PDA

View Full Version : ABIT KD7 ~ VDD mod ?



SKYNET
01-17-2003, 10:14 PM
the following problem has ! my ABIT KD7 does not go stably over 205FSB ! now I would like to make VDD mod! already did the gesammte InterNet find searched however nix to such a thing, has one of you an idea ?

plz help me :(


sorry for my bad english :rolleyes:

KS1
01-19-2003, 11:59 AM
i think you should look for KX or some older abit m/b mod in this forum to see if it's a similar one?

SKYNET
01-19-2003, 07:32 PM
is a other IC

KS1
01-20-2003, 01:00 AM
look for the word HIPxxx or IC number near MOSFETs and Inductor coils.

SKYNET
01-20-2003, 01:04 AM
i found the hip for VIO, V-dimm & V-core:

HIP6521CB
P0149ABBA

KS1
01-20-2003, 01:41 AM
then goto a regular V mod procedure.
pls find the info on this forum (Vmod basic)

Arz
01-22-2003, 10:25 AM
Had an AT7, which could handle 203 FSB, BIos offers two settings for VIO, 3,50V and 3,65V. I made a mod to get more VIO, but no use of it, tried with 3,82V but the best FSB I could get was 195. I don't know what happened, but after removing the mod it went 200 again?

BTW, I don't think that HIP6521 controls the Vcore, though it's possible. With my AT7 there is HIP6301 for Vcore and HIP6521 for Vmem. HIP6012 was for VIO. Again these are for AT7.

trodas
01-29-2005, 03:01 AM
SKYNET - so, any sucess with the Abit KD7 mobo? :stick: :confused:

I tought I should to a OCP mod for it, since it is not able to deliver steady voltage for 2500MHz AXP-M at 1.850Vcore... :(


KS1 - Vmod basic ...? :rolleyes: Im a bit confused, you mean just try what was tried for the HIP6521CB and P0149ABBA chips? :confused:
Hmmm, I did not even know, witch one is for the Vcpu.. :(

trodas
01-31-2005, 04:31 AM
Other than that, I tried the Enermax 485W PSU with exactly the same result as the Zalman 400W B revision of PSU - 1.950Vcore stays longest, almost a hour of Prime95, then error.
1.975 - 15min.
1.900 - 25min... :rolleyes:

It has to be the OCP or something like it :( :rolleyes:

Hardly it can be the memory, since I use this test to stress the CPU:

http://ax2.old-cans.com/prime95_cputest.gif

...so amost no mmory is used. The board is kind of funny... :( It offer unbelieable 2.3Vcore, but can't deliver steady 1.850Vcore to 2500Mhz running Barton... :mad: :mad: :mad:

persivore
02-01-2005, 12:27 PM
Trodas, if you can get a pic of the area around the Vcore MOSFETS, I will try to find a mod. I think you will need a Vdroop mod to help stabalise the Vcore under load.

trodas
02-26-2005, 02:15 PM
You think you should? :D This soulds good, but... I already added a huge heatsink on the mosfets - could you not like more the picture of the regulating chip? (+surroundings) ;)

It will be pretty cool if you can figure something out :toast:
(more folding :D )

...actually, I run the machine at 2500Mhz for quite some time, but it gets unstable now... :( And no, no capacitor mod, pls! :rolleyes:

persivore
02-26-2005, 02:24 PM
A picture of the Vcore regulator chip area would be great :)

trodas
03-01-2005, 06:56 AM
persivore - so I opened the poor mobo, took PSU and CNPS7000Cu out and took these pictures:

http://ax2.old-cans.com/g.php?p=KD7G&d=1&v=v2

...hope they coud help, determining why the mobo always end with error in prime95 CPU torture test at 2500Mhz (166x15)... :( (166x14 working flawlesly)

However Im a bit concerned about the temps of the CPU. My O/C might be as well limited by these :(

persivore
03-01-2005, 07:34 AM
On the HIP6301, can you get a resistance reading between pin7 and a Vcore read point, and pin10 and a Vcore read point please.

trodas
03-01-2005, 11:18 AM
http://ax2.old-cans.com/TooHighTempForAS3.jpg

That's why Im worried about the temperature :p: :(
(sadly, on the picture is not much visible the difference between the color of normal AS3 and the dark gray one, yep... :(

Anyway, by Vcore read point you mean just any pin in the CPU socket, that supply Vcore to the CPU? :stick:

persivore
03-01-2005, 11:53 AM
What sort of temps do you get?
You could use a Vcore pin from the CPU socket, but its probably easier to use one of the points where one of the Vcore chokes meets the motherboard.

trodas
03-01-2005, 12:09 PM
Well, I did not saw any trace or something labeled Vcc or Vcore, so I just get the ADM guide there:
http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/25175.pdf
...and find out the pin Vcore location for itself - there are plenty of Vcore pins, labeled Vcc:
http://ax2.old-cans.com/AXP-pinoutdiagram.png :D

I used the AM/2 one (top left, the middle into the left shaved corner :p: ), sticked a cooper wire with similar diameter as the CPU got, tried pull-out a little, if the socket holding it (holding, but I tought it should hold it harder :rolleyes: ) and measured.

On the Abit KD7G mobo, from pin 10 of HIP6301 to the Vcore is the resistance exactly 1K ohm (measured 0.999K) and from pin 7 of HIP6301 to the Vcore is the resistance exactly 2.5K ohms. (measured 2.496K) :D

Temps? I tried pushing hard, very hard, a little over 86C :slap:
Too much, I quess... :( But the mobile is rated to 100C ( http://doublescan.wz.cz/2500Barton_mobile.htm ) and I use the Zalman CNPS7000Cu, so I getting no idea, why the temp is so so damn high... :stick: Even if I run it at 1.725Vcore (166x14) and 2000RPM on the Zalman, temperature (core reading on this mobo) getting over 63-66C after day of folding :rolleyes:

persivore
03-01-2005, 12:46 PM
Can you check if the point I have marked is connected to the Vcoreread point.
http://www.drunken-student.co.uk/tom/mods/Abit%2520KD7G%25202.jpg

trodas
03-05-2005, 06:01 AM
persivore - sorry, to keep you waiting, mate, I got some urgent stuff to do, so give my pls a day or two... ;) :p: Hope you understand... :(

persivore
03-05-2005, 06:14 AM
persivore - sorry, to keep you waiting, mate, I got some urgent stuff to do, so give my pls a day or two... ;) :p: Hope you understand... :(

No problem :)

trodas
03-14-2005, 10:35 AM
I think a bit SORRY is necessary for my extreme lazyness :(
You trying to help and discover the mod and I just did not getting to do... Shame on me.

No results - yes, the marked pin IS connected w/o any resistance (measured 0,6 ohms, the measuring wires have 0,3 ohms for theirself :lol:) to Vcore read point :)

What next measurments is need? :D

PS. I pushed the CPU out for the work, but I wonder - if is better this way, or with the CPU presented? I better took it out - just to avoid removing the mobo from the case (10 screws, the Zalman CNPS 7000Cu got only 2 screws :lol:), but if the CPU is present, it should probably not harm the measuring, Im right? :rolleyes: :confused: :stick:

persivore
03-14-2005, 11:06 AM
Having the CPU in the socket shouldn't effect the readings, but I always remove the CPU before taking readings/do a mod to reduce the risk of the CPU being killed.

For the droop mod, you need to connect a 50k variable resistor between pin 7 of the HIP6301 and the point I've shown in the picture (A point connected to the Vcore supply output).
You will need to 'tune' the VR to get the right level, but I think that people have found that setting the VR to 28k and then soldering it to the board is the optimal value, but it could be different for your board.

http://www.drunken-student.co.uk/tom/mods/Abit%2520KD7G%25202-Vdroop.JPG

trodas
03-17-2005, 08:50 AM
I get my balls together and did it. Because of the lack (I hope I can find them somewhere :( ) of the 50k variable resistor, I just bought a there 27k resistors and the one, what get almost 28k I soldered there :p:

Surprisingly, the mobo boots :lol:

Good points - BEFORE, even the mobo is Prime95 stable (for few hours), when I try memory testing ( http://ax2.old-cans.com/prime95_memtest.gif ) or CPU test ( http://ax2.old-cans.com/prime95_cputest.gif ) - however THIS pwm test stoped at error in 5min or so:

http://ax2.old-cans.com/prime95_pwmtest.gif

...but after this Vdroop mod, it hold pretty well for about a hour, so I decided to give 166x15 again a try. But nope, it seems better, but it still stop at error after some time.

Decrease or increase the resistance now? :p: The variable resistor will be probably better solution, since the soldering is a bit hard :p:
Size of the resistor and it's wires... well :) No pain, no gain :D

PS. thank you, persivore, for improving stability of my folding machine :p: :D

persivore
03-17-2005, 09:21 AM
Decreasing the resistance should reduce the droop, but at the cost of Vcore, so you might have to increase the Vcore in BIOS if you decrease the droop. Have you checked what the droop is like with a multimeter with and without the mod?