaxed timings it will not help you when you got a few extra mhz on the ram imho, the timings will rule out all gain. Btw, I don't think you could use my bios unless you also got a leadtek A400le tdh.

Yeah, with 0.05V higher vmem i can get up to 890mhz without Bs, only the core doesnt go any higher than 350. Any hints for the core?

Btw. how many stamps do i need from germany to findland? I want to send sikarippa my old capacitors but i dont know how much it is :)

I'm doing the vmem mod now, but it isn't working ( eg turning vr knob doesn't change vref readout must be bad contacts either tp13 or tp14 :( )

Quote:

---

Originally Posted by MartianInvader

Do you want the original bios or the modded one I am using atm? I am running 1.5vcore with modded bios, also bios clocks are set to 400/800, removing hassle to oc when transfering mobo's ( rt used for higher oc's )?

Must notice that with relaxed timings it will not help you when you got a few extra mhz on the ram imho, the timings will rule out all gain. Btw, I don't think you could use my bios unless you also got a leadtek A400le tdh.

---

both of them... or just right me the timings you have... and what mem clocks you can have without artifacts...?

Artifact free depends on what I'm running. In benching it's around 900 depending on temps, gaming is around 880 100% ( even for hours gaming it's still good ). Atleast these are my day to day settings. And yes 900 without artifacts mean 950 while benching looks... pixelated ;)

I mostly play 440/900, but if after some hours I see some arti's I back down ram to 880 and it's fine after that.

I will post timings when I'm done doing vmem mod, is giving me allot off pain. It seems that no matter what I turn my vr I always get 3.32vmem with vr in place, when I take it out it goes back to 2.74vmem ( stock! )... Wondering I can bench 3.32vmem cause off active cooling and cold weather today so can see also if bs goes away in 06 with more volts.

ok almost killed the card with 3.3Xv :(

Was alright as desktop but when switching to 3d instantly screen got garbeld up real bad. Reset pc, didn't boot. Took out vc, out in 9600pro booted fine. Put vc in old sempron rig, didn't boot. Thought last chanche, put back in A64 rig and booted fine. Treid some nfsmw, screen looked horrible quit game, check settings and mem was still at 980mhz, took down to 900 and started game again and everything was fine.

However, I think I might have killed my max oc because I got a bs on nfsmw at 920... only one so far but still bad because only bs I had before was with 3dmark06.

@jlovrek

I would give you my bios but it's a diffrent bios ( manufacturer ) so you would be stupid if you'd use it. I can post a screenshot off the timings with nibitor tomorrow but not today to bussy and on the main rig not internet atm so can't dl nibitor :)

Quote:

---

Originally Posted by jlovrek

can you post your bios? sikarippa, do you have some bios with relaxed timings cause the highest clock i can achieve is 870 withous artifacts...

---

I am using default bios, so I wonder no need to post it.. Anyway, I think timing setting tweaking can not be used to remove artefacts, but get some boost in the case (frequencies) memories still working properly. What is the speed of you memory chips ? Maybe 870 MHz is (hard) limit of your chips, Vmem mod *might* help little bit.

A different supply voltage is used/suggested for the memory chips having different speed specs (page 3):

http://www.hynix.com/datasheet/pdf/d...F(Rev.0.7).pdf

.. e.g. 2.8 V for 450 MHz (=> 900 MHz)

Quote:

---

Originally Posted by MartianInvader

ok almost killed the card with 3.3Xv :(

Was alright as desktop but when switching to 3d instantly screen got garbeld up real bad. Reset pc, didn't boot. Took out vc, out in 9600pro booted fine. Put vc in old sempron rig, didn't boot. Thought last chanche, put back in A64 rig and booted fine. Treid some nfsmw, screen looked horrible quit game, check settings and mem was still at 980mhz, took down to 900 and started game again and everything was fine.

However, I think I might have killed my max oc because I got a bs on nfsmw at 920... only one so far but still bad because only bs I had before was with 3dmark06.

@jlovrek

I would give you my bios but it's a diffrent bios ( manufacturer ) so you would be stupid if you'd use it. I can post a screenshot off the timings with nibitor tomorrow but not today to bussy and on the main rig not internet atm so can't dl nibitor :)

---

There is something terribly wrong with mod when you get Vmem of 3.3 V :slobber: I suggest to remove mod ASAP, since that voltage level is dangerous to card !!

What is resistance you have above you Vres ? ((How about resistance between TP's when Vres removed ?)) Having Vmem of 3.3 V is obtained with very low value of resistance (much lower than ~ 1 kilo ohm). Is it possible that you have made a short by mistake somewhere in PCB ?

Quote:

---

Originally Posted by Sidolin

Yeah, with 0.05V higher vmem i can get up to 890mhz without Bs, only the core doesnt go any higher than 350. Any hints for the core?

Btw. how many stamps do i need from germany to findland? I want to send sikarippa my old capacitors but i dont know how much it is :)

---

Core OC'ing capability might be also increased with Vmod (by bios), but take care of cooling the MOSFET's !!

--

From Finland to Germany (or everywhere to Europe) the postage cost is exactly the same as in side Finland (in the case of letter) i.e. minimum postage cost less than 1 euro. Thank you before hand :)

Quote:

---

Originally Posted by SikaRippa

There is something terribly wrong with mod when you get Vmem of 3.3 V :slobber: I suggest to remove mod ASAP, since that voltage level is dangerous to card !!

What is resistance you have above you Vres ? ((How about resistance between TP's when Vres removed ?)) Having Vmem of 3.3 V is obtained with very low value of resistance (much lower than ~ 1 kilo ohm). Is it possible that you have made a short by mistake somewhere in PCB ?

---

Mod removed already, going to measure resitance off vr's I got, maybe I killed them all before doing the mod. No short in pcb, very sure off that because with mod in place but without vr in fanconnector vmem is back to 2.73v so definetly something wrong with either vr, or inside fanconnector there is a fault ( short? ) when I put the vr in there :(

Ok mod worked fine just I got a 20ohm vr instead off 20Kohm.

SikkaRippa do you still have those caps for the ultimate mod? Very interested in them because I'm hoping it will prevent my card from dying so slowly ;)

Pm me with shipping costs ect if you got them :)

Quote:

---

Originally Posted by MartianInvader

Ok mod worked fine just I got a 20ohm vr instead off 20Kohm.

SikkaRippa do you still have those caps for the ultimate mod? Very interested in them because I'm hoping it will prevent my card from dying so slowly ;)

Pm me with shipping costs ect if you got them :)

---

Nice you got Vmod finally working ! How high level you set voltage?

--

I have still a few capacitors left, and I can send you a pair. Just send PM for me (your address..). However, I think in your case changing capacitors is not needed (at least yet). Your BS-free frequency is quite high and BS appears rarely (only with 3dmark06 ???). So, it is possible you will not gain too much with new capacitors. In fact there is (tiny) risk to get even more poor performance because of desoldering/soldering of the capacitors might fail..

I suggest to first try out adding extra capacitors between TP's and check out if you still get BS with 3dmark06.

--

I downloaded 3dmark06 and tested myself. I did NOT get BS at stock frequencies with GW 6800le (version 1b). My processor is not very fast (XP 2400+) which makes score quite low. That might be one reason why I did not get BS: slow processor is not capable to fetch data in a fast manner. What prosessor you have ?

Venice 2.7ghz something newer then you got but still not fast ( enough lol )

Vmod went fine this time I got pics but very bad quality so no pics yet I will post some later when I will clean the case out so I can do some more modding to my board. Also I measured actual gp uand it was still 1.1v! So I remodded the bios voltage tabel and are now at a actual 1.4v. Benchable on 2k1 470/970 1.4v/2.9v. I'm currently doing a cleanup on my win2k install because with these clocks and 500mhz cpu increase but no major tweaking my score was actually lower in some cases because off drivers and dx version but I'm hoping to gain some ranking maybe on the orb :)

PM sent, I want to do the full mod think it will allow a much higher stable overclock. With 2.9v sometimes I got to startup my pc twice or so because off no mem voltage I think this is caused by capacitators dying out on me.

Allright I finally got my Hiper PSU and believe it or not but the BS's are gone!

Power levels are 12.0 5.05 with this baby and I need Asus utilty (3 digits after comma) to notice any fluctuations. Aparently this xfx xt card (possibly other too)really likes a high end PSU, I just hope it stays this way and my card isn't damaged already.

Oh jeah some stats to supplement the previous ones:

UT2k4 linux amd64:

380x800 no prob
423x817 no prob (the highest the selftest would allow me, before becoming consistent in failing)

Will test some more in Windows later on.

Sorry, answers take toooo long - *little* busy in real life :(

Quote:

---

Originally Posted by MartianInvader

Venice 2.7ghz something newer then you got but still not fast ( enough lol )

---

Have ever seen (or anybody, ever) a TOO fast computer :D

Quote:

---

Originally Posted by MartianInvader

Vmod went fine this time I got pics but very bad quality so no pics yet I will post some later when I will clean the case out so I can do some more modding to my board. Also I measured actual gp uand it was still 1.1v! So I remodded the bios voltage tabel and are now at a actual 1.4v. Benchable on 2k1 470/970 1.4v/2.9v. I'm currently doing a cleanup on my win2k install because with these clocks and 500mhz cpu increase but no major tweaking my score was actually lower in some cases because off drivers and dx version but I'm hoping to gain some ranking maybe on the orb :)

---

OK.. Bios Vmodding seems to have problems in many cases = only way to verify voltage level is physical voltage measurement.

Quote:

---

PM sent, I want to do the full mod think it will allow a much higher stable overclock. With 2.9v sometimes I got to startup my pc twice or so because off no mem voltage I think this is caused by capacitators dying out on me.

---

I have sent caps for your, I'll hope your mod will be success :) I wonder your capacitors are not yet fully ruined since you can have such high frequencies etc. Maybe windows clean-up will/would help a lot to startup etc problems (?).

Quote:

---

Originally Posted by scorpion77

Allright I finally got my Hiper PSU and believe it or not but the BS's are gone!

Power levels are 12.0 5.05 with this baby and I need Asus utilty (3 digits after comma) to notice any fluctuations. Aparently this xfx xt card (possibly other too)really likes a high end PSU, I just hope it stays this way and my card isn't damaged already.

Oh jeah some stats to supplement the previous ones:

UT2k4 linux amd64:

380x800 no prob
423x817 no prob (the highest the selftest would allow me, before becoming consistent in failing)

Will test some more in Windows later o !!n.

---

Nice you got your card working better :up: A better PSU is a good solution to minimize BS-problems (even prevent in some cases).

--

At least some cards get ruined more and more when old, but there are so far too little info about XFX 6800XT cards to say anything for sure. The purple capacitors (Sanyo SVP's) are much better than 'green' (CV-EX', so the deterioration rate has to be much lower.

In the case any problems/BS-performance decrease in future, please post results !!

Hi all,
firstly i want to apologize for english that will be used further, then i want to thank all who participate on this BS problem, special thx to SikaRippa, Bilabong and Zadah.:woot:

I bought my card before one year, it's specs are:
GW 6800LE
caps 4 7 470 6E
made in China
codes are identical to card version 1b
Problem was quite a typical:
-BS after several hours in 3D mode (almost immediately in HL2).
-3Dmark03 passed on default memory frequency but not on higher freqs
-3Dmark05 never passed on defaults
-BS was not dependant on core freqs
-BS underclocking mem freqs wasn't solution because of large artifacts
! I'm ruing the theory that versin 1b is non BS !
I decided not to RMA and rather do some enhancements because of good probability to "upgrade" this card to 6800GT. According to Zadah's and Bilabong's suggestion i made:
-simple cap mod (TP1-GND) with two parallel 4700uF caps
-bought stronger PSU (voltages under full load never drops more than 1%)
Effect:
-3Dmark05 passed
-BS occured ocassionaly after hours of playing HL2
So i decided to test further and made:
-opened rest of shaders to 16v6 (no artifacts at 350/700)
-flashed to Aopen 6800GT
Card was realy stable, but in the last months BS occured again and more and more frequently, so i decided to make:
-advanced cap mod (1000uF between TP13-TP14)
-higher mem voltage by 0,5V (6k8 resistor between TP9-TP10)
Card is more stable than with simple cap mod (tested only one day).
-3Dmark05 passed withou problems at 350/700 but not at 400/800
-Aquamark passed even at 400/800
-played 6 straight hours of WoW without BS
-audible fps-dependant noises at higher than 350/700 freqs are present

That was my situation, but finaly i want to solve this problem with ultimate cap mod. BUT there is a problem, in our country it's impossible to buy on shop (even e-shop) such quality caps in small lots. I can only hope that any good soul will be able to sell me 2-4 of these caps.

I have to ask SikaRippa or anyone else, can you send me couple or more of these quality caps. I'll repay you for sure ;)

Quote:

---

Originally Posted by Mula

Hi all,
firstly i want to apologize for english that will be used further, then i want to thank all who participate on this BS problem, special thx to SikaRippa, Bilabong and Zadah.:woot:

I bought my card before one year, it's specs are:
GW 6800LE
caps 4 7 470 6E
made in China
codes are identical to card version 1b
Problem was quite a typical:
-BS after several hours in 3D mode (almost immediately in HL2).
-3Dmark03 passed on default memory frequency but not on higher freqs
-3Dmark05 never passed on defaults
-BS was not dependant on core freqs
-BS underclocking mem freqs wasn't solution because of large artifacts
! I'm ruing the theory that versin 1b is non BS !
I decided not to RMA and rather do some enhancements because of good probability to "upgrade" this card to 6800GT. According to Zadah's and Bilabong's suggestion i made:
-simple cap mod (TP1-GND) with two parallel 4700uF caps
-bought stronger PSU (voltages under full load never drops more than 1%)
Effect:
-3Dmark05 passed
-BS occured ocassionaly after hours of playing HL2
So i decided to test further and made:
-opened rest of shaders to 16v6 (no artifacts at 350/700)
-flashed to Aopen 6800GT
Card was realy stable, but in the last months BS occured again and more and more frequently, so i decided to make:
-advanced cap mod (1000uF between TP13-TP14)
-higher mem voltage by 0,5V (6k8 resistor between TP9-TP10)
Card is more stable than with simple cap mod (tested only one day).
-3Dmark05 passed withou problems at 350/700 but not at 400/800
-Aquamark passed even at 400/800
-played 6 straight hours of WoW without BS
-audible fps-dependant noises at higher than 350/700 freqs are present

That was my situation, but finaly i want to solve this problem with ultimate cap mod. BUT there is a problem, in our country it's impossible to buy on shop (even e-shop) such quality caps in small lots. I can only hope that any good soul will be able to sell me 2-4 of these caps.

I have to ask SikaRippa or anyone else, can you send me couple or more of these quality caps. I'll repay you for sure ;)

---

Hi,

I wonder you have been 'dealing' with the problem quite long time, since your remember Bila and Zadah. Did you also take part to that BS-problem thread in the GW-forum (which has already been destroyed by GW) ?

--

'Version 1b' has been working BETTER than many other version BUT it is now 100 % sure some of those are also BS-affected when old. The first BS-case with 'version 1b' was reported in last summer (!) it was quite inaccurate. Only a few days ago I got exact codes about BS-affected card having "4 7 470 6E" capacitors i.e. 'version 1b'. That card has identical coding to the card shown in document, so I will modify document (hopefully soon..) with this new PROOFED info + many other things :)

--

My capacitor 'supply' is running out :( but I still have one extra pair :) Send a PM with address, so we can arrange this.

--

I know how and where can those capacitors can be obtained. Problem is a large minimum amount of capacitors they sent/provide and high postage cost. Therefore only reasonable way to get those capacitors would be (for somebody) to buy a whole reel (!!) and send samples to users having BS-affected cards. (((In theory I could do it but how many capacitors will be ever requested 10 ? 20 ? 100 ? )))

Quote:

---

Originally Posted by SikaRippa

Hi,

I wonder you have been 'dealing' with the problem quite long time, since your remember Bila and Zadah. Did you also take part to that BS-problem thread in the GW-forum (which has already been destroyed by GW) ?

---

Yeah i remember very well, GW forum was the only source of information for me. I still have that destroyed site in my favorites, just to remember GW behavior.. ::mad:

Quote:

---

Originally Posted by SikaRippa

I know how and where can those capacitors can be obtained. Problem is a large minimum amount of capacitors they sent/provide and high postage cost. Therefore only reasonable way to get those capacitors would be (for somebody) to buy a whole reel (!!) and send samples to users having BS-affected cards. (((In theory I could do it but how many capacitors will be ever requested 10 ? 20 ? 100 ? )))

---

But it looks like the only way how to obtain these caps...
I'm still trying to contact any electronics repair center and ask them for these caps.

I got the caps sikkarippa you the man :banana:
When I saw the envelop I thought wtf is this, I openend it and again wtf is this but then I rememberd these :)

Will try to get some good pics this time around ;)

Quote:

---

Originally Posted by Mula

Yeah i remember very well, GW forum was the only source of information for me. I still have that destroyed site in my favorites, just to remember GW behavior.. ::mad:

---

Me too :D (have that still bookmark'ed). Did you noticed that there was FOUR threads considering BS-issue in GW's forum and *THEM* deleted those all... In one thread (yes I saved that just before :D :D ) the moderator confirmed BS is ' a design flaw' :

Quote:

---

Originally Posted by GW's moderator

The current revision of exchanged LE's are indeed fixed, the problem occured thanks to a somewhat erronerous BOM once upon a time; thus generating issues further down the road, this has been acknowledged and dealt with.

---

That happened ~ may of last year, I wonder whether they have dealt the problem :( .. or deal = forget in GW's language !!!!

Quote:

---

But it looks like the only way how to obtain these caps...
I'm still trying to contact any electronics repair center and ask them for these caps.

---

Sent PM for you about capacitors.

Quote:

---

Originally Posted by MartianInvader

I got the caps sikkarippa you the man :banana:
When I saw the envelop I thought wtf is this, I openend it and again wtf is this but then I rememberd these :)

Will try to get some good pics this time around ;)

---

:toast: Have you already tried out those ?

Sorry been very slow lately but will do it this week ;)

Hi! I'm new here and my English isn't so good.

I encountered also this not funny problem with my Leadtek Winfast A400LE TDH Rev. A2, an used one. The Green Capacitors on that card are labelled: 46 - 470 - 6E.
The seller described it as fine working with 12/5. I don't know whether I can believe this or not. It gave Black screens from the the day I put it into my computer (these ones where the screen goes black like standby the power consumption of the hole machine rises, boot up again only possible by holding the Power Button down a few seconds.
First I tested with 3DMark03, and my favourite game Guild Wars, I tried to fix the problem with several nvidia drivers and the rivatuner. It gave me only few seconds more ingame. First thing I suspected was the Power supply because of my hungry Prescott 3.2 GHz. Weird thing with my 300W noname Power supply it ran best and occasionally I could finish a 3DMark03. My 400W and 550W Power Supply caused earlier BS. From my side no OC of GPU and RAM has been done, I only tried to unlock further pixelpipelines and shaders with
the riva tuner.

After 2 weeks of trial and error with other components and a lot of frustration on my side I found Your interesting article yesterday. Immediately I desoldered a few Capacitors I.Q. 1000µF from an broken CUSL-2 Mainboard because I found a similar Capacitor right in the first picture of your article. I attached the 1000µF/ 6.3V/ 105°C carefully to TP9 and TP 10 on my card and in order to hold it in place I fixed it with glue tape for testing.

http://people.freenet.de/japanblue/Kondi-Mod1.JPG

After that modification I played Guild Wars whole evening and no BS occured, looks like the 6800LE is working fine now. Interesting thing would be if the opening of pipes or shaders increases the chance of destroying the capacitors and how long does this modification work until the card ist suffering from BS again?
The capacitor is soldered now and the card still working fine.

Thx a lot for this discussion.

it's working for now, but with the time, bs will be back... maybe in the month or two, maybe in the half of year... ultimate mod is solution for ever (that's the way it looks for now)...

Hi Guppifil !!

Thank you for comments and providing new info :)

--

Leadtek-cards have been so far better that e.g. GW & MSI-cards. However, during last few weeks several case have been reported. For instance in Finish forum was reposted BS-affected Leadtek card ("4 9 470 6E", Rev:A, Assy rev:C, thx Tiputtaja !!). I have proposed that ALL 6800(le)-card using this PCB-design are propably BS-affected when old. The reports with Leadtek-card seems to confirm that assumption: Leadtek's are also BS-affected but mean-time-of-failure is longer.

--

Extra capacitor modification may fail when old, as jlovrek already said. Maybe you can run long time without problem, maybe not. Some models (e.g. "4 3 470 6E" GW or MSI) are usually working a very short period with the extra capacitor. Certain models with a GOOD added capacitor have been working so far quite long time (year).

--

Interesting info about PSU's :) I have demonstrated myself that lower 5 V voltage value increases BS-propability (in addition to unstable level = poor supply). Could you check what voltage level you have in 5 V line in your 300 W and 550 W (400W) supplies ? That might give solid explanation why 300 W provides the best performance.

Hey SikaRippa,

long time no see,

I've recently done the Vmem mod again, after getting a little itchy to do some volt mods (TP13-TP14). Also, for a long while, pretty much regardless of clock speeds, memory seems to cause problems.

XFX 6800 Rev2 16/6 + NV5
Autodetect is quite low. I flashed very loose timings, and 1.4V (1.33V measured). Although I can set, and run 3DMark2005 as high as 414/950 it occasionally 5-30min "glitches". During benchmark or whatever 3d, the screen flickers, and suddenly its like its running at 75% speed. I have forced performance level enabled, and fastwrites off (enabling even at default clock speeds = havoc)

After voltmod (default 2.72V), 2.85V, max on memory still about the same... up to 978 in ATItool for 5min+ without artifacts, BUT internal driver test, coolbits or rivatuner fails even at 790.

Could it be wrong Vddq, or other memory voltage parameter. I tried capacitor mod before but had very limited results... not sure what the problem could be. I could measure whatever points you need....

Hello.
I am using AXLE 6800XT AGP card.
My card sometimes get BS, so I tried Enhanced capacitor modification
with 10V 2200uF ESR caps. It worked and never BS or other hang.
But the parts named R47, and 3 voltage chips are so hot. I can't touch
these chips even though cooling with 12cm FAN beside. The plastic base
of purple color caps near the R47 parts are melted little. But I can't
remember these parts get so hot because of Caps mod or from bought the
card. What is the problem? Thank you.

Quote:

---

Originally Posted by ***Deimos***

Hey SikaRippa,

long time no see,

I've recently done the Vmem mod again, after getting a little itchy to do some volt mods (TP13-TP14). Also, for a long while, pretty much regardless of clock speeds, memory seems to cause problems.

XFX 6800 Rev2 16/6 + NV5
Autodetect is quite low. I flashed very loose timings, and 1.4V (1.33V measured). Although I can set, and run 3DMark2005 as high as 414/950 it occasionally 5-30min "glitches". During benchmark or whatever 3d, the screen flickers, and suddenly its like its running at 75% speed. I have forced performance level enabled, and fastwrites off (enabling even at default clock speeds = havoc)

After voltmod (default 2.72V), 2.85V, max on memory still about the same... up to 978 in ATItool for 5min+ without artifacts, BUT internal driver test, coolbits or rivatuner fails even at 790.

Could it be wrong Vddq, or other memory voltage parameter. I tried capacitor mod before but had very limited results... not sure what the problem could be. I could measure whatever points you need....

---

Hi,

Nice to *see* you again..

Yep, I remember your case (vmodding, capacitor did not help too much/at all). It quite hard to say anything wise (or give killer solutions) to solve your problems. Some comments/thoughts. There is only one memory voltage in this card (i.e. no seperated VDD VDDQ) so there no room to play with highly optimized VDD/VDDQ ratio. However, playing with Vmemory and Vcore ratio might have some effect (???). Capacitor mod is (designed) to overcome BS-feature, so when having high (~900 MHz) memory OC-capablity by defualt no extra gain can not be expected. The idea with the mod is to compensate the poor performance of the ruined (too 'weak') original capacitors.

It is possible those glitches are due to throttling i.e. overheating ? What is your OC-performance having default pipes ?

--

Thank you for info about your card version and measured value of "1.4 V" :up: I have assumed that bios settings can not set Vcore in a linear manner, and that somehow confirms this. Since, some of the XFX LE and XT -cards (even with "purple" capacitors !!) are BS-affected, it would be nice to heard whether there is any other coding/version number in addition to "rev2" ?

Quote:

---

Originally Posted by dkr09

Hello.
I am using AXLE 6800XT AGP card.
My card sometimes get BS, so I tried Enhanced capacitor modification
with 10V 2200uF ESR caps. It worked and never BS or other hang.
But the parts named R47, and 3 voltage chips are so hot. I can't touch
these chips even though cooling with 12cm FAN beside. The plastic base
of purple color caps near the R47 parts are melted little. But I can't
remember these parts get so hot because of Caps mod or from bought the
card. What is the problem? Thank you.

---

Hi,

Thank you providing new info :up: I have never before heard such manufacturer "AXLE" !! Is your card similar to this one:

http://www.axle3d.com/i22.htm

Could you specify which code you have top of "green capacitors" e.g. "4 3 470 6E" ?

--

The high temperature of components is **feature** of this card model. I wonder the capacitor mod do not have any affect to temperatures, maybe mod I have measures temperatures of the components:

http://img401.imageshack.us/img401/7...emp27nx.th.jpg

In that case temperature of "R47" was 62 C. The R47 is an inductor providing current for core (part of regulator) and high temperature means extremely high power (=current) through that device. Have you increased core voltage ? A too high Vcore (high temperature) may cause several problems for instance components could burn.

auto-detect with 12/5 and 16/6 is about the same. top 3dmark speed also same. Likewise, it was at similar level before I installed NV5. Although increasing Vmem increased top memory speed before artifacting in rthdribl or atitool, autodetect still same or even lower. I think there may be some damage on the card. When I made a mess with some AS5 one time, I used contact cleaner to clean it up. I sprayed it everywhere, and made puddles of AS5 that even went under the chips (GPU, mem etc).. I sprayed a lot, until it was nearly clear... and it vaporizes fast. Dont know how big issue AS5 residue could be.

I will try the capacitor mod again, because last time the legs to the capacitor were too long. I will post some capacitors I have. I'm not sure which has lowest ESR.. I think they are all hybrid, 6.3V-10V and 1000uF-6000uF...

I just thought of something!
Typically these glitches happen when it starts a 3dMark game test, ie change in load on GPU/MEM. Perhaps there is cold-solder and poor contact somewhere, or voltage regulators cant regulate voltage properly. It really "feels" like something is throttling.

Also, to elaborate, it happens at 350/800, 350/960, as well as 410/800 and 410/960... so I didn't think power supply or heat are the issue but you never know. I'm still trying to figure out what kind of shims to put uner the NV5 because not perfect contact with the mem chips... back of the mem chips feels only lighty warm during benchmarking.

Thank you for your reply, SikaRippa.
Thanks read my post and give an advice for me in spite of my poor english.

My card seems to made that site,
and almost same that card linked without
cooler and some parts. I think it is OEM products.
The value of green capacitor is 40 470 6E.
I bought this card a week ago and the cards get BS from when I bought.

I tried plug in and out the card several time for test. Changed VGA and
motherboard and power and so on. Now the R47 parts and 3 chips are
not so hot with additional 12cm cooling fan with composed first PC parts.
I don't know why the situation was happened. I don't have tester so
I checked core voltage with Rivatunner monitoring program, but It says
core vid is 2.0v. It is incorrect value, isn't it? Another report tab,
the voltage was 1.2v on the BIOS info.

Anyway all is good.
My card driving with 16/6 pipe and 400/800 clocks without any problem.
And BS situation disappeard completely.
I am very happy. The cap mod was very helpful for me.
Thanks and Good luck SikaRippa!