X800 Pro / X800 XT Voltmods?

ok, i finally got thru all R420 reviews available online and made next purchase decision. so the only thing that's left is to find voltmods for X800 Pro / X800 XT Extreme Edition.

i believe only Driverheaven.com posted sort of hi-res pictures of X800 Pro (X800 XT seems to share exactly same layout). take a look...

so, it apperas that vgpu mod is exaclty the same as one for 9800XT that bmg told us about here.

but what about vmem mod? x800 series is missing one of the Intersil's ISL6522 compared to 9800xt?

does anyone of u guys, volt mod gurus, have any idea by looking at those pics?

Jumping the gun a little aren't we? lol

hehe, i know :) but that would be kind of fun to have the mods ready even though the cards didn't make it to the shelves yet :)

Theres already a thread for this..

Quote:

---

Originally posted by twiggy
Theres already a thread for this..

---

i don't think so, man... i couldn't find anything yet unless u can point us out to it.

What makes you say that the vgpu mod is the same as the 9800xt? Do you know this for sure?

well they say that it is because ATI fixed the 8 broken pipelines on the card thus making it 16, but still 4 are suppose to be non-working.

Quote:

---

Originally posted by k|ngp|n
What makes you say that the vgpu mod is the same as the 9800xt? Do you know this for sure?

---

i just looked at the pics...
x800 Pro and x800 XT both have same voltage controller (Fairchild FAN5240MTC) that 9800xt has... all that area around it seems to be exactly same on x800 and 9800xt.

that would probably mean that vgpu mod is to be done same way... that's what this thread is for - to confirm it and find vmem mod as well...

there're good pics... we just need to look at them... mod's can be ready even before cards are out on the market ;) could be fun...

There are another 2 controllers that look like VDDQ/VDDR - on the back end of the card there are 2 ISL6522CB chips. These are easy to mod. :)

But where ist the Vmem Chip ??

I got my card yesterday.....
Yes, the layout looks very much like the xt. the resistors for gpu mod seems the same. It also has res 1596 in same spot too. I wonder if the over current mod is the same?
Im going to see what it does stock first on air and vaporcooling, then I'll probably go for it and mod it up. I'll post back either way if it works or not. I'm probably going to try first gpu mod, then vddr.

Quote:

---

There are another 2 controllers that look like VDDQ/VDDR - on the back end of the card there are 2 ISL6522CB chips. These are easy to mod.

---

:)

great k|ngp|n, let us know as soon as u get your solder iron...

btw, can u post good hi-res pics of this card as soon as u have some time?

The memory modules on the top edge have been lowered closer to the core somewhat. Could be a challenge for some air coolers, and water blocks hitting ram sinks.....

where can I see some close up pics of the card?

Quote:

---

Originally posted by bachus_anonym
i don't think so, man... i couldn't find anything yet unless u can point us out to it.

---

My bad. It was Tom asking about the 6800 vmods.:stick: :toast:

http://pic8.picturetrail.com/VOL256/...6/53486846.jpg
http://pic8.picturetrail.com/VOL256/...6/53486841.jpg
http://pic8.picturetrail.com/VOL256/...6/53486836.jpg

memory modules are staggered as well.

Well, I can confirm that the vgpu mod is same for x800 :)
http://pic8.picturetrail.com/VOL256/...6/53531027.jpg

I am using the same reading point as on xt too.
Just getting started on core......so far, Im at 680/529 wiTH 1.6V GPU
Those clocks got me almost 32000 3dmark01' with 2.87ghz 3400+.


The memory on this card must suck. On air, the best I could do was 510 on mem :(. On r507, it got to 540 and crapped out. I'm bummed about that. I definitely need to do the vdd mod on this thing. I see guys getting up to 560-570 mem clocks with the stock hsf.
IM going to see what the mem can do today...should increase nature and drags quite a bit judging from the other scores.

sweet.. i cant wait till i can get one.. sometime in the summer when i have money. i hope theres a way to make a pro into an xt so i can save money.. im sure theres a way to unlock the extra pipelines.

Kingpin, why aren't you up on the orb yet? When I checked a few hours ago 32k would have put you in 3rd :toast:

Quote:

---

Kingpin, why aren't you up on the orb yet? When I checked a few hours ago 32k would have put you in 3rd

---

I just posted a score this morning. I am pretty sure ,I will go alot farther once I get this vdd mod done.
Hmmm...now which one do I use for mem mod? Neither? I wonder if reading point for vmem is the same....
http://pic8.picturetrail.com/VOL256/...6/53583483.jpg
anyone know what the stock mem volts are on this card so I know what to look for?

x800 has Samsung 2ns GDDR3 DRAM memory labeled K4J55323QF-GC20.
according to the datasheet, vdd and vddq are 2.0v, so poke around looking for that value +/- 0.1v.

Quote:

---

Originally posted by k|ngp|n

The memory on this card must suck. On air, the best I could do was 510 on mem :(. On r507, it got to 540 and crapped out. I'm bummed about that. I definitely need to do the vdd mod on this thing. I see guys getting up to 560-570 mem clocks with the stock hsf.
IM going to see what the mem can do today...should increase nature and drags quite a bit judging from the other scores.

---

Nice k|ngp|n!!! Hopefully your ram will better with volts. I have gotten mine up to 565 thru some tests no sinks, on air. Great job dude!!

:toast:

Have you run into overvolt protection yet?

Quote:

---

Have you run into overvolt protection yet?

---

Not yet. I got it up to 1.68 volts so far and still running like a champ. This mem is truly horrible on this card. It will not go past 540....even on the r507. I am seeing such good mem overclocks with stock hsf from others..I might exchange this one while its still working :)

You know what is weird is that I am getting way more condensation with this card than with with 9800xt, even at 680/540 and 1.69v on the core. My r507 prommie is almost overkill for this card. It just seems like it doesn't put out anywhere near the ammount of heat my xt did @ 2.15v/3v....for obvious reasons.

The card is still very cold even after like 10 straight benches.

yeah because it generates so much less heat.

Have anybody news about the vmem mod? It looks so far like only Vgpu (and Vddq + Vddr perhaps) works.

bump for vmem mod....

k|ngp|n,
I read somewhere that you were able to do the vmem mod (at least I saw a pic of some mods close to the memory)...

Were you succesful with the vmem mod??

If so, please share the wealth:D

:toast:
Minnyboy

good pic for vgpu
i really like my Minolta Dimage A2 :D

good mod :D

http://sites.rapidus.net/maxbleau/x800%20vgpu.JPG

Nice ram but isnt 500Mhz at cas4 about the same speed as 250Mhz at cas2?....as far as total throughput?

As a curiousity, anyone ever written a video memory test along the lines of SiSoft sandra memory tests for CPUs? Just move blocks hither and fro... and time the action?

can you show me how to do the vmem mod please.

Vgpu, vmem and overvolt mods are the same for X800pro and X800XTPE?

Thanks

Yes, mods are the same on both X800 series cards. Can someone confirm that for the vmem mod, it's pin 5 to ground of the chip on the left from the picture above. What size VR also?

Found it somewhere else already...vmem mod is @ pin 5 to ground with a 20k VR. @ 20k it increases vmem .08 or so.

Thanks stasiu:)

You are right about the vmem

may you indicate, please, were is the measurement poin for vmem?

Thanks

Look at the last pc ice pic and on the left side it's where the green wire is soldered....

Thanks

well i did the vcore mod for the X800XT, the damn solder wouldnt stick to the resistor! Almost burnt off the resistor next to it, but its fine. Damn solder wouldnt stick but eventually it worked ok.

Crunchtime, use expensive solder! With flux already built in. :)

My Vmem mod works, but I am getting far worse results in Mem Overclock! WTF?
Card is a X800 VIVO modded to 16 pipes. Mod works - I can raise the voltage. But the higher I set Vmem, the carppier is my MemOC. :confused:
It's not an issue of temp. I have BGA Freezers put on and a 80mm fan blowing over them.

I'm getting worse mem. clocks as well since doing the vmem mod. Before I could do 625 no problem...now it artifacts like mad.

Quote:

---

Originally posted by pc ice
kinda wierd i could no way do 640 mem before on stock voltage but after doing the mod and removing it..i can

---

Strange :confused: .....but I hope the same happens with my card when I remove it. ;)

OK that's weird. Anyone did the Pin5 to Pin7 method for Vmem?

Maybe i can give a little hint. I cannot solder myself.

The Vddq must be raised together with the vdd (memmod).

Overclockersforum:

hxxp://www.ocforums.com/showthread.php?t=309528


Picture of vddq mod:

hxxp://www.ocforums.com/attachment.php?attachmentid=31428&stc=1

Picture of vdd and vddq mod, different view:

hxxp://www.ocforums.com/attachment.php?attachmentid=31429&stc=1

Important words i think:

"Partially right. With previous GDDR-I and GDDR-II memory running Vddq high and/or above Vdd
would ding the memory OC. With the exception of Samsung 8E's which like the Vddq a little hotter
than nominal spec (and right where ATI had it BTW) Vddq at or a bit below nominal spec worked
best in most cases.

This new Samsung GDDR-III is a different animal and likes it's Vddq run hot and above Vdd (again
exactly how ATI have the memory voltages setup stock) which is against decidedly against
Samsungs spec that Vddq should always be lower than Vdd. ATI knew what is was doing lol!
These new Sammy GDDR-III chips typically do not respond to Vdd increases without a
corresponding increase in Vddq to go along with it. Kraigs Vdd is running right at the high end
of samsungs nominal spec for it and Vddq is running .050 volts above it. Fairly soft really
considering Samsung DDR-I RA chips were often run with VDD a full 1/2 volt above nominal spec.

Note that the GDDR-III chips ran a bit cooler than say Samsung 8E chips at stock voltages but
heat dissapation went up rapidly and radically with increased memory voltages. TM's have no
place on Vmodded Samsung GDDR-III chips"

Can anyone post good pics of measuring points for the vgpu and vmem?

Is there any guide already for the vddq mod???

Is the value of the pot needed to the OVP mod the same that ppl use in 9800xt...20k?

I guess i read somewhere 50k...confused now hehe!

In vr-zone shamino says 500k.

Thks in advance!

I have done the pencil Vgpu mod and it worked a treat :) I have the card running 549 Core from about 500 before the mod.

However the memory on my card will only do 460 at stock voltages. I took a reading of the vmem voltage and found it to be at 1.93v :(

If I do the Vmem mod I can push it to 560 with about 2.1v, but I still can get graphical glitches in Soldiers-Heroes of world war II and a few other games so I know it is not 100% stable :(

I am thinking that Vddq mod might settle things down. Especially if the Vddq is low like the Vmem was :(

I find it very odd that I have seen quite a few posts where people can't get the 1.6n rated ram to run at anywhere near its rated speed, myself included. Maybe the ram has been crippled in some way? Or there was a really bad batch put on some cards?

Is the Vddq mod the same mod as Vmem ? I can't really find much info on it. Is it the same style pin 5 to ground mod that it looks like to me?

On a side note I can also confirm that the pencil vmem mod does not work :( At least not on my card ;)

Cheers

the CAS is a delay timings, which they are equal, but 500MHz gives more bandwidth ...

Quote:

---

Originally posted by uwackme
Nice ram but isnt 500Mhz at cas4 about the same speed as 250Mhz at cas2?....as far as total throughput?

As a curiousity, anyone ever written a video memory test along the lines of SiSoft sandra memory tests for CPUs? Just move blocks hither and fro... and time the action?

---

Quote:

---

Originally posted by arrgh5
Maybe i can give a little hint. I cannot solder myself.

The Vddq must be raised together with the vdd (memmod).

Overclockersforum:

hxxp://www.ocforums.com/showthread.php?t=309528


Picture of vddq mod:

hxxp://www.ocforums.com/attachment.php?attachmentid=31428&stc=1

Picture of vdd and vddq mod, different view:

hxxp://www.ocforums.com/attachment.php?attachmentid=31429&stc=1

Important words i think:

"Partially right. With previous GDDR-I and GDDR-II memory running Vddq high and/or above Vdd
would ding the memory OC. With the exception of Samsung 8E's which like the Vddq a little hotter
than nominal spec (and right where ATI had it BTW) Vddq at or a bit below nominal spec worked
best in most cases.

This new Samsung GDDR-III is a different animal and likes it's Vddq run hot and above Vdd (again
exactly how ATI have the memory voltages setup stock) which is against decidedly against
Samsungs spec that Vddq should always be lower than Vdd. ATI knew what is was doing lol!
These new Sammy GDDR-III chips typically do not respond to Vdd increases without a
corresponding increase in Vddq to go along with it. Kraigs Vdd is running right at the high end
of samsungs nominal spec for it and Vddq is running .050 volts above it. Fairly soft really
considering Samsung DDR-I RA chips were often run with VDD a full 1/2 volt above nominal spec.

Note that the GDDR-III chips ran a bit cooler than say Samsung 8E chips at stock voltages but
heat dissapation went up rapidly and radically with increased memory voltages. TM's have no
place on Vmodded Samsung GDDR-III chips"

---

Thats a beautifully done mod. Hats off to the man with the iron skills :toast:

But a value for that VR would be nice ;) I read 770 ohms from pin 5 to ground, so would 20k be a good starting point?

I take it the Vddq measurement point is the bottom edge of C141? Or is it off C142 (which is directly connected to the FET drain)?

Vddq measurement...

http://img56.exs.cx/img56/2067/x800x...urement.th.jpg

No matter what capacitor, choose one.

I did both the vmem and vddq mods, both with a 50k VR. Pin 5 and pin 7 ( ground ) were used on both mods and it did work :)
The read points work as stated too :D

The strange thing is my ram would not run stable over 470 even with the mods :(

I did notice that AGP voltage ( 1.5-1.8 on the NF7-S Rev2 ) had a direct effect on artifacting. Higher voltage gave less artifacts as you would expect ;)

Just for the hell of it I dropped my AGP down to 4x from 8x and damn if all my problems went away :D :D

I have my ram at 570 now with no problems at all and no volt mods attached !! I have not tested this VGA card on another motherboard so I have no idea if its the board or the VGA card causing the problem.

Out of interest I assume the drop to 4x AGP will give a minimal performance drop if any ? The increased ram speed will easily offset any performance drop I would imagine.

Just thought I would let you guys know my results :) I don't even think I will re-attach my mods now, as 570 is more than enough for me ;)

Cheers

@all

By what Voltage from Vcore do i need the 500k VR resistor ???


Do i need the 500k VR for stbilisation or what ???

can i leave the 500k VR out , or is this not good vor my card ???

Or can i used an other VR as 500k ??


Bye Bloody

Quote:

---

Originally posted by Bloodymaster
@all

By what Voltage from Vcore do i need the 500k VR resistor ???


Do i need the 500k VR for stbilisation or what ???

can i leave the 500k VR out , or is this not good vor my card ???

Or can i used an other VR as 500k ??


Bye Bloody

---

the 1596 mod allows u to achieve greater core speed

in my opinion u should only do it if u phase change ur card

from my experience it allowed me a 20Mhz increase in core speed... but im still experimenting :D

@ Blaster

ok thanks for your opinion

so i leave him out why at this moment i cooled with water , and later when my Mach 2 @R507 ist coming i do it for better core Speed !!


How high did you get your Card with Stock , with Water and then with comp ???

:slobber: :slobber:

Quote:

---

Originally posted by Bloodymaster
@ Blaster

How high did you get your Card with Stock , with Water and then with comp ???

---

Stock max was 560/610

prommy cooled no vmods max was 620/650

prommy and VGPu max was 685/650

prommy, VGPu and 1596 max is atm 710/650

... never tried water, prefer beer :D

... still have to do the vmem mods

later

My readings are vddq 2.11 and vmem 2.11. Can vmem not go any higher than vddq?
Hve moved the resistor for the vmem mod down from 20k to 8k but no change in the volts.

:confused:

@blaster

i have got a big problem , i set VGPU Voltage to 1,6 V for testing out the max core speed!
during the test it froze because VGPU went down to 1,4V,but i don't know why ???
Have you got any ideas why this voltage is so unstable?
But I don't change the VR it is set to 1,6 V and not for 1,4V!!But after rebooting VGPU is at at 1,4V
Can anybody help me ?Any ideas?

Bye Bloody

Bloody:

take it slowly man, if ur still stock cooling u have to try to overclock it from a step at a time, what clocks were u trying with the gpu at 1,6 volts?

Gomer:

I still havent tried the vmem mods bro, gonna do it tonight prolly :D

Quote:

---

Originally posted by Bloodymaster
@blaster

i have got a big problem , i set VGPU Voltage to 1,6 V for testing out the max core speed!
during the test it froze because VGPU went down to 1,4V,but i don't know why ???
Have you got any ideas why this voltage is so unstable?
But I don't change the VR it is set to 1,6 V and not for 1,4V!!But after rebooting VGPU is at at 1,4V
Can anybody help me ?Any ideas?

Bye Bloody

---

Bloodymaster,

The reason your vga is doing that is because Vcore
regulator is initiating a Vcore Over Current shutdown of the Vcore power supply.

Or u get a better cooling for the vga or u can make the ovp mod...better do both ;)

btw nice score on 2k3 blaster...always nice to see ppl from our country on the top :toast:

Quote:

---

Originally posted by VoRtAn_MaDgE
Bloodymaster,

The reason your vga is doing that is because Vcore
regulator is initiating a Vcore Over Current shutdown of the Vcore power supply.

Or u get a better cooling for the vga or u can make the ovp mod...better do both ;)

btw nice score on 2k3 blaster...always nice to see ppl from our country on the top :toast:

---

You just solved my problem :D

I was under the impression that the overcurrent protection forced the card into reset, but now you are saying that it initiates current limiting and hence the Vgpu starts to drop. When I set the Vgpu to 1.80V it drops to 1.68V under load and is really limiting my o/c.

Time to get the soldering iron back out :D

Are you guy's sure the R1597 is right? here is how the 9800xt vgpu mod was done, using the R1592/R1597 resistors and to make it easier I only used the R1592 to ground on my 9800xt.

The R1592 resistor to ground mod for the 9800xt, this was drawn up by charlie.

Quote:

---

Originally Posted by Blaster

Stock max was 560/610

prommy cooled no vmods max was 620/650

prommy and VGPu max was 685/650

prommy, VGPu and 1596 max is atm 710/650

... never tried water, prefer beer :D

... still have to do the vmem mods

later

---

For igpu you have used 500k VR resistor? Fixed to 500k? I have seen others mod with 250k under 1596!! :confused: These Igpu mod it increases the voltage or it stabilizes the voltage?

Thx

Up :D

Quote:

---

Originally Posted by CarNabY

For igpu you have used 500k VR resistor? Fixed to 500k? I have seen others mod with 250k under 1596!! :confused: These Igpu mod it increases the voltage or it stabilizes the voltage?

Thx

---

The 1596mod helps u going higher on the voltage that the gpu can handle! Regulate it to 100k, done then!

Hi all!

I did VGPU and OCP mod as is shown here:

http://www.techpowerup.com/articles/101/images/vgpu.jpg

I thought that maybe the OCP mod would reduce this damn drop in VGPU under load from 1.83V to 1.72V. No?
Or is my OCP mod just not working?

btw, pics are made by W1zzard. :)

The vmem mod is killing ocs because Gddr3 just isnt voltage hungry like the ram on our 9800s and what not was :D

-CaT

Great thread, I am new to these forums. Mods worked like a charm. Thanks guys. :toast:

Quote:

---

Originally Posted by SIK_L_CELL

Great thread, I am new to these forums. Mods worked like a charm. Thanks guys. :toast:

---

That's why people are here...to learn and to help...
Nice have u here :)

Quote:

---

Originally Posted by Goldlocke

Hi all!

I did VGPU and OCP mod as is shown here:


I thought that maybe the OCP mod would reduce this damn drop in VGPU under load from 1.83V to 1.72V. No?
Or is my OCP mod just not working?

btw, pics are made by W1zzard. :)

---

Same problem here. Below 1.6Vgpu I'm getting a 0.03V drop. From 1.6-1.7V the drop is around 0.06V...and closer to 0.10V above 1.8Vgpu. By tweaking the +5 and +12V rails I have been able to reduce the drop to around 0.03V at 1.6-1.7Vgpu. My board seems to like low 12V rail (12.2V is the lowest I can get), and a really high 5V rail (I have mine at 5.8V). The OVP/Igpu mod does nothing for me :( On the positive side my card clocks to ~700MHz on the GPU (room temp water). My current theory is that perhaps I got a GPU that pulls a bit more amps than usual. I have seen this when oc'ing low batch number AMD CPU's (runs hotter, and doesn't like Vcore as much).

I'm also measuring an additional drop of ~0.03V at high loads from the Vgpu caps to the voltage feed pins on the GPU itself:

http://xtremesystems.org/forums/atta...id=19070&stc=1

As for the 0.03-0.06-0.10V drop, I think one of the current sharing channels is dropping out. I haven't been able to understand this fully, but this is what I currently got:

http://xtremesystems.org/forums/atta...id=19071&stc=1

There's two hexfets from +5V->channel2->GND, and two and two in parallell from +12V->channel1->GND. By just considering the number of hexfets pr channel, I think 2/3 of the amps are pulled from the +12V rail and 1/3 from the +5V rail.

Quick question I used a 100k fixed for the OVP but I still get lockups around 680 and any voltage over 1.85 I get shutdowns. Thanks.

Quote:

---

Originally Posted by lsevald

By tweaking the +5 and +12V rails I have been able to reduce the drop to around 0.03V at 1.6-1.7Vgpu. My board seems to like low 12V rail (12.2V is the lowest I can get), and a really high 5V rail (I have mine at 5.8V).

---

thx for your input Isevald.
I thought that maybe my cheap VR could have something to do with that. But now I see you use a cermet type VR and also have this "VGPU droop" prob. :(
I will try changing the PSU rails and the VR to a cermet one soon...

Quote:

---

Originally Posted by SIK_L_CELL

Quick question I used a 100k fixed for the OVP but I still get lockups around 680 and any voltage over 1.85 I get shutdowns. Thanks.

---

Cooling?

Quote:

---

Originally Posted by VoRtAn_MaDgE

Cooling?

---

Ya think? I have never seen the temps go over 38c. Hmmm....

Hehe. Thanks to Isevalds suggestion I upped the 5V rail on my Powerstream to 5.50V. The VGPU now stays at 1.80V under load. :D
(Upping 12V didn't help, instead made the drop worse. So, let the 12V rail low)
thx!

Glad I could help :)

I suggest you watch the volts realtime under 2k1 Nature, as it seems to push the card really hard (at your highest Vgpu and GPU clock settings). Max I can do in that test without Vgpu dropping like crazy (with 12V@12.2V and 5V@5.8V), is ~1.7Vgpu load (1.73ish idle) and 708MHz. Perhaps I would be better off if I could get the 12V rail lower?

I have a 100k fixed for my OVP and it still drops. Should I use a different resistor? Thanks.

Sik L Cell, no. The OCP mod does not help with the VGPU drop. Only increasing the 5V rail of your PSU probably will.

Quote:

---

Originally Posted by Goldlocke

Sik L Cell, no. The OCP mod does not help with the VGPU drop. Only increasing the 5V rail of your PSU probably will.

---

Correct!

Ati loves high 5v rail in generall!

The OVP only "cuts" or "retard" the reset of the power supply of the gpu according to the value of te resistance, 100Kohms is fine!

Still droppin though! 5v at 5.50

Big problem with my OVP mod :(
I have used one fixed resistance from 50k to 37k
Ok,the value originates its passed from 41k to 19.5k..I do not have problems with voltages 1.5v <--->1.85v....Over 1.85v i get shutdowns :upset:
Somebody help me?
Thanks.

Quote:

---

Originally Posted by lsevald

Same problem here. Below 1.6Vgpu I'm getting a 0.03V drop. From 1.6-1.7V the drop is around 0.06V...and closer to 0.10V above 1.8Vgpu. By tweaking the +5 and +12V rails I have been able to reduce the drop to around 0.03V at 1.6-1.7Vgpu. My board seems to like low 12V rail (12.2V is the lowest I can get), and a really high 5V rail (I have mine at 5.8V). The OVP/Igpu mod does nothing for me :( On the positive side my card clocks to ~700MHz on the GPU (room temp water). My current theory is that perhaps I got a GPU that pulls a bit more amps than usual. I have seen this when oc'ing low batch number AMD CPU's (runs hotter, and doesn't like Vcore as much).

I'm also measuring an additional drop of ~0.03V at high loads from the Vgpu caps to the voltage feed pins on the GPU itself:

http://xtremesystems.org/forums/atta...id=19070&stc=1

As for the 0.03-0.06-0.10V drop, I think one of the current sharing channels is dropping out. I haven't been able to understand this fully, but this is what I currently got:

http://xtremesystems.org/forums/atta...id=19071&stc=1

There's two hexfets from +5V->channel2->GND, and two and two in parallell from +12V->channel1->GND. By just considering the number of hexfets pr channel, I think 2/3 of the amps are pulled from the +12V rail and 1/3 from the +5V rail.

---

where do I find a guide for those cap mods?