Tec Ram Cooling

[n00b 0f l337]

Just had a dumb/wierd idea, fugger and such get amazing results when uber cooling ram. But ambient cooling, normal water and air, dont help at all. So, what if you took like a small 80watt, hotplate with just a good heatsink (yeah i know I know, ive tec'd before, air is bad), and make the cold plate have fins off of it to use as like heat spredders.....

Insulation wouldnt be to hard with so small a tec, and you could water cool it if you wanted to...

[QuikSilver]

It would work, although it would not benefit the RAM much, if at all. It would only improve the ram clocks by 1-2Mhz. It would be better to invest in higher quality ram instead. Just my 2 cents.

[snipper_cr]

I dont know if this is true or not but as long as the ram isnt HOT, you probably wount get much better results. I know ram experiences problems (like all other semi condoctor items) when it gets HOT, but a heat spreader or heat sinks would do the job.

Peltier cooling would be an extreme was of power which could be used to cool something else... but what i call waste, one may call inginious.

[n00b 0f l337]

Yeah, I've got quite a few 80 watts around, and they aint good for anything else. I may give it a try. It would be 1 80 watt peltier for like 2/4 sticks of ram. And ram does work really well when cold.

[labrat23]

This gave me another 25mhz on ram that would not overclock at all at ambient temps.




I cant remember the load temps but they were a degree or two over 0C.

[-(CH)-]

That looks...Ghetto... I like it
Didnt u have any problems with condensation???

[n00b 0f l337]

Thats some good gains. Probably much better on good ram or on high voltage ram!

[Holst]

I tried spraying my memory with freeze spray once (on 8kha+)

Didnt clock any higher but I suppose it might have been the motherboard holding me back...

I think the freeze spray held it at about -10 , but I didnt have a probe.

Just kept on spraying it and trying to clock higher...

[n00b 0f l337]

Well... For those tec ram coolers what size dimensions for the crap and were'd you get the peices... Im going to build my own as well I think.

[forey]

nice to see some1 have a go at it nicejob

[labrat23]

I got all the material from onlinemetals.com The heatspreaders are 1/8" x 1" x 5,1/2". The waterblocks are just 1/8" flatbar with channels cut in them and a piece of sheet metal soldered on to the top. The inlet/outlets are copper tube crushed in a vise over a 1/8" bar for uniform thickness and flatness. The ends were crushed flat and soldered. A rectangular hole was cut in on side of the inlet/outlet so it could feed the waterblock from the side. Also soldered on of course. The barbs are cheap home depot barbs I cut and necked down to fit inside the copper tube. Soldered. Since there is only one tec per ram stick, there has to be a heat bridge/pipe to the other side. That is just another 1/8" piece of copper soldered on. That is the trickiest part. I had to mic the ram and then clamp the heatspreaders over shims stacked to the same width when I soldered it together. It is a tight fit and when I put the clamp on the holds the waterblock on, it is even tighter.

Good luck. I will warn you though that all the ram I tested so far either responds very well or not at all. I give it a 50/50 chance.

[datura3]

I got all the material from onlinemetals.com The heatspreaders are 1/8" x 1" x 5,1/2". The waterblocks are just 1/8" flatbar with channels cut in them and a piece of sheet metal soldered on to the top. The inlet/outlets are copper tube crushed in a vise over a 1/8" bar for uniform thickness and flatness. The ends were crushed flat and soldered. A rectangular hole was cut in on side of the inlet/outlet so it could feed the waterblock from the side. Also soldered on of course. The barbs are cheap home depot barbs I cut and necked down to fit inside the copper tube. Soldered. Since there is only one tec per ram stick, there has to be a heat bridge/pipe to the other side. That is just another 1/8" piece of copper soldered on. That is the trickiest part. I had to mic the ram and then clamp the heatspreaders over shims stacked to the same width when I soldered it together. It is a tight fit and when I put the clamp on the holds the waterblock on, it is even tighter.

Good luck. I will warn you though that all the ram I tested so far either responds very well or not at all. I give it a 50/50 chance.

what memory have you tested? If you could give us a list and the mhz increase that would be extremely helpful!

[sluflyer06]

i've seen large gains my super cooling memory in the past......on my old 6800ultra i had ambient at about -3C and a large fan blowing on the card and memory OC went from 1200 to 1296...they had copper ramsinks on them attached with Arctic Alumina Adhesive.

[Holst]

I think TEC cooling graphics memory might give you better results than system ram, but ive never tested this.

Graphics ram TEC cooling would be more of a technical challenge.

[labrat23]

The Ram that got me the extra 25mhz was Corsair xms. It was made by Samsung. Interestingly, a piece of ram made and sold by Sumsung did not see any increase at all. Kingston Value Ram is a no go, but that is probably useless info . The other ram, which I don't have anymore, I dont remember what they did. I would think the only reliable 'databse' would have to go by chip numbers. I dont think I have tested bh-5, sry.

[n00b 0f l337]

I think its only gonna help things that love high voltage. pc2-5400 corsair UL will get better me thinks and so will bh-5.

[labrat23]

My first thought would be to test the ram with air heatsinks and without them. If they do perform better with some air heatsinks, chances are going colder will help even more. The ddr500 that I had that wouldn't overclock at all until I froze it did go a few mhz over if I strapped a fan on them but wouldnt even hit 250fsb with no airflow.