PDA

View Full Version : Air Cooling Stacked Pelts.


Kunaak
09-14-2004, 04:48 PM
I have been testing a few of my Pelts with Air Cooling.

I have my Zalman 7000 CU, a 110 watt pelt, and a 80 watt pelt to play with.
I used a barrier box and a regular 550 watt Antec PSU, so I have 12 volts and 5 volts available to power the Pelts depending on what barrier box I attach the pelt to.

when I use the 110 with the 12 volt line, it's very cold and gets frosty in about a minute.
when I use the 110 with the 5 volt line, it gets kinda cold, but not icy.
the heatsink on the other hand doesn't even get remotly warm with either option.

when I use the 80 watt pelt on the other hand, with the 12 volt line, it's fairly cold, and gets frosty in about 2 minutes.
when I use it one the 5 volt line, it's cold to the touch, but little else.
the heatsink alone doesn't even get warm.

last, when I use the pelts stacked...
the perfect combination seems to be.

heatsinks.
12 volts on the 110 pelt.
5 volts on the 80 watt pelt.
cold plate.

this combination seem to get incredibly cold, cold enough to numb my hands in a few seconds. it also gets frosty in about 10 seconds like this.
this heatsink isn't really that warm like this either.

my question is, how many watts of power is this combination putting out??

saaya
09-14-2004, 07:41 PM
putting out? like how much heat does the zalman heatsink have to remove? 160-190W if its a 12v vmax pelt and is running at 12v.

thats really a lot... what do you mean with the heatsink doesnt get really hot? it should actually be hot, actually VERY HOT ;) prescott=100-120W , this is 160-190W!

its either not a 110W tec, or it has a vmax of 24V, or its damaged, or the psu volts drop to 10v and it has a vmax of 16v... but even then it should get much warmer...

weird...

so when using both tecs at 12v the 80W one cools better than the 110W? then the 110W is def fcked up or it has a vmax of 24.

Kunaak
09-14-2004, 09:18 PM
no the 110 definatly cools better then the 80.
the 110 is a industrial pelt from Tekra. has "110 watt" printed on it, and even tells me what side is hot and what side is cold :)

but no, the Zalman also, does not get hot, barely even warm to tell you the truth.
with the stacked pelts it gets alittle warmer, but nothing I'd call hot.

I used to have a Swiftech MC462 with a 229 watt pelt, with a Meanwell PSU, and I can tell you, that thing was burning hot, hot enough to burn your fingers if you touched it. this is nothing like that.
even with 2 pelts on there.

saaya
09-15-2004, 09:35 AM
theres something wrong... i mean think about it... the zalman gets warm with a northwood or a64 wich both produce less than 100W under load, but it doesnt get even warm with a 110W tec under it wich produces at least 160W?

thats impossible... there must be something wrong with the tec, again, what volts is it rated for? can you meassure how many amps it is pulling?

Kunaak
09-15-2004, 04:49 PM
ok, one thing I forgot to mention, this is just testing.
on air, not on a CPU.

(I mentiond that the first time I posted this)

saaya
09-16-2004, 04:35 AM
yes, but it doesnt matter.

a 110W tec always produces 110W of heat itself and always pumps between 1/2 of that to 3/4th of that of heat from the cold side to the hot side, so you will ALWAYS have 160-190W of heat on the hot side of that tec unless you insulate it.

and an 80W tec pushing at least 60W of heat to the cold side of the 110W tec isnt really insulating the cold side ^^ even if it would be very unnefficient at 5v and only push ,lets say ,15W of heat to the cold side of the 110W tec (very very very unrealistic) then the 110W tec will still have 110W+15W on the hot side the zalman has to remove.

thats a fully fledged prescott there... impossible for the hatsink to be just mildly warm if warm at all... hmmmm i dont get it...

plus you say the heatsink doesnt even get warm with the 110W tec hooked up on it alone... i dont get it...

yes there was even an air cooled 227W tec from swiftech, but that heatsink had like double to 4 times the surface area of the zalman 7000 cu and it was using an 80mm tornado and it was getting cooking hot!

so that heatsink could manage to dissipate 400W rather bad than good, so i think this prooves again its impossible to cool a 110W tec wich produces almost half the heat of a 227W tec with 160-190W with a 7000 cpu and have a cool to slightly warm heatsink.

there must be really something wrong with the tec if the heatsink doesnt get hot.

this is confusing me a lot...

assuming the 110W tec pumps no heat at all then we are talking of 110W of heat, even then the heatsink should be friggin hot!

but it HAS to pump all the heat from the 80W tec at 5v, if it wouldnt then that tec would overheat and die in a matter of seconds or minutes with some luck, and the cold side of the 80Wer wouldnt get cold but heat from its hot side that doesnt get carried away would diffuse back to the cold side...

it all doesnt make sence... did you really try the tecs or was this all a theoretical question? :D

Kunaak
09-16-2004, 06:07 AM
haha...

do you want me to send you my pelts and heatsink to try this yourself??

you know me, I don't just think up stuff and hope it works.
this is something I did.

also, no, the heatsink doesn't get even alittle warm until you stack the pelts.
each pelt alone, barely does anything to heatsink alone.
but stack the pelts, and turn it on, and it gets super cold in seconds, while barely doing anything to the heatsink.

saaya
09-16-2004, 06:18 AM
well yes, stacking pelts of course means lower temps, BUT!!!!! the heatsink that cools the 110W should always have the same temp! tecs always pump the same amount of heat, unless you insulate them of course :D

so it doesnt make any ence at all, this is making me mad :D because it doesnt make sence following the usual tec rules...

Kunaak
09-16-2004, 02:08 PM
I don't know the pelt rules.
I am not that into them, they are a rare thing for me to play with, I always use waterchilllers or phase change or dry ice.
I am also no expert on electronics, but heres my thinking.

take the 80 watt pelt.
I run it with 5 volts, I know most pelts run best about 12-15 volts, running it with only 5 volts should mean it's putting out less cooling power.
less cooling, means less heat.
I doubt either pelt is running at full wattage.
thats why I ask...

how much power does a 110 watt pelt use at 12 volts?
how much power does a 80 watt put out at 5 volts?

saaya
09-16-2004, 07:54 PM
depends what the vmax of those pelts is.

i assume the 80W tec has a vmax of 12v and the 110W a vmax of 16V.

but that still wouldnt explain it. even if the 110W tec is a 24vmax tec and is running at 12v its producing at least 80W of heat total, and theres no way the zalman can remove 80W without getting warm.

to answer your questions i need to know some more things, please meassure how much amps each tec sucks at 12 and at 5v and check whether the 12v or 5v rails drop while you are running the tecs.

volts and amps, thats all i need to guess how much heat they pump and produce :)

Kunaak
09-17-2004, 12:25 AM
well, if you tell me how to measure amps I will.
I've never measured anything relating to pelts.
I have a volt meter and know how to use it, but I don't know how to measure pelts amps.

also...
theres tons of CPU's that put off more then 80 watts of heat. even some AMD Xp's.
but even the lowest 10 dollar heatsinks will cool them without getting warm.

I have the Zalman on my A64 3700 right now at 2.7 ghz and it's not even remotly warm.
surely a 3700 at 2.7 ghz and 1.6 volts puts off more then 80 watts of heat...

but even now, the heatsinks cold to the touch.

saaya
09-17-2004, 12:51 AM
your a64@2.7ghz and 1.6v is cooled by the zalman cu 7000 and the heatsink is cold? :confused: in idle or what do you mean?

about meassuring the amps, use your multimeter and plug the red cable into the unfused slot, then connect the multimeter psu and tec like this:

feed (12v/5v) -> red wire of the multimeter
black wire of the multimeter -> red wire of the tec
black wire of the tec -> ground (psu black wire)

most multimeters have cables that can stand 10A officially and 25A for a short period, so there shouldnt be a problem. turn off the psu after 10 seconds to prevent the cables from running hot.

Kunaak
09-17-2004, 01:59 AM
yep, heres my zalman on my 3700 at 2.7 ghz at 1.6 volts.

the screen shot from speedfan is just to show it does cool very well. this is idle of course, but as you can see it does pretty well.
it's not exactly warm in alaska at the moment, nor is it all that warm in my house, but I am pretty sure I can tell if a heatsink is warm by touching it, especially if it's cold in my room.

speedfan doesn't report the 5 volt line correctly, but all the other voltages and temps were checked against MBM5, and they matched.
plus, the temps on sensor 1 go up when I do super pi tests so thats how I confirmed it was the right one.

it is about 40 degrees out right now, and my house is alittle drafty, because the temps are normally arond 32c, 28c is alittle low in my opinion.
but whatever...

just trying to show, the Zalman does cool very well.

saaya
09-17-2004, 02:07 AM
you call 40C cold to the touch? :P

and that 30C is idle temps, right? what are your load temps? and why is your mobo 5C warmer than the cpu 0_o guess the mobo probe is set up weird, same here on my infinity, sometimes the mobo is 10C hotter than the cpu :lol:

Hiwayman
09-17-2004, 03:53 AM
Stacking Pelts is a problem area:

The critical reality here is that the second TEC must not only pump the heat from the thermal load plus its own internal power dissipation (I2R), but it also must remove the heat dissipated within the first TE device.

dTload = (1 - (heat load/max cooling power)) * max temp difference

where heat load = CPU cooling requirement and max cooling power = peltier rating and max temp difference = peltiers max dT under no load.

Let's see what a single peltier setup for a 60 watt CPU and 72 watt peltier with a Delta T of 65 C can deliver:

dTload = (1-(60/72)*65 = 10.8 C


or with 2 pelts:


Peltier 1: dTload = ( 1 - (60/72) ) * 65 = 10.8 C
Peltier 2: dTload = ( 1 - (132/72) ) * 65 = -54.2 C

Uh Oh...not good. The second peltier is actually overwhelmed by the first one, because it has to cool the CPU AND the first peltier. This is why many folks are seeing temps rise when they attempt to stack peltiers.


I tried using a 226W stacked with a 122W and it seemed to work but not that well.

Kunaak
09-17-2004, 03:58 AM
you know saaya...

I feel like you sitting here saying everything except directly calling me a liar.
so I am gonna abandon this topic.
I didn't start this topic to be harassed, and so far, thats all I feel you've done.
I am not gonna open this topic again, so say whatever you want in here.

see ya.

Hiwayman
09-17-2004, 04:29 AM
Well, there is more info here as well:

http://www.overclockers.com/tips313/

saaya
09-17-2004, 12:57 PM
you know saaya...

I feel like you sitting here saying everything except directly calling me a liar.
so I am gonna abandon this topic.
I didn't start this topic to be harassed, and so far, thats all I feel you've done.
I am not gonna open this topic again, so say whatever you want in here.

see ya.

no! im sorry if it sounded like that. :(

im very curious about this as i feel my tec knowledge challanged :D i dont know THAT much about tecs, but so far it all made sence with the few rules i knew, now this doesnt. im really confused by it, ive talked to other people and they dont understand it either and couldnt help me to explain it.

im sorry if i might have sounded harsh :toast:

please post your results from meassuring the tecs! i cant wait to solve this riddle :D


and hiwayman (cant write your real name lol :D) , stacking tecs can give great reults, ive seen really amazing temps with stacked tecs, but yeah, its really hard to tweak and as soon as you go only close to the limit the tecs can pump your temps are up and its all for nothing... :/

Andrew
09-18-2004, 06:06 PM
saaya, you have to remember that the guy is in Alaska.. I think that he was giving the outside (outside the house that is) temp as 40F.

saaya
09-18-2004, 06:26 PM
40F? thats 5C... hmmm that would make sence... but why didnt he mention that? but even with 5C ambient temps the heatsink should be HOT with around 180W under it... hmmm i think its a vmax 16v tec so its probably only putting out 120W of heat at 12v, and with 5C ambient the heatsink might be in themid 30s or maybe around 40C. i wouldnt call that cold to the touch, but you could call it cold for a heatsink with a 110W tec under it.

sandman
09-19-2004, 06:20 PM
My 7000-CU the base barely gets warm on my A64, that's at 2.2ghz 1.6Vcore and my cpu temps are around 45C this is full load all the time as I'm folding.