I updated some missing links for figures, those figures were DISAPPEARED from 'imageshack', wierd...
I updated some missing links for figures, those figures were DISAPPEARED from 'imageshack', wierd...
The table is not so nice, but I could not tabulate it since you can not put any extra spaces etc. between numbers :mad:Quote:
Originally Posted by markiemrboo
--
Please, check the figure 13, to be sure you have measured correct points. Here, a short list (maybe more clear one) what I got (measurement accuracy +/- ~10mV):
TP1 = 2.72 V (VDD_MEM)
TP2 = 0 V
TP3 = 1.37 V (VDD_MEM/2, VTT/VREF for memory ?)
TP4 = 0 V
TP5 = 1.36 V (VDD_MEM/2, VTT/VREF for memory ?)
TP6 = 0 V
TP7 = 3.35 V (some REFERENCE ?)
TP8 = 0.60 V (Internal REFERENCE buffered out ?)
TP9 = 0 V
TP10 = 2.72 V (VDD_MEM)
TP11 = 1.12 V (VDD_GPU, not directly)
TP12 = 0.61 V (REFERENCE for Vgpu, for feedback)
TP13 = 3.31 V (VREF, used for Vmem generation)
TP14 = 2.72 V (Generated from VREF with resistor division determines VMEM)
--
If you measure VERY different values, something is badly wrong ..
--
Are you still ready to measure capacitances :)
I found a way to measure ALL feedback component values without removing those components from card :D The accuracy is not 1%, but I think that enough the compare BS and non-BS cards and cards modified afterwards !!
I'll try to make instructions&figures quite soon..
That table is a lot clearer, thanks.Quote:
Originally Posted by SikaRippa
My main concern is that I don't seem to be reading anything at all from TP3 or TP5. All the rest are ok, but aren't TP3 and TP5 for the memory? I was hoping that maybe I had found a cause why my memory won't clock as high as it should? Could this perhaps be why? I assume with a memory voltage of 2.8v I should be getting readings of 1.4v from those two points?
I finished my last exam today, so I would be more than happy to measure anything you want... although I perhaps don't have the same black screen as others are experiencing, perhaps my results could still come in useful. :)
:toast:
That is very wierd indeed.. as you can I measured VMEM/2 also with Gainward 1b (which had wrong memory voltage). So, there IS something wrong with your card - maybe this causes your problem.Quote:
Originally Posted by markiemrboo
I assume that these TP's corresponds to VTT/VREF voltages (not sure !), and by having 0 V in those point means for me that card cannot work !
--
Do you think that layout of your card is EXACTLY the same as shown in those figures (e.g. GW-cards) . Could your measure resistance & capacitance values of those TP's (you can compare to values I have posted!).
--
Measurement instructions are not ready, I will add they to next update
(nice that you still want help :) )
It's incredibly weird, and I am rather worried about it :) It's definately odd that I get 0v from both TP5 and TP3 but my card still actually still works. I've double checked the readings, i've put a wire in the TP point, tight fit, and measured by the wire... but still 0v readings.Quote:
Originally Posted by SikaRippa
If anyone else with a Leadtek 6800NU could measure TP3 and TP5 and tell me if they also get 0v that would be interesting for me.
The layout of my board seems to be the same. I'll measure those values tomorrow morning and get back to you.
Ah ha! Turns out I was measuring TP2 not TP3, as the board is reversed from the picture (I was measuring the back of the card and the picture I was using as a reference was a shot of the front of the card), oops!!
SO, I get 1.40v on TP3, but still 0v on TP5. Definately TP5, I can just about read the writing saying TP5 on the hole from underneath the Arctic Cooler, as it's close to the edge of the board!
OK :), nice that TP3 is clear now ! Are you still sure about TP5 and you are not measuring TP4 (see Fig 13). At least I measured that point :rolleyes:Quote:
Originally Posted by markiemrboo
Absolutely sure, I can see TP5 written on the spot underneath and it reads nothing :(Quote:
Originally Posted by SikaRippa
I'm prolly gonna buy a Leadtek 6800NU monday. Saw that there had been one incident of BS on a Leadtek 6800NU. Had an MSI 6800LE before which was BS, RMA'd it but only got aprox 80% of my money back, no explanation why. I'd hate to have BS problems again. Do you know if the Leadtek 6800NU still has these problems?
Thanks. :toast:
That would be my card. My black screen is something totally different I think. I still get voltages when it black screens and I can still ping my machine, only happens when I overclock the memory too high.Quote:
Originally Posted by krille
If you do get it, would you be able to measure TP5 with a multimeter for me, just to see if you get a voltage reading there? I have a feeling there may be something wrong with my card :)
Ah ha! I was looking through all your figures and saw one that you had drawn arrows pointing at every TP point (http://img112.exs.cx/my.php?loc=img1...ge=fig82gl.jpg)
It seems either you have TP5 and TP6 mixed up, or TP5 is labelled wrongly on my card. The point I have labelled TP5 on your diagram definately reads TP5 on my card and not TP6.
Anyway, the good news is that the voltages are correct, both at 1.41/1.42v, now that I found the right spot. The bad news is...... that my memory still just won't overclock like it should. I think I will just have to give up on trying to get the memory on this catd running at the spec sheets rated stock speeds (900MHz) :slobber:
I'm beginning to think that the memory chips on my card have been labelled wrong and are actually 2.5ns chips :( Ah well!
Heya mate! Sorry, about your card/mems. I'm not that into electrics, but I'll do what I can if you supply further instructions. :)Quote:
Originally Posted by markiemrboo
I'll prolly get the card tuesday, and since you think it's your specific card that's wrong, hopefully I won't (shouldn't you RMA?). Anyway, my previous experiences with Leadtek have been very good so far. Once again, just tell me exactly what to do and what equipment I need and I'll be happy to help out, bro! :toast:
It's actually OK, I think. It looks to me like Sika possibly got TP5 and TP6 mixed up, or TP5 is a different point no the Leadtek boards? But I got the reading I was after on a different point :) Back to square one again now lolQuote:
Originally Posted by krille
The black screen issue I was having definately isn't the same as the black screen other people seem to get.Their cards shut down when they black screened, mine didn't.. there was just no picture on the screen. I think it's just badly marked memory chips on my card and the issue I am having *may* happen for everyone if you push the memory too far. They're supposed to be 2.2ns chips on mine, but they're acting very very much like 2.5ns. There was no difference in overclocking from 2.7v to 2.8v (800MHz), from 2.8v to 2.95v it looks like I can manage 850MHz... but from 2.95v to 3.05v I get no improvement at all, again, just like the jump from 2.7v to 2.8v!
Yeah .. those labels ARE very bad. In Fig 1 the TP5 is 'clearly' TP6 and it looks like a Leadtek-card (i.e. green). In many other cards (e.g Fig 3) label could be identified as TP5 (I'll check mine after opening case!).Quote:
Originally Posted by markiemrboo
Anyway, nice that you got the same voltages :)
--
Have you tested any timing tricks to get more speed ? You can find info at least in guru3d-forum:
http://forums.guru3d.com/showthread....ht=6800+timing
I tried adjusting timings too over the past few days, made absolutely no difference at all for me! I think it's just a lost cause :) May be one day I will figure it out ( probably when there's a Geforce 9 out and my card is totally obselete :) )Quote:
Originally Posted by SikaRippa
I think you were absolutely correct with TP5 and TP6 ;) . I took a close look about those labels and it is more likely that TP6 is for memory and located at lower part of card (=AGP side). The reason I am (was) not sure is that there are holes (vias) under TP5 and TP6 labels, which makes labels unclear (in GW-card you can see only a part of those labels).Quote:
Originally Posted by markiemrboo
Thank you very much for this information !
--
No performance increase with any timing settings ? And you are totally stuck to 800 MHz with any voltages ? Maybe you have indeed 2.5 ns chips..
If I remember correct, you got some boost when adding a capacitor (?) That would indicate some mechanism/correlation (also in your case) between OC-capability and memory voltage (stability). Therefore, it is wierd that voltage increase does not boost OC.
Have you any experience about Vcore overclocking/voltage mod vs. memory OC ? Just wondering if the core would be a bottleneck when OC the memory.
No problem :)Quote:
Originally Posted by SikaRippa
No performance increase at all. Still stuck at the same speed as the default timings! I went ahead and increased the voltage to 2.9v and I can get 845MHz-850MHz on the test, but then I went to 3.05v and I was still stuck at 850MHz, timings didn't change that either! In short:
2.72v -> 2.82v = no increase
2.82v -> 2.95v = 835MHz -> 850MHz
2.95v -> 3.06v = no increase
I think I may have found somewhere I can get a nice quality capacitor, 10v, 2200uF, 105C and apparently low esr / 2.3A ripple. I'm gonna buy one of those and see if it helps.
I've also tried lowering the core to default speeds / default voltage and relocking the pipes, *just* overclocking the memory and I hit the exact same limits.
EDIT: Do you have those measurement figures yet? :)
Did the enhanced capacitor mod and it all works now :) . Thanks
P.S. I had to my capacitor in a pack of 5 so if your located in the UK and need a capacitor that works for the mod then send me an email at storm_gecko_2001@hotmail and I can post one/two to you.
So, just for the record the Leadtek 6800NU cards are *NOT* defective, your individual sample just had some bad mems, while most don't? The leadtek 6800NU is supposed to have 2.2 ns mems?
Oh, and btw, I never understood how to increase my vtt/vddr/vmem or whatever it's called the voltage to the mems on my prev MSI 6800LE, I just knew how to bump my core.
Btw, my old MSI card worked 100% fine when i literally halved the mems freq. But this you probably already knew, no?
They don't seem to suffer from the big nasty black screen, no. Just seems like the chips on my card, which are marked as 2.2ns, actually aren't 2.2ns so it doesn't overclock as high as it should do. Either that or my card has a strange fault :)Quote:
Originally Posted by krille
You need to do a volt mod to increase memory voltage, can't be done by a BIOS flash like the core voltage can!
Ok, cool. I'd love to do that with my card, any link/guide please? How far would it be safe to go? Is it permanent (i e, I won't have to keep doing stuff after vmem is increased), no really, I'd love to pump up them memories! :DQuote:
Originally Posted by markiemrboo
On a side note, do you know what would be better - stock Leadtek cooler or NV Silencer 5 or 4 or whichever for 6800 I have here laying around somewehre, thanks.
It's pretty easy, and yes it is permanent. From the Hynix spec sheet 2.94v is within tolerable limits, so I should think you're probably safe running 3.1v 24/7 so long as you keep the RAM cool.Quote:
Originally Posted by krille
The stock cooler isn't too bad, it's quiet if you unscrew the nasty grill. But I would recommend an NV5 if you're going to be increasing voltages :)
I used http://www.xtremesystems.org/forums/...4&page=4&pp=25 to do mine, although there's no really any "instructions" to follow as such. If you want to see pretty much exactly what I did, see attached drawn on photo.
You can use any of the red parts + ground for the memory mod. I just found the little round solder point the easiest to solder to personally :) I need to take my card out of my system in a few days to add an extra cap, see if it has any effect on my crappy memory overclock.. so if you would like a close up picture, not drawn on, of how I did it I can probably get you one. Pretty sure the drawn on diagram should be good enough though? :)
EDIT: Oh and you can ignore that 12k ohm + arrows part. I didn't draw that. That may be another way of doing it.. but I don't really know :) Looks like it to me, but... im no expert :)
krille:
There is only little information wheather Leadtek-cards (le/nu) would suffer 'black screen' etc. problems. Therefore, I think (?) that those cards are 'clean' or at least much better than e.g GW/MSI-cards
If you consider Vmem-modification, please take a look in post #1 chapter '3.5 Memory voltage modification' there is explained A LOT easier way modify Vmem .. no soldering required :rolleyes: ..
Nice to heard :) Your card is born again...Quote:
Originally Posted by KoR_Wraith
Another no solder required mod. These cards seem good for those...Quote:
Originally Posted by SikaRippa