172w Pelts

[Player0]

Hi guys, hope this is the right forum. Anyway, I'm looking for people with experience direct mounting 172w peltiers on the CPU. Im uprading my chiller system from 3x156w pelts to 4x172w peltiers, and also want to add a 5th peltier to the waterblock itself.

The PSU im getting is 24v 65a, which is more than enough for 5x172w peltiers. So i could put #5 on the XP2000+ (soon to be 2200+). Of course, I havent seen anyone use the 172w pelts since the 226w units came out. Does one 172w unit have enough Qmax to handle the XPs heatload?

The other alternative is running a 226w peltier on the CPU, powered by the Antec 550w PSU. Im just not sure if that PSU has enough current to do this. Also, at 12v, the 226w peltier only has a Qmax of about 180w...not much more than the 172w unit actually.

I hope my chiller blocks don't melt :) I'll be cooling this with 4x Black Ice Extreme radiators.

[DaGooch]

Extremely complex setup configuration there. To answer just one of your questions, the 226W Peltier junction requires approximately 25A at 12V and if the Antec is rated for continuous duty at that load, it should not be a problem. Personally, I would prefer to use a dedicated 12V or 13.8V external power supply.

[Bravo]

Why use expensive BIX cores when a decent heatercore, or two for a 1/6th of the price would do the job?

[Bravo]

'Fox, get this guy off me!'

heheh, sorry, couldnt resist :(

[Player0]

Gooch, actually...at 12v the 226w peltier only draws about 19.5amps, because it is only running at 80% voltage. Its also only running at 80% cooling capacity, 180w or so.

The problem is, I will be swapping out my 12v dedicated powersupply for a 24v unit a friend is hooking me up with. I ain't got no more room in my case for a 3rd powersupply just for the 226w.

The most important question is...will 170-180w of heat pumping be enough to handle a highly overclocked XP at like 2.2vCore? Or am I better off just not using a peltier directly on the core and focusing on the chiller portion?

Surely someone has a 172w running on their XP somewhere :)

Majestik, yeah...your probably right about the heatercore thing. The BIX itself actually IS a heatercore, just modified with the shroud and proper fittings I guess. Im not sure a heatercore would necessarily work better, unless it was actually bigger. Figgure I spent over $200 on the four radiators...which is probably silly too. It just seemed easier to me I guess. I had gotten one, it outperformed all my other radiators, AquaCoil, cooling cube...and was way small.

So yeah, thats why I spent 2-3x more for the BIXes...has all the mounting and shroud already, and is pretty much plug and play for my cube :) Or, im just dumb...i dont know lol :)

[OPPAINTER]

I tell you Player these 220 pelts with chilled water jam.

Idle temps with water at -6c cooling the pelt.

[dmitriyaz]

while we're at it,
OPP,
i see you MoBo temp is 33', is that measured in the north bridge?
the question is, is it even worth to supercool the NB? or will it do just the same or only a bit better compared to air?

[OPPAINTER]

Originally posted by dmitriyaz
while we're at it,
OPP,
i see you MoBo temp is 33', is that measured in the north bridge?
the question is, is it even worth to supercool the NB? or will it do just the same or only a bit better compared to air?

Yea my room is frying, about 29c no breeze:D
I"ve never tryed super cooling the NB but cooling it does help. Like when my system is at 30 plus I have a tough time with High fsb. I don't think super cooled would do much better then just real good cooling though.

OPP

[dmitriyaz]

Originally posted by OPPAINTER
I don't think super cooled would do much better then just real good cooling though.

OPP
thanks OPP :)
real good cooling--do you mean a huge heatsink+fan? or water? or a little pelt action?

[OPPAINTER]

I'm thinking a Good heatsink and a Good fan:D

OPP

[dmitriyaz]

gotcha :)

[Player0]

I watercool the NB on my KX7...doesnt seem to have any added benefit.

Opp, is that 220w on the CPU a full 15v or only 12v?

What waterblock is that?

[OPPAINTER]

I'm using a custom waterblock that Eli made me, must wiegh 3bls. It has a 1/2" cold plate and is a spiral interier.
When I first started testing the 220 pelt a ways back from 12-15 I found that 12v was the best at the time, then again my water was warmer back then so It may be more effiecient now with higher voltage with this chilled water. I haven't tested with the chilled water, I still run it at 12 straight up.

I don't know if MBM or the mobo probe will read below -30, it stayed there for a while but didn't get colder and it was going up real quicktoo, umm.
Coldest I've ever had it :D

[mackanz]

Thats freezing!

LOL!
Opp, you said you run the 220W at 12V, i am getting a 220W soon, and my q is, can i run that pelt on a separate half-cheap psu?
I have 340W Antec lying for no use and i though that might come in handy for a pelt or 2.
I also have 2 smaller pelts at 80W, would one of those be of any use for the G4 or can it be of use on the cpu even? They run perfectly on 12V by the way.

Any tips?
I´m not reaching for North-pole or anything, but colder than just plain water.

Thanks,

Marcus

[Player0]

Thats some impressive temp Opp. Yeah im sure the thermsistor flakes out so far out of spec like that.

I guess I will run 4x172w pelts on the chiller then, and the 226w pelt on the CPU (but only 12v) since the 180w seems to be enough.

I just hope my cold water temperatures stay cool enough to benefit the GF4, NB and Ram that im currently chilling.

Need to decide on a waterblock though for the pelt sandwhich. Trying to get my hands on some review samples so I can figgure out just what really is best.

[dmitriyaz]

Marcus,
tell me what is the Vmax and Qmax of the pelt you want to get, tell me what voltage will you be running it at (probably 12v), and what PSU do you want to use for it.
and i'll tell you if its sufficient or not, by using the Ohm's law.

[OPPAINTER]

Originally posted by Player0
.

Need to decide on a waterblock though for the pelt sandwhich. Trying to get my hands on some review samples so I can figgure out just what really is best.

Talk to Gabe at Swiftech, he's got a direct water to the Pelt block he's testing right now. In other words the pelt is cooled by direct water contact. With chilled water on the pelt it should be a winner. I asked for one the other day but he only has the one proto type. Maybe you can talk him into it:D

OPP

[Player0]

dmitryaz, how does ohms law figgure in to finding a powersupply for the tec?

Opp, direct cooling the peltier sounds like a good idea. As long as it doesnt leak or anything :) I wrote gabe seeing if he could hook me up a couple days ago, I hope we can work something out. Did you ever have a chance to test out Leufken's direct cooled pelt system?

Im still surprised Gabe doesnt use some sort of spiral design to help keep the water in the block longer, rather than relying on turbulance to keep the water in the block. His waterblocks are still upside down, with the majority of the metal mass actually above the gasket, which reduces thermal transfer surface area even more. When I removed the gasket in exchange for thermal epoxy, and pitted the inside of the copper base plate on the old MCW462, I actually picked up a few degrees improvement. Maybe the new MCW works better...need to try it.

[dmitriyaz]

Player,
Ohm's law: voltage = resistance x currect.
you take the Vmax (voltage), the Imax (currrent), and find the resistance of the pelt using the Ohm's law.
now that you know the resistance, you take the voltage that you will be applying to it, in our case 12v, and calculate the "new current".
like so:
the "new current" = the new voltage (in our case--12v) / resistance.
than you take that "new current" and times the new voltage (12v, in our case), and get the wattage that the pelt will be consuming. (it will be more than how much heat the pelt will be removing. pelts are inneficient, remember?)
now, to see if your PSU is sufficient, compare the maximum current for the 12v line (should be written on the side of the PSU, usually 10 to 25 amps); you compare it to that "new current" number that we we got from using Ohm's law formula. if your PSU's number is bigger, it is sufficient to power the pelt.

i should include this along with some other stuff in the "guides" section :banana:

[Player0]

Qmax is not a measure of the current. Qmax measures the amount of heat (in watts) that the peltier can transfer from hotside to coldside at Vmax.

Imax is a measure of the peltiers current, and you can use this to figgure out the draw of the peltier, although it has nothing to do with ohms law.

Take the 226w peltier. It has a Vmax of 15v and Imax of 24a.

To find the current draw at 12v, first find the percentage of power decrease:

15v - 12v
~~~~~~
15v

= 20% decrease.

So, 24a * 80% = 19.2amps

The 226w peltier will draw about 19amps at 12v. So your PSU needs to have a rating of at least that (20amps to be safe) to run the peltier.

Do note that it IS only a rough estimate, as because of the complex nature of the peltier, the amp decrease is not completely linear with the voltage decrease. Tedist.com has provided me with peltier performance graphs in the past from Kryotherm. These included a graph that charted voltage/current measurements. Using one of these charts, or actually measuring the current draw, is the only completely accurate way of knowing the exact amperage usage of the pelt at non-Vmax voltages.

But, like I said, the curve is not that steep, so using the simple efficiency equation gets close enough for most people.

[dmitriyaz]

of course its Imax, that was a typo, i corrected it.

now back to the topic:

Player, my way to calculate it IS EXACTLY the same as yours.
we will get the exact same results for every case.
except i explained in detail how it works,
and you explained it in a nutshell.
it does, indeed have something to do with Ohm's law. what you said IS the Ohm's law, only simplified for this particular case.

[Player0]

dmi,

I realize that the results are the same, but like I said these numbers can only be used as a rough estimate. The pelts resistance actually changes with temperature, which is why ohms law doesn't apply for TECs.

[dmitriyaz]

yeah, your right, Player,
Ohm's law applies to conductors, and TEC's are semiconductors, aren't they?
as far a efficiency goes, we can say nothing,
but for the power consumtion but we can still, as you said, give an estimate.
do you know where i could find the graph of V. vs. I vs. R vs. Efficiency ? or something like that for the peltiers, expecially the 226w one?

[DisposableHero]

im using a direct water touch pelt block.. leufkens cfi block... with just a heatercore and 2 120's i get around 18-19 idle and 23+ load.. i dunno if this is good or not for a 226w pelt at 13.8v on a 1.9 chip with 1.85 vcore

[dmitriyaz]

oh yeah, you'll be aight :)

[DisposableHero]

Originally posted by dmitriyaz
oh yeah, you'll be aight :)

you talking to me? lol

[dmitriyaz]

yep :D

[DisposableHero]

i saw in another post talking about flow with pelts and they said the higher the flow the better especially with pelts.. well.. my 350gph pump is probably 200 gph or less now with this heatercore..

i should probably get a better pump?

[mackanz]

Originally posted by DisposableHero
im using a direct water touch pelt block.. leufkens cfi block... with just a heatercore and 2 120's i get around 18-19 idle and 23+ load.. i dunno if this is good or not for a 226w pelt at 13.8v on a 1.9 chip with 1.85 vcore

Hero,

Are you using neopreme for condens caution?

Marcus

[dmitriyaz]

seems weird to me....
heater cores don't usually restict flow,
are you sure thats a heater core and not just a radiator?
and in any way, i'd say its something else than that...
what size are the original fittings on the pump?
what tubing are you using?
what are the complete specs or your watercooling system?

[DisposableHero]

Originally posted by mackanz


Hero,

Are you using neopreme for condens caution?

Marcus

yeah and have dielectric greese all over!


Originally posted by dmitriyaz


seems weird to me....
heater cores don't usually restict flow,
are you sure thats a heater core and not just a radiator?
and in any way, i'd say its something else than that...
what size are the original fittings on the pump?
what tubing are you using?
what are the complete specs or your watercooling system?

well its a radiator for a GM truck.. fits 2 120mm's perfectly... well going from a BIX to this rad. my flow has dropped bigtime..

i dont know what orig fittings are but i made it 1/2..
i just have some cheap tubing from homedepot.. crimps easy so i watch out for that...

umm i have a danner 350gph pump.. submersed cuz its leaking like crazy.. so that could add 1C or so i guess...

[mackanz]

Then i guessed right about the proofing, then.
Another question, will there be a lot of difference if i have the pump above the reservoir or below it?
Normally below it would be the better but it would be much handier if i could have it above it for casedesign purpose.
Its a Eheim 1250 btw.

Marcus

[dmitriyaz]

Originally posted by DisposableHero
... well going from a BIX to this rad. my flow has dropped bigtime..
your heater core is clogged, my friend :)
get another one, and you'll be aight :toast:

[DisposableHero]

just have it about even with everything... this stuff wouldnt really matter THAT much...


btw guys.. here is my rad before i modded barbs and put shroud around...

http://blitzkriegwars.netfirms.com/P4110012smaller.jpg

[DisposableHero]

Originally posted by dmitriyaz

your heater core is clogged, my friend :)
get another one, and you'll be aight :toast:

is there any way i can "unclogg" it? i got this from autozone.. i think it was "new"

also.. ive always heard that heatercores restrict flow bigtime..

[dmitriyaz]

"new", huh?
if it was new, i can't tell you it's clogged, that seems unlikely...
the barbs that were originally on it were bigger than 1/2", right?
to unclogg it, try some hardcore chemicals...talk to the wal-mart people, they know more about that than i :)

[DisposableHero]

Originally posted by dmitriyaz
"new", huh?
if it was new, i can't tell you it's clogged, that seems unlikely...
the barbs that were originally on it were bigger than 1/2", right?
to unclogg it, try some hardcore chemicals...talk to the wal-mart people, they know more about that than i :)

hmmmm... i guess i could try some drain uncloggers or something lol..

just in case.. adding to my watercooling specs.. my water is usually around 37-39 C

[dmitriyaz]

the temps are pretty high...there is definetly more room to tweak it :D
and:

Originally posted by dmitriyaz
the barbs that were originally on it were bigger than 1/2", right?

[DisposableHero]

Originally posted by dmitriyaz
the temps are pretty high...there is definetly more room to tweak it :D
and:


yes on the pump and rad the barbs were bigger

[dmitriyaz]

Originally posted by DisposableHero


yes on the ... rad the barbs were bigger
thats good :)


Originally posted by DisposableHero


yes on the pump ... the barbs were bigger

there we go, found a problem :( ; the pump was designed to do 350 gph with those bigger barbs, by putting the smaller ones in, you restricted the flow. but don't think that the bigger barbs/tubing will solve the problem; your channel is only as wide as the tightest spot. so if your pump used big fittings, and the waterblock used smaller ones, the only way to resolve that is to either buy a new pump, with 1/2" fittings, or make a custom waterblock with very wide channel and barbs.

[DisposableHero]

Originally posted by dmitriyaz


there we go, found a problem :( ; the pump was designed to do 350 gph with those bigger barbs, by putting the smaller ones in, you restricted the flow. but don't think that the bigger barbs/tubing will solve the problem; your channel is only as wide as the tightest spot. so if your pump used big fittings, and the waterblock used smaller ones, the only way to resolve that is to either buy a new pump, with 1/2" fittings, or make a custom waterblock with very wide channel and barbs.

oh ok i see what your getting at.. since the inside of the rad and pump are bigger than 1/2 the flow has been cut down.. but when before with my BIX the inside of that was for 1/2 = more flow...

well a more powerful pump should do it right?

[dmitriyaz]

Originally posted by DisposableHero
well a more powerful pump should do it right?
not exactly...
when you are making a pump run on the channel smaller than the spec., you are torturing it, and its gonna eventually die. make sure that if you do buy a pump, its 1/2". its not too hard to find. Rio pumps are dirt-cheap, and for $50-$60 you could find one that does 500 gph and has 1/2" fittings. but it can't run in-line, only submersible.

[DisposableHero]

i havent heard much about rio pumps... i see they're cheap after a quick search... 48 bux for a 999 gph pump


edit: http://blitzkriegwars.netfirms.com/mbm5pic.jpg

[mackanz]

Originally posted by DisposableHero
i havent heard much about rio pumps... i see they're cheap after a quick search... 48 bux for a 999 gph pump

999GPH...overkill?
Or a typo?

[dmitriyaz]

yeah...its a cheapass pump,
its no EHEIM,
but it'l do.

[DisposableHero]

Originally posted by mackanz


999GPH...overkill?
Or a typo?
no typo ;) http://www.aquadirect.com/catalog/pumps/taam.htm

was just showing how cheaply priced they are

[mackanz]

Fireki´n e´ll!!


Mac

[DisposableHero]

Originally posted by mackanz
Fireki´n e´ll!!


Mac

:confused:

[mackanz]

ehh....kinda.... jesus crist!
But in slang english...

[DisposableHero]

meaning good pump or bad?

[dmitriyaz]

its, well, um, good.
yeah...

[DisposableHero]

Originally posted by dmitriyaz
its, well, um, good.
yeah...

lol.. oh thats really reasuring... well one thing is for sure.. im not paying 120+ for the ehiem or whatever its called

[mackanz]

Well i got it for 70 yesterday! :)
Closing sale...pretty fortunate.

Mac

[DisposableHero]

Originally posted by mackanz
Well i got it for 70 yesterday! :)
Closing sale...pretty fortunate.

Mac

but still for 350 gph? thats what mine is now.. :(

[dmitriyaz]

hehe....
Via pumps are OK,
what can you expect from a pump?
pretty much all of that Via pumps have.
keep in mind though, they are submersible, so you can't use them in-line.
for in-line, i'd say go with ViaAqua.
they are cheap as sh|t too.

[mackanz]

Well, i had to take what i could and this pump is recommended all over so it can´t be that bad.
And you dont have to use it submersed either...and it is dead quiet.

Mac

[Player0]

Dmi, not sure who actually makes the 226w peltier. Tedist only deals with kryotherm modules, which is where I got the spec sheets for my 156w and 172w beasties. Might want to check with one of the distributers to see if they can't tell you.

Hero, those are some mighty fine cpu temps already, but you can get those water temps down to closer to 30c easy, and pick up a big improvement. Waterblocks benefit from higher flow but radiators usually like less flow. With that said, mixing tube sizes (including pumps and rads) inside a watercooling system has bad results. I use eheim 1250 pumps with 1/2" tubing. I got new adapters for the pump itself so its completely inline with 1/2" id. Rio pumps are okay, and you can actually convert them to run inline mode. Many people like Danner-Mag pumps. Nothing beats the quiet and reliability of the Eheims though, but your gonna pay premium for them. $70 ish for the 1250 at like dangerden or petwarehouse.

Dmi, heatercores and radiators are basically the same thing, and they all restrict flow to one degree or another.

[DisposableHero]

yeah i have a mag-drive danner pump right now... its leaking like crazy

[Player0]

Yeah I hear about those problems all the time. They are mostly designed for submersed usage. I hear you can use a special nylon seal (mcmaster.com should sell them) that helps improve the friction on the impeller fittings and stops the leaks. I dont own one of these drives so dont know.

I will only use eheims.

[DisposableHero]

i tried to seal the whole pump up but the water would just find its way through :mad:

i just think ehiems arent powerful enough

[Player0]

Well what kind of power do you need? For a bong, no probably not. But for a case with less than 4' lift height and 1/2" tubing, the 380GPH is more than adequate. I have resevoirs bolted right on to the front of the Eheim with 3/4" fittings for a 0-head input feed, and my 1250 is strong enough to move water through the peltier block and 4 black ice extreme radiators.

[DisposableHero]

btw.. just so you know a 175w pelt will be fine for an amd chip.. i ran a 156w for some time.. worked fine


i upped vcore
http://blitzkriegwars.netfirms.com/mbm5pic1966.jpg :)

[dmitriyaz]

Originally posted by Player0
Dmi, heatercores and radiators are basically the same thing, and they all restrict flow to one degree or another.
not exactly.
heater cores are a type of radiators, yes,
but regular rad is just a pipe with fins around it,
and in heater core, water first goes to the tank,
than from the tank, there are many tiny pipes coming out, they are usually not round, but kind ot flat. for increased surface area. the fins in the heater core are touching those little pipes. then the water goes to the 2nd tank, and from there to the barb etc.

[Player0]

Hero, I had much different results with the 156w on my CPU. This was running at 12v however. My temperatures were better without the pelt above 1500mhz. But again, at 12v. At 15v, the story may have been different.

Im actually going to hook the 226w pelt up on to the Antec 550w, and give that a go. If someday, god willing, I find out how to mod the Antec PSU, I will crank it to 13v on the 12v line to give my pelt a little more juice. I just hope the pelt at 19ish amps leaves enough current left over for the rest of my PC. I will be moving my fans over to the 24v PSU, so that should help.

Dmi, a heater core is simply a radiator that was designed for the specific purpose of heating the interior of a car. The design of these types of radiators are usually built to be 1.) compact in size and 2.) fan/duct friendly. Both reasons why they make excellent rads for watercooling. However, that fact alone doesn't mean that heater-cores are designed to be less restrictive. That usually depends on a model by model basis.

I was simply commenting on your "heater cores don't usually restict flow" which isn't exactly true. Everything restricts and reduces flowrate in a system because of friction. Radiators with long passes work better, but create more friction. Some radiators have built in resevoirs which helps balance pressure between rad tubes and slows the flow down a little.

The more restrictive a radiator is, the better job it probably does in cooling. Of course, slowing the water down in the waterblock has the opposite effect. That is why you want your flow rate to be not to fast and not to slow. Somewhere in the middle usually works the best for the waterblocks and radiators equally.