ATI 5750/5770 voltmod thread

[Theorw]

Theorw

Adding extra capacitors is for lowering ESR (series resistance). More caps > more energy stored > more current ripple can be filtered. If you add caps with long wires, + thin wires => you will add MORE parasitic resistance and noise to power planes, that without caps. It's totally stupid idea, to use that long wires from you first pic, sorry

http://en.wikipedia.org/wiki/Equival...ies_resistance

I know pal,i soldered them exactly on the PCB now!!!

[pr@$r1g]

guys please help me out

Mine is Sapphire 5770 & where to what to do ??? > vmem,vmod


mostly i need vmem


& vmod

[Kubis17]

Vmod GPU is same as on HD 4770 because both cards has got same controller L6788A. (Datasheet)



Vmod MEM is realized by controller uP7701 (Datasheet)

[Riska]

If i am seing wright you have to mod both controllers for mem because ther is one for both sides!! correct me if i am wrong..

[oohms]

I don't think those control Vmem

http://www.xtremesystems.org/forums/...2&postcount=11

or maybe

http://www.xtremesystems.org/forums/...63&postcount=2

I havent tried any of them yet.. but tempted to

My card won't even do 1000 core at 1.3v

[pr@$r1g]

Vmod GPU is same as on HD 4770 because both cards has got same controller L6788A. (Datasheet)



Vmod MEM is realized by controller uP7701 (Datasheet)

thanks for ur quick replay

If i am seing wright you have to mod both controllers for mem because ther is one for both sides!! correct me if i am wrong..

some one correct this please

I don't think those control Vmem

http://www.xtremesystems.org/forums/...2&postcount=11

or maybe

http://www.xtremesystems.org/forums/...63&postcount=2

I havent tried any of them yet.. but tempted to

My card won't even do 1000 core at 1.3v

i have really not tried OCing this card ,but my card is @ 950/1300

^^^ please make this a proper vmod & vmem thread for HD5770

Theorw has soldered some caps on his cards ???

[CL3P20]

IMO- Using caps with wire leads is not bad if done properly at all..in fact the added ESR from using wire leads is next to nothing, especially when you consider using high uf caps with a low voltage. The amount of additional filtering introduced to the power circuit is still way beyond that of stock ESR values. For example:

Using SMD caps gives the highest amount of stock ESR value over all other types of caps...
Using caps with a lower voltage rating [closer to that of stock vcore], will have a lower ESR value than those with a higher voltage rating
Using a high uf rating ensures low ESR

...So obviously, it would not be in vain to add 4v, 750uf caps to a GPUv circuit...and use wire leads. The amount of added filtering is still beyond that of stock capability.

useful info I found here: http://kakopa.com/ESR_meter/index.html

A capacitor which is working at 100 KHz and is 2000 times smaller is going to have an ESR which is ten times greater and is going to generate much more heat in much less space. Miniaturization tends to increase ESR as well as impede heat dissipation. We can see everything here works against us.

*I have capped plenty of GPU's myself..and adding electrolytic caps to a GPU circuit regardless of method can bring benefits when done properly.

[Theorw]

Right after i opened the card,i found out that the resistor C304 is connected to the leg of the memory controller and then to the ground....So its the pencil mod right???
HOWEVER WHEN I SHADED FROM STOCK 5,17 TO 4,44 KOHMS i saw no increase at all!I am gonna try shading the other contrell together and see if something happens.If not then probably we should do as shamino says...

[Theorw]

USE ON YOUR OWN RESPONSIBILITY!!!

OK GUYS just did some testing and found the PENCIL and HARD VMOD!!!!
Its the upper uP controller that manages the vMEM.

Default value of the RED circled resistor is 2,35kOhms.
Shading with pencil to 2,13 kOhms went from the stock 1,618~20 to
1,628~30 idle and 1,640 LOAD!
So doesnt this make it and OFFICIAL PENCILMOD?
And of course the green lef to the ground with a 20kohm VR is the GARDMOD right????
Any corrections accepted.
USE ON YOUR OWN RESPONSIBILITY!!!

[Riska]

Where did you measure the voltage?
And did you try to measure on the front mem with that first try?

[Theorw]

I measured in the SMD caps located NEAR the chips!!!Dont worry!!
No i didnt measure the front memory chips BUT I MEASURED THE R301 resistor that is in contact with the leg of the OTHER uP controller and it was 5,72 kOhms so i dont think that the controller apropriate for the back side chips would give different voltage and therefore have different resistor RIGHT???
But for the hell of it i am gonna solder and measure the front side chips too!
Wait 10 minutes!!

[Riska]

thanks man

[Theorw]

Well it wasa A PAIN IN THE *SS to solder in the front cos of the mem heatsinks but i managed to do it and i verify that it is the same on both sides of the PCB!
Somebody YET has to verify thsi mod too because its better when its cross checked right???????????

USE ON YOUR OWN RISK!!!


EDIT:1 little problem though...I had 2*20kohm VR s soldered to gether from a previous mod and i put them in this mod BUT i didnt notice any increase in the volts except i now have 1,635 LOAD instead of the stock 1,620...
Strange.I am gonna put a regular 100k soon and check again....

EDIT No2:i CHANGED THE VR BUT THE PROBLEM PERSISTS...So i am having doupts about the hardmod...Pencil mod work great though!

[CL3P20]

If the mem control IC's have different resistances..then they are most likely for vDD and vDDQ...just like the 5850/5870's..having 2x separate voltages for core and driving voltages... Seems to me this would be more logical, than thinking 1x controller for each side of the PCB..

[Theorw]

I think so too...So we want to change the vDD right?
My successful pencil mod shows that the uper controller is for the (vDD?)
The thing is that i think i need a very small VR like 10kohm...
I try tomorrow...Off to bed...

[CL3P20]

Yes..increase in vDD for mem OC..only slight bump in vDDQ is really needed if anything..doesnt really increase clock potential a whole lot, from the OC's I have seen..

[Astennu]

Are there also people here that have mod a 5750 ?

I already own a 5850. But i like to play with hardware but i dont want to risk my 5850 just yet so i was thinking of buying a 5750 to have come oc fun.

And what kind of gains do you see with the cap mods ? do you also see improvements at lower voltages ? or is the gain bigger when you use it in combination with a v-mod ?

Edit:

I dit some research and i now know why people take the 5770. I thought the 5750 also had 800 shaders active but this is not the case. And off course the stock cooler is better. But i find the 5770 to expensive compared to the 5850's.

[Theorw]

Yes dut 5750s price doesnt justify the difference from 5770...
Its like 10-15 euros here with which in 5770 u get far better cooler,voltage regulation AND 80 extra shaders...All this for 10-15 euros...
So i think that if u want vmod and experiments then 5770 is the ONLY way!!!!

[Astennu]

Yes dut 5750s price doesnt justify the difference from 5770...
Its like 10-15 euros here with which in 5770 u get far better cooler,voltage regulation AND 80 extra shaders...All this for 10-15 euros...
So i think that if u want vmod and experiments then 5770 is the ONLY way!!!!

Here the difference is bigger. A 5750 = 115 euro and a 5770 is like 140+. Both 1 GB. With those prices i would go for the 5850 thats 220 over here.

[Theorw]

Can someone explain the VDDC CURRENT%OCP value in the shot ?What deos it calculate?Maybe when it reaches 50% the card shuts down?I heared that this happens on 1,7volts...AND WHY DOES IT GO UP AND DOWN in regular intervals???????????
iN 1,34V THE VALUE WAS 42%.i INCREASED @1,4V and the value went 45%...Didnt want to torture the card but is there a way to determine the AMPS the gpu consumes by this value?

[CL3P20]

45% + 1.4v = 2v ...I dont think you have anything to worry about..Your screeny doesnt show any OCP value that I can see...am I missing something? The only value I can spot is your GPUv...which says its at 1.12v..a long ways from the 1.4v you stated...

OCP works off of hysteresis ...so changing VID bumps OCP...but with these Voltera IC's it appears that VID does NOT change..only GPUv..so VID/OCP limit stays the same..Have you tried to see if the OCP mod for the 5870 is the same as your GPU..?

[Theorw]

am I missing something? The only value I can spot is your GPUv...which says its at 1.12v..a long ways from the 1.4v you stated...

OCP works off of hysteresis ...so changing VID bumps OCP...but with these Voltera IC's it appears that VID does NOT change..only GPUv..so VID/OCP limit stays the same..Have you tried to see if the OCP mod for the 5870 is the same as your GPU..?

Yes u are missing tha i hardmodded the card!!!
I dont use software i mod with a VR!Far steadier!!!
2)How do i do that?
From what i saw u need to remove a resistor completely and then solder a 18kohm one right?But 5770 has different volt chip...What leg of the chip was the 5870 resistor connected to???
So i could search my chips datasheet and experiment...NO WAY I REMOVE THE RESISTOR THOUGH!!!!DID U?

[CL3P20]

Well..according to the datasheet, OCP is set to trip at 140% of stock VID..

I marked the pin for OCP..and the pads most likely linked to it..you will need to confirm they are connected though.

[Theorw]

Thanks man!Will confirm as soon as i can!

[Riska]

Here is the points for the ocp mod .
The leg on the ic is connected to a smd cap and not the resistor to the side but the little cap