PiLsY's mod for Prometeia/P4's is up

[MrIcee]

Just posted up PiLsY's guide to improving contact pressure on the Prometeia when using the P4 CPU.

PiLsY's Prometeia Mod guide (http://www.xtremesystems.org/mricee/images/Modding%20the%20Prometeia/Prometeia%20mod.htm)

Enjoy my friends !!

Randi:D

[Its-Freezing]

Hmmm, something is wrong here, (or maybe I am, in that case help me out here).

It stated a P4, the MB looked like a P4, and the mounting bracket and rear cover was also for a P4........ so what's wrong ?

The good damn (sorry for the swear words) CPU did'nt look like a P4, at least not the ones we see around here.

Could it be because the heatspreader was taken off, well it could, but then the storry just doesn't quite fit (you need to cut down the legs of the Mounting bracket too, if you want to pull that stunt ;) ).

edit: on the new P4 kits that is, as there is not sufficient Gap in the shorter legs and narrower opening to press anything into the opening in the leg under the tap. So on the new P4 kits, the legs of the mounting bracket has to be cut off instead (which still might be a tiny bit too litle :(

Look more to me like it was a P3, but then again I must be wrong :slobber:

edit: ohh, now I see, it seems to be the old P4 kit, that he has been using, so it is a P4 indeed without the heatspreader, in which case he must have won at least 10 degrees or more at full load (leaving out 1 thermal bridge as he is). However it might be a good precaution to use one of those shims that we have for the AMD's, as an extra precaution

In any case, there are 2 other important points which didn't make it into this little "12 steps to improve your contact", so it does not turn into "2 missing steps how to create a bathtop for your CPU" :wiggle:

The first one is about the SMD component residing under the mounting bracket. If any part of it sticks out (which I kind of recall it does on that MB) it needs to be covered fully, legs and end of IC too. There is a nice little tunnel under such an IC, that perfectly guides a fresh supply of humid air into the cell :cool: , it's allmost like putting a sign up right next to it, saying "Free passage here fellows" (Fellows being H2O molekyles just pleeing to enter the Cell, to have a taste of the cool dry air inside :D )

Now what is the next one, well the screws on top of the Coolinghead, remember !! it's in the manual --- Read the ....... manual boys and girls ( The ....... is me trying to speel the word User, just not counting the letters properly ;) )

Put some sealstring (and it's not made from, well you know...) on top of the screws, both to prevent the small entry through the thread on the screws (Yes, that all it takes), and to prevent that the screwheads gather condensation, cause they are Coooold.

Apart from that, I better not comment too much on it (liability and legal issues and all that)


Stay cool
Steen, Chip-con

[MrIcee]

Steen:)

Thank you for your words of wisdom my friend:)

And yes, that is a P4 without it's heatspreader, PiLsY is known to experiment with everything he gets his hands on in the name of reducing temps:)

Of course, I included the extensive disclaimer to remove any responsibility for damages to the respective parties, including the fact it will completely void the warranty.


Randi:D

[PiLsY]

Steen - the doubling up of the seal string over the HIP regulator has the effect of squashing the stuff right inbetween the legs of the HIP chip. Trust me I made absolutely sure :). As for the screws, ive got a bit of seal string on the threads and inside underneath the screw "cap". Thought it looked a bit cack with a big chunk of black gloop on top so I put it underneath instead :). Its still nicely sealed though.

PiLsY.

[Its-Freezing]

Well it maybe looks a bit "cack", but with nothing covering the top of the screws, they can still produce outside condensation, as they get very COLD, so they should be covered nevertheless.

It may not be a problem right now in the northern hemospfere with low humidity indorrs, but it will turn into a problem when humidity in the air increase, and for somebody else having a higher relative humidity right now !

Concerning the HIP (power regulator SMD component) on the MB, it is imparative to present that if a SMD device is sitting crossing the barrier, it must be covered all around also at the end. On another MB it might be sitting differently. Please remember that if the end is sticking out into free air, there is a gap underneath it between the MB and the component !

My point to both topics is that you display this proposal to the world, therefore it should be as fail-safe and as caretaking as possible, with all possible "loop-holes" covered (both the ones for the free air and the ones for the reader not get everything done as thorough as possible ;) ).

I would appreciate very much if these points were entered in your review, as not everybody take the time to read this thread perhaps.

By the way if you only got a 4 to 5 degree improvement of the Core temp, it is either because the MB cannot present the real decrease in temps, or because the microfreezer is having a small tilt, something one of those shims could help to prevent/reduce, you should see at least a 10 degree decrease in Core temp. at full load, more like 15 degrees in fact, trust me :cool:

Steen, Chip-con

[Merenfreak]

@ Steen

Hallo,

I got a Prometeia unit one month ago and i have to say i like it :D

But i would also like to see something like a "chillcontrol" ( like in the new Vapochill) in a Prometeia System, where i can change the load of the heater ( Can i disable the heater now, without having the fans running at 100% ? ), the speed of the fans and maybe also the speed of the condenser. That would be really nice :toast:
Are you working on something like that ?

And i have a special question on my system :

I have increased the pressure on DIE using an AMD 2400+ by taking like 8 - 9 spacer rings instead of the 4 you usually recommend... That gave me an extraboost of ~ 250 Mhz ;)
I also remounted the System like 30 times to get the best contact and this score :

2845 Mhz with my cpu @ 2,15 Vcore ( First try was 2550 ! )
But i feel like the Prometeia seems to reach their limits.
I cannot drop the temps to 0°C Idle ( Temps around 4 °C ) and the Evaporator shows temps up to -23°C at full load after a few hours. ( 22°C room temp.) ( Full load CPU temps are up to 15°C )

Sometimes it also happens, that the system is going under -20°C just after the first boot up and it reboots. ( but very seldom )
In this case the temps are rising very rapidly, like 1°C every 2 seconds.

Did i reach the limits or is there something wrong with my setup ?

Ahh, one more question ;)

Which parts inside the Prommy should be cooled down ? Which parts does it not effect ? I tried to cool down the radiator by an extra fan, but i had like no improovement. There is also no improovment by blowing in fresh air into the system... But something is lowering the temps rapidly after 30 minutes, because i have start idle temps of -7 °C on CPU and after a few minutes i have like 10°C more :( , What is getting so hot ? The condenser ? The microfreezer ( through the heater inside ) ?


thx and cya,

[ARGON]

Ok, maybe I am missing something here...

In the article he said that the microfreezer would have a HIGHER temperature than the CPU?!? If that is the case, there would be no heat transfer from the CPU to the cold plate (Second Thermodynamic Law). I suspect there is some callibration that must be done with the temperature probes.

If I am wrong, would someone please educate me? I can't reason how pilsy's explanation might be true.

[MrIcee]

ARGON:)

You misunderstood what PiLsY wrote my friend:)

What he was infering was that with better core contact made, Evaporator temps will be higher than previous because of the increased heat transfer, while at the same time core temps would be reduced by 3-4c.:)

Randi:D

[ARGON]

Thanks Iceman! lol

While it pains me to see my reading comprehension is slipping, I am glad that the Laws of Thermodynamics aren't!

[PiLsY]

Steen,

WHen I first got my prometia I completely covered the chip youre talking about in seal string. However, as its the vreg chip it does put out a bit of heat. The higher the vcore, the warmer it gets. Under normal operation its at about 4c above room temp give or take. However, if youre running 1.75v or higher its about 7c above room temp. Covering it in seal string acts like an insulator (as the stuff is designed to do) and its gets VERY hot. CPU voltage starts spiking and the whole system becomes unstable.

A lot of this is due to the excessive temps at the mosfets shown by all IT7 series boards and their proximity to the HIP chip, and covering in seal string compounds the problem. Its one of those no win situations - damned if you do, damned if you dont.

I will ask Randi if he can add in a small note in that section of the review about pushing seal string inbetween the pins to ensure a 100% blockage of air flow to try and stop the pump effect. I'll also ask him to mention that you recommend covering it completely. However, I will have to add a warning that doing so can cause instability at high vcore settings.

BTW - core temps without the heat spreader are nowhere near 10c lower :). I presume this si what youre talking about when you say 10c lower. The spreaderless cpu is for a later article and was only in the picture as it was the best picture I had to illustrate contact. The difference is only around 3c without spreader (with mods in the article that makes it 8c lower). However, the heatspreader does just that. Its a thermal buffer between cooling and cpu and is 100% necessary to stable CPU operation. High cpu speeds and voltages of over 1.63v actual cause the cpu to hang coming off 100% load without it. The actual article will compare a normal P4, with the same p4 with no spreader, and with the spreader replaced with aluminium, copper and silver coldplates, along with different thicknesses of coldplates. Needless to say silver gives best results. Around 3c lower than with stock spreader and a whole 80mhz higher cpu speed. Anyway, as I said thats for a later article :).

Thanks again for your input,

PiLsY.

[KenKunen]

Hi guys.. just wanted to say thnx to Pilsy for the guide, as a prometia "noob" it helped me get my p4 2.66@3.45 from -12c idle to -22c or lower, and dropped the full load temp from +1c to approx -9c, maybe not the best I've seen but it seems reasonable for 1.77vcore :)

a quick question.. what's a safe vcore to run 24/7, I know to keep it 1.75 or less on air, but can I go higher with the prom ?

thnx

[PiLsY]

Personally id stick with 1.73v actual vcore or less for 24/7 usage. For short busts though (benching) I do go higher. However, I think youll find with the Prometia that once you get over about 1.68v actual that diminishing returns kicks in and you dont see much improvement for the voltage you use.

Glad the article helped :).

PiLsY.