PDA

View Full Version : temp difference?



Frozendeath
08-02-2004, 10:34 AM
Hi all,

Just a question:

Homemade chiller chilling 1.5 gallons of liguid to -24 C or so , that is being pumped via a 100gallon / hour pump ( submersed) to my maze 4 cpu block.

Now I dont know what the block temp is because I dont have a temp sensor on it, but since all tubing is insulated I cant imagine its higher than say -20C at most.

So then Why is it that if I am making good contact on my cpu (p4 3.0c in an IC7-G)

That as soon as I start it up and get into the bios the temp sensor for cpu says 1 to 3 C ..... MBM shows me idle at -2C ..... but never goes lower.

So wheres the 20C difference between the block and the cpu core? Is this normal? I have to be making fine contact , theres even pressure on the mounting brackets and the entier mobo is getting really cold around the cpu (12C power mosfet temps down from 40 C prior to chiller )

So is this just a sensor reading wrong is what?

Just confused,
Thanks for any input

KODE
08-03-2004, 03:22 AM
Frozendeath,

How much anti-freeze did you added to the loop?

I had a similar experience when my chiller was operating at -30c. To achieve such low temps I has to add a 50/50 ratio between Water and Anti-freeze and we all know that anti-freeze is not so good as water to carry the heat from the Water block. If I'm not mistaken I was running my Barton 2500+ with 2.3V at 2633MHz with water temps at -30 or so and CPU temps at -5c and my pump was doing a lousy 140 liter per hour...

I decided to make some changes to this chiller bacause waiting 40minutes to 6 liters of water reachs -30c sucks.

Frozendeath
08-03-2004, 05:48 AM
Im using straight de-icer windshield washer fluid

let me get a link:

http://www.midwayautosupply.com/detailedproductdescription.asp?982

its methonal + water (unknown ratio as it doenst say) with a bit of ethelyne glycol in it as well

Doesnt freeze at all , but im wondering if my pump isnt a high enough flow rate to keep the block cold under load, but the strange thing is that if I let my chiller go with the pc off (no load) it gets to -25ish C ... then I turn the pc on and within 10 seconds im in the bios looking at temps.... and the temp of cpu core says -1 C.

I cant imagine that in 10 seconds the block temp went up 20+ degrees, the heat load cant be that much.

KODE
08-03-2004, 07:56 AM
The sky-rocket temp issue in my opinion can be caused by two factors in your case:

1- Pump is not being able to maintain the high flow due to higher liquid viscosity. My water/Anti-freeze mix got Really viscuos at -30c. How's your liquid looks like?

2- I don't know much about the product you're using since we don't need this kind of product where I live. I suggest you to try some 50/50 - Water/Car radiator anti-freeze and see what happens.

Do you have any pictures of your Chiller?

Frozendeath
08-03-2004, 08:27 AM
Let me get some pics, and ill see what the liquid is like at the low temps

i_jester
08-03-2004, 10:09 AM
Regular automotive antifreeze will get fairly viscous at -30 degrees. Many people have had good success with Denatured Alcohol. It is flammable though.

Blergo
08-03-2004, 10:30 AM
imo stick with the windscreen washer fluid, it is the best stuff i have found to use in my waterchiller and i know a lot of people on the forums use it. mabey try a larger pump?
owen

Frozendeath
08-03-2004, 10:44 AM
I checked the liquid at -25 ..... its basically like water so I dont think its to thick.

As for a pump Im starting to think that could be the issue, whats a good pump to use? (flow rate and such) my current pump is submersed but I can do inline without much issue.

Oh and if its inline (thus not in my res) will it crack as the room air and the water temps are so different?

OCme
08-03-2004, 11:46 AM
Originally posted by Frozendeath
IAs for a pump Im starting to think that could be the issue, whats a good pump to use? (flow rate and such) my current pump is submersed but I can do inline without much issue.

i would recommend swiftech mcp600... no seals to worry about & plenty of head to push liquid through multiple wb's


Originally posted by Frozendeath
Oh and if its inline (thus not in my res) will it crack as the room air and the water temps are so different?

don't worry about yur pump cracking... insulate the front of the pump & leave the motor or the back portion of the pump open to air.

Titan151
08-05-2004, 11:38 AM
Hi,

I get a similer issue, I have always figured it was caused by one of two things.

1st, the use of a chiller in my case required me to insulate the center of the socket. This is where the mobo temp sensor on my board is at. So I am insulating this sensor (w/ neoprene square) from the cpu. Not sure how much of an impact it has but it must be at least several degrees C.

2nd, my chiller also keeps the liquid below freezing, and my cpu is around 8-12 C. I would assume that even with adequate clamping pressure there is always some loss in heat transfer from the die to the block, so it should always be a little warmer then the water (10-15 C).

But overall, I guess it is hard for me to tell what temp sensor is better and what is not. The one on the mobo is insulated and the one I installed on top of the chip, with the tip touching the edge of the die is questionable since this is bumped up against the block. I tried for several hours the other day to get one installed under the block but I failed to get the CPU to sit properly several times. It would be nice to know what sensor people use when they have a chiller or peltier installed.

-just my thoughts, easily could be wrong. :D

Frozendeath
08-06-2004, 04:29 AM
Hmm interesting, I too insulated the socket (neoprene square and such) never gave that a thought. But anyone know where the IC7-G's cpu temp sensor is..... I was told they measure right from the cpu (current drop off a diode) so is that diode on the cpu or is it below the insulated socket?

Frozendeath
08-08-2004, 06:29 AM
Ok yesterday I got a 350 gph pump, put it inline and my temps dropped 5 C :) So thats probably the best I can do atm as the heat is really transfering well from the block to the resivoir (with my old pump it wasnt transfering well at all)


Thanks for the help all

KODE
08-09-2004, 03:45 AM
Glad it worked out Frozendeath, 5c is a big temp decrease :toast:

kayl
08-10-2004, 10:18 PM
Originally posted by KODE
Frozendeath,

How much anti-freeze did you added to the loop?

I had a similar experience when my chiller was operating at -30c. To achieve such low temps I has to add a 50/50 ratio between Water and Anti-freeze and we all know that anti-freeze is not so good as water to carry the heat from the Water block. If I'm not mistaken I was running my Barton 2500+ with 2.3V at 2633MHz with water temps at -30 or so and CPU temps at -5c and my pump was doing a lousy 140 liter per hour...

I decided to make some changes to this chiller because waiting 40minutes to 6 liters of water reachs -30c sucks.

hey kode have you tried to oc with not such a high vcore.
with that oc at 2.3v @ 2633MHZ totals to 190w plus extra heat added from the submerged pump and heat lost through insulation.
if you lower vcore to 2vs this will reduce the heat to 144w or so and you should have better temps, which might yield a better temps getting a similar oc or beta.
too many volts can make a cpu less stable.

KODE
08-11-2004, 03:48 AM
Hi kayl,

Thanks for the tip :up: but this was my old setup with a XP1700 DUT3C T-bred B, Iwaki in-line pump, etc... not running it anymore. The method I use to overclock is first up the clock a bit, if not stable I increase the Vcore a tad. This way I use just the Vcore needed. Unfortunately 2.3V was required to run at 2.633MHz 24/7 Prime stable.

PS: I'm currently running at 1.9GHz Vcore @ 1.38V trying to find my maximum Prime95/3DMark01 stable FSB.