Printable View
There is a known mod that feeds RAM directly from 3.3V line. So, RAM voltage = PSU 3.3V line voltage. Is it possible to change RAM voltage without changing 3.3V PSU line? Does anyone know what type of resistor do I need between those points to change RAM voltage? If not then how low can 3.3V line be? For example, 2.9V ok or not? The problem is that I can only use RAM that likes high voltage. Or I should use some switch to enable/disable mod.
http://home.no.net/comerade/pc/p4gd1_3,3v-vdimm.jpg
Can you post any pictures of the DIMM slot area on the board please.
Here you go, front and back
I measured voltages on the modded board already. PSU 3.3 line was around 3.14V. If I increase it to say 3.4V then all voltages at green points also increase.
From the top to bottom,
1st chip: purple 12.11, green 3.14
2nd chip: purple 2.99, green 3.11
3rd chip: purple 2.99, green 3.09
Last chip: left both 0.8, right 10.7 and 4.83
what? with a 3.3v line set to 3.4v you only get 3.14v (real) to the dimms?
that's pathetic. even less then with the normal vdimm mod (connecting pins
on that lm-chip) on the p4p800-se.
isn't there a straight 3.3v to vdimm pins mod like kyosen did with some other
boards? the pic you posted way at the top looks somewhat different. i used
to think you connect the 3.3v spot on the 20pin (or 24pin) psu-plug straight
to pin x off the dimmslots, not to a point somewhat off the socket (that may
be connected to the dimm-sockets via some resistors, etc etc...) ?
^^^ this is me thinking, nothing to be taken at face value.. (just in case)
example pic of what i mean.. some aopen board (i855gmem lfes iirc):
No,
When PSU line is @3.4V then RAM voltage (and readings at green points) is approximately the same i.e. around 3.4V :D I have tested it up to 3.6VQuote:
Originally Posted by kiwi
But as I said I can regulate voltage only on my PSU :(
OK volt measured on a P4GD1 set to 2.90V in bios running 2 sticks of RAM
From this image the following voltages were measured with reference to ground -
For MOSFETS starting with top most MOSFET
Green pin - 3.41v
Pink pin - 11.86v
Green pin - 3.06v
Pink pin - 4.61v
Green pin - 2.73v
Pink pin - 4.27v
IC starting top left going anti clockwise
Pin 1 - 3.38v
Pin 7 - 4.10V
Pin 8 - 4.51v
Pin 14 - 10.37v
Hope this helps
JB
How can there be such difference between P4Gd1 and P4gpl-x? PCB is almost the same. Maybe because my board's been modded already? :rolleyes:
Dunno, tho I would expect the mod to have a big effect. Gues we'll have to wait till someone else give in their readings!
Done a bit more testing, discovered with MOSFET ouputs the vdimm voltage :
http://img360.imageshack.us/img360/2...d1vdimm4xq.png
Hope this is of some use to you all :)
Can someone check the resistance between the pink pin on the Vdimm MOSFET (the MOSFET in Johnny Bravo's post above this post), and pin 7 and pin 8 on the LM324.
Edit: check resistance with the power off
1.5K for both tho I don't trust the reading, slow to settle out, I suspect a capacitance may be charging up somewhere on the line. These readings mean anything to you?
Actually got the data sheet and see notch, Pin 7 has infinite resistance and pin 8 is reading 1.513Kohm atm
yeah pins 5 &6 are tied to ground so pin seven is out of the running
Are you checking the resistance between the pin on the LM324 and ground, or between the LM324 pin and the pink MOSFET pin?Quote:
Originally Posted by Johnny Bravo
between MOSFET and Pin
hmm, I get 2.99K (pin 7). Nothing @ pin 8. Probably I should remove the mod and measure then.
http://www.xtremesystems.org/forums/...id=40162&stc=1
connect a 50k VR between pin 9 and ground. Decrease resistance to increase voltage. I've marked pin 9 on the attached picture :)
Note: XS owners and members will not be liable for any damage caused to you or your hardware, neither can be liable for any breach of the warranty.
will tell u how its goes some time after 5 tomorrow Persivore, got to take labs in the afternoon so won't be able to get anywhere near the board till I get back home.
vdimm mod confirmed! Measured upto 3.3v fully adjustable so far :-
http://img507.imageshack.us/img507/86/imgp13515ai.jpg
nice! i'm sure many people will be happy. good work as always guys, good work
Thanks a lot :) Good job Persivore and Johnny Bravo
:toast:
We may not be finished here guys, was doign some testing today and it seems that the max is ~3.33v regardless of how low a resistance goes between these two pins. Unfortunatly my BIOS became bored with living and now I have to get the chip reprogrammed...*sigh*
If anyone else has done this mod and can get higher vdimm please report in here!!!
Cheers
JB
The max Vdimm will be limited by the 3.3v line. You will need to have the 3.3v line about 0.2v higher than Vdimm :)
what will this achieve persivore? Does it cause the circuit to go substancially higher?
The higher the 3.3v line is, the higher you will be able to run Vdimm :). Vdimm can go as high as your 3.3v line minus ~0.2vQuote:
Originally Posted by Johnny Bravo
ahh right, i see ur point, I'll give it a shot and see how it goes. Whats the max a 3.3v can be safely lifted to?
The max I would run the 3.3v rail at for 24/7 usage would be 3.7v, but up to 3.9v should be ok for benching.
So it is connected with the 3.3v rail? If I don't mod the psu I wont be able to achieve higher then 3.3v?
You are correct :)Quote:
Originally Posted by NiCKE^
So there is no other way to make a vdimm mod which doesn't requiers that?
Something like do 3.3v rail mod but use the 5v rail instead but in some way regulate the 5v which goes to the vdimm cap with a VR.
Maybe just my imagination playing now? :P
Isn't 3.9V - 0.2V = 3.7V enough for benching? :)
I am not sure but theoretically you can feed RAM from 5V using VR
I spoke a bit with persivore and I'm going to make a little mod to see if we can get 5v to feed the memories.
cant you mod the psu with 3.8 3.9 vio , and make mod vio = vddr ?
that s what i had on my old giga nf3
I wont mod my PSU. Easier to mod the board.
whats the mod NiCKE^, any details on it yet?
I can say it works! :P Succesfully I did the mod, I will post pics in a few.
First:
http://www.xtremesystems.org/forums/...1&d=1134850120
2nd
http://www.xtremesystems.org/forums/...1&d=1134850120
3rd
http://www.xtremesystems.org/forums/...1&d=1134850120
4th
http://www.xtremesystems.org/forums/...1&d=1134850120
Final
http://home.no/nickeh/finale.JPG
BY DOING THIS MOD YOU WILL BE ABLE TO SUPPLY AS MUCH VOLTS TO YOUR RAMS AS THE 5V RAIL CAN HANDLE!
I'm not responsible for any dead hardware because you have made this mod and killed something.
Wow, interesting :)
But I will stick with 3.3V mod. And I think it is easier to mod PSU 3.3V line to make it adjustable :)
Looks good, cheers NiCKE^ and persivore :up:
Hm if someone tries this mod please report results about overvolting.
Seems like I got some big issues, running 3.3v idle will give 3.8v during load!
I don't know if it did this last night but I believe it didn't.
Me and persivore (mostly pers) are trying to solve this problem somehow.
By the way, is it possible to mod any motherboard in a similar way by connecting 3.3V to capacitor leg (the one which reads vdimm) like in the picture of the 1st post?
Yes, I believe you can do it on any motherboard. I used it on the MSI K8N NEO2 s939 motherboard.
Cool :D
But what if there are several capacitor legs that all read vdimm? Which do I use then? Is there any difference?
Dunno, I just look at the pics what others have done :)
The problem with my P4GD1 mod is that the MOSFET get so hot that the lod warms up and gets liquid.
That's probably the cause of droops and suddenly very high volts.
vdimm mod ok like first post
3.85vdimm with 2x256 bh5 ok on p4gd1 :)
when the vdd mod ? :D
Right guys, i've build a variable vdimm power board and want to confirm that the best way to connect it would be to solder where the vdimm measurement point is. Should I remove the existing MOSFET or not?
http://www.xtremesystems.org/forums/...id=40162&stc=1
It would probably work ok if you just connect the new power supply to the Vdimm read point on the MOSFET. If you want to be sure that no damage will be caused, remove the Vdimm MOSFET, but it should be ok without removing it IMOQuote:
Originally Posted by Johnny Bravo
Cheers Persivore, have installed external power board and now have super stable vdimm supply, max swing is 0.005v @ 3.5 volts on BH-5 :D