• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Air-cooled pelts?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Skiing Squirrel

Member
Joined
Jun 22, 2002
Location
Charlotte, NC
Can it be done? Im looking around ebay, and seeing these very cheap prices on durons, and cellerysand thinking about getting one and overclocking the poop out of it. Could I air cool a pelt? Provided I would need a big fan and a very good heatsink. Can it be done?
 
Up to about 80w, maybe. After that you are into water. A very cheap water cooling setup could be done as water blocks for the pelts do not have to be very fancy.
 
You could, just get a good socketA/370 HS, put a big fan on it, and increase your case airflow.
The main problem would be clamping the Pelt to your HSF. With Skt370 /Skt A HSF's, they are usually too small to properly mod them to add clamping holes for the coldplate bolts.
You could just AS2 epoxy the coldplate to the pelt, and the pelt to the HSF, but I'm not sure how much it would decrease the potential performance of the pelt.
 
Celeys and Durons...........reach about 37W max output@stock, so an 80W with a big Tornado would work
 
I have mentioned this before and will again just briefly. I have not had the time to do it but I would think that one could pull off dual 80w tecs (run off added psu's to the case) on a verticle cold plate on air. This I suspect would prove to be a much more efficient setup and would get one some rather extreme cooling given the wattage of the tecs.
 
Now, I do not know if this will work or not (my geuss is that it will). You can picture a regular cold plate 1/2 thick, right? Well cut it down to 2x2 and instead of laying it down flat on the cpu, put it up on its' edge with the 1/2 edge on the core. Now try using 2 80 watt pelts (one on each side) with two copper 1U Jag heat sinks on each tec. You will have to change the fans as they will not handle the heat load. Go to a couple of 30cfm or better 60mm fans. You now have a sandwich. The tecs (much like my water cooled setup) will try to overlap each other and at the junction of this overlapping will create an intense cold area. This will help keep the hottest part of the core (much like a flame) from reaching very high and thus kill much of the insulating heat that the core needs in order to penetrate. As I said in the beginning, I have not built this as I go from project to project however I have built this setup using tow 156w water cooled tecs and it has proved effective. The nice thing about the above setup is that it is concievable that three psu's or 2 (one being powerful) could be placed into a case with a little creativity and thus the whole thing would be within the case. Think of it, one case, no exterior devices and enclosed within its walls are a verticle plate with dual 80w tecs, air cooled...powering an xp1600 to .....oh say ....2.0Ghz. Maybe, maybe not, but definately a fun project.;)
 
I thought about cooling my tualeron with a 80W pelt + air cooled HS but after doing some calculations, I figured the gain in temps would be ridiculous. That's because the C/W of air cooled HS are just too high to dissipate all the heat generated by the peltier+cpu. Remember that although these procs only generate around 30W stock, they go up to almost 50W when OCed. You have to add the 128W (around) that a 80W peltier will generate and you get around 180W. That's way too much for an air-cooled solution. The best HS out there (SLK800) has a C/W of 0.23 so you'd get temps of around 35C. With a WC solution (C/W of around 0.1) you'd get much better temps of around 12C. With a SLK800 without a pelt you'd get temps of 32C. (that's all assuming an ambient temp of 20C). As you can see the theoretical temps from using a pelt are WORSE than those using a normal air-cooled solution. Also you might have a hard time mounting the SLK800 with a pelt while most WB come with the option of a special mounting for a pelt. Also the SLK800 needs a 5500rpm 80mm fan to achieve a C/W of 0.23 which puts out an ear splitting 55dBa while a WC solution is much more silent. Hope I've been of some help :D
 
Humm I have a duron 1.1 here. Anyone have some pelts for me to play with? Id like to try Silver's idea.
 
Tamasha_Strife said:
I thought about cooling my tualeron with a 80W pelt + air cooled HS but after doing some calculations, I figured the gain in temps would be ridiculous. That's because the C/W of air cooled HS are just too high to dissipate all the heat generated by the peltier+cpu. Remember that although these procs only generate around 30W stock, they go up to almost 50W when OCed. You have to add the 128W (around) that a 80W peltier will generate and you get around 180W. That's way too much for an air-cooled solution. The best HS out there (SLK800) has a C/W of 0.23 so you'd get temps of around 35C. With a WC solution (C/W of around 0.1) you'd get much better temps of around 12C. With a SLK800 without a pelt you'd get temps of 32C. (that's all assuming an ambient temp of 20C). As you can see the theoretical temps from using a pelt are WORSE than those using a normal air-cooled solution. Also you might have a hard time mounting the SLK800 with a pelt while most WB come with the option of a special mounting for a pelt. Also the SLK800 needs a 5500rpm 80mm fan to achieve a C/W of 0.23 which puts out an ear splitting 55dBa while a WC solution is much more silent. Hope I've been of some help :D

Sound thinking and you will find little disagreement from me. We do however know (as Donny paycheck proved) that an 80w can be cooled on air. Though it is about the max. Dual opposing tecs on a verticle plate would however change the landscape some and acceptably cold temps might be achieved through this means. I do concede that some creativity would be required in its implimentation as you would have to work within the constrains of the area allowed around the socket. The nice thing is that the 1U coolers are copper and adapting them to allow for bolt downs using the existing mount holes would not be too difficult. Brass screw adapters could be soldered onto the heatsinks allowing for usage of the full area.

Rough first estimates would be 1 5/8 width on each heatsink/fan setup, 3/8 inch cold plate plus 1/4 for both tecs. Total width would be 4 1/8. This is going to be a tight fit and some careful measurement and modification will have to be done to make it work but it is still within doable range. Careful modification would bring this under 4".
 
Last edited:
What do you think the temps would look like, with a set up like this on a celeron or duron?
 
{PMS}fishy said:
What do you think the temps would look like, with a set up like this on a celeron or duron?

First, realize that you are entering into uncharted waters here. I have found this setup on water to allow me to achieve oc's close to some very extreme measures, short of LN2 and am therefore very pleased with the results. I attempt to keep temps somewhat elevated so as to keep condensation to a minimum. I do not think that temps have improved all that much however the form of the heat 'mushroom' within the cold plate has been altered and this has allowed for a higher oc. For example, lets assume the temp has not changed at all however the absorbtion of heat from the processor core has been dramatically improved. Thus though the processor is booting up at roughly the same temp as before, the ability of it to achieve higher temps under load has been somewhat negated allowing for a higher stable oc. Kind of make sense? At any rate, I would suspect that you would get a very nice oc on this setup and far in excess of anything achievable on air. Actual temps could be controlled to a degree by core voltage thus negating condensation while achieving a 'safe' stable oc. Extremely low temps (I am learning) is not as important as the ability to absorb energy and to control its form as pictured within the cold plate. Lowest temp is already defined within the tec specification and is really not nearly as important as the tecs ability to transfer energy. Heck if ambient alone could be maintained then max post speed would be max oc (and max post speed on some chips is rather high). I am sure that there are engineer types that could possibly explain this better as I am working with limited educational levels here and rather logic and first hand experience. Does not really answer your question and I realize that.

Do realize that on my setup some said it would not work and a select few thought that it might. Logic dictated to me that it would. (You have no idea how many nights I lay in bed picturing the tip of a red hot needle and the best means by which to cool it-energy being constant.) Some thought the tecs cooling one another would not work (fail to see the logic in that one), some thought the small water blocks would not cool the pelts (knew the melting point of the solder and suspected the bong where well in excess of that especially given the increase in cooling performance as the temperature increased). BTW, I now have dual bongs in place and would expect a small increase in oc due to the increase in tec cooling efficiency. At no time did I feel the tecs would be in jeopardy. Probably the most important thing in an aircooled experiment of this sort would be whether or not you could keep the soldered leads on the pelts below thier melting point through the use of the jag/fan heatsinks. Were it I, I would most likely purchase a 1u and a high flow 60mm and play with it on a stock setup (with duron) in order to iron out the cooling of the pelt before purchasing the second setup and setting up the verticle plate. Drop a thread on this idea in cooling and pay attention to the engineering types, especially the ones that think it might work. There will be many who will purport to know what they are doing but in fact are more interested in what they have to say than the idea. Pay attention to those that really sound to you like they know something and are willing to conribute to the idea. Remember standard ideas will get you standard results. Think ouside of the box and pull in those that know more than you. Just direct the thinking to achieve the desired goal and this idea should work. Big question is whether or not whether the 1U with the proper fan can dissipate the heat. Wattage off the cpu (I would think would be roughly split between the tecs so do not buy off on the single pelt single cpu wattage idea as this would most likely overwhelm the heatsink and I believe would be inaccurate). If the jag (with the proper fan) can handle the wattage on a single cpu (regardless of how successful the oc or temps) then the experiment should be a go as the heat load should not increase but the cooling ability (ability to absorb energy) would be dramatically increased on the dual opposing setup.

One more point. Dual tecs on a conventional coldplate are not nearly as efficient as they are attempting to cool the entire cold plate (which is more or less achieved - with the accompanying condensation) however at any given point in time the heat mushroom within the plate is not extending more than about 3/4 of an inch from the center of the core in a v type pattern (my belief only and held by one other engineering type as well here). This being true, a second tec on a conventional cold plate is attempting to cool the entire plate with very little impact on the core mushroom within the cold plate. Yes it narrows it a touch and thus you get a touch more of an oc but not much. Overlapping cold arcs as naturaly contained within opposing pelts makes sense as it is well within the range of the overlapping arcs of the tecs creating a usasable area of much more cold intensity, thus altering the mushroom and bringing into play the sides of the tecs which continue to overlap.
 
Last edited:
This sounds like a really interstring experiment. Im off to do some sketching and maybe some CAD. Ill take some measurments and see what dimensions I have to work with. Sounds like a good project.
 
I have thrown in a couple of edits, sorry. Get thinking and ideas keep coming to mind. The thinking (I believe) is sound and has proven itself on the dual watercooled idea. Take Care and I highly recommend you through up a thread and direct it so it does not get off track. I believe that there are others that would love to help (in a serious fashion), just don't let some take the thread where you do not want it to go. I.E. keep it on track. Take Care and if I can be of any assistance.

If you get stuck, might use these thoughts. Candle, yellow is insulation for blue (intense heat). Candle represents xp2200 heat profile. Stove, Lower height but wider area, represents xp standard. Duron, stove on low. Kill the yellow, (insulator) and the profile of the blue must change. XP2200, profile is candle, intense over a small area. Picture tec, amount of tec cooling area brought into play on a small intense area? Tec is taken out as cold plate (standard setup) is penetrated and small area of tec must cool intense heat. Must bring full 48mm of tec into play to achieve cooling. Idea, overlap tecs. Double cooling in overlapped area and candle can not penetrate tecs. Result profile changed to small area at bottom of tec where arcs do not overlap and heat spreads out. Insulation is narrow immensely. Blue is very low and stove type situation (on low flame) is achieved.
 
Last edited:
Silver, I followed your thread with great interest although I did not contribute anything at the time. Basically you divided the thermal load between the 2 WB. If we did the same thing with 80W pelts and dual SLK800 we could lower the temps down to 7C (that's assuming the 50W the CPU puts out is split equally between the 2 pelts and that the special cold plate is 100% efficient). I'm not sure what the C/W on the U1 coolers you're talking about but I doubt it's as good as the SLK800. Overall I think it's a nice project but a WC TEC would be more efficient and be in the same price range. I'll be happy to try and help whoever wants to try this, who knows I might be dead wrong...
 
The heat sink is limited by the area he has to work with and that is why he would be looking at the most efficient (yet lowest height) heat sinks he could find. Width will become an issue with ram and capacitors on the side of the socket. The real question to the whole thing is whether or not the 1u (with the fan changed to a higher flow) can handle the heat load of an 80w tec. The thinking after that I find to be very sound. I am presently using dual opposing water cooled 156w tecs on 2" water blocks and even pumping from one water block to the next the tecs could not overwhelm the bong. Ran about 3c over ambient on the water temp. Please, if you have ideas then do join him in the link he posted above. As far as I know this (on air) has never been done and should he pull it off I think he would be rewarded with very good (far beyond normal) air cooled tec results.
 
Basically what we need to know is what the C/W of those U1 coolers is, from there it's pretty easy to figure out wether it's worth a shot or not.
 
Last edited:
Back