how much cooling is enough for a peltier setup?

[Squishyface]

lets say i'm running a 156W pelt on a 1.2 tbird.
The cold plate is gonna be some dt colder than the hotplate...and the hotplate is gonna be as cold as i can get the water.
So...if the water can theoretically be cooled to near room air (22C)....will that allow the pelt to cool the cpu to near 0C or less? And...is the ability to cool the water to room air just a function of how many/big your radiator is?
I read a bit about chillers...seems like that would just heat your house up

------------------
1.2Tbird @1456 with winter air cooling...yeah baby!

[SYX]

Well you won't get a 156wat peltier cuz those use 15v and PC only gives off 12, so thats around 120 wats. Yeah radiator plays a big thing. I have a 4 pass one that came with AquaStealth and it's plain garbage, it keeps water 5-8cels over ambient. Get the cube from dangerden, that should cool things well.

[Cablepro]

hey dood, guess what...i have a 156W pelt...and guess what...my ps does 15V...

[mfpmax]

According to Overclockers.com article a 172w peliter at 12v will handle a 1 gigger.

------------------
Watercooled K6-2 450 at 600 (http://mfpmax.tripod.com/oc2/)

[Kevin]

Those 172w peltiers are 24v. And standard ATX psus have 12v and 5v lines... no 15v lines so I'm not sure what that other guy was saying. You can get temps a bit under zero NOT UNDER LOAD w/ an athlon AS LONG AS you make sure that your water stays room temp.

[Rich W]

Hehe.... watch for an upcoming article from me http://www.procooling.com/ubb/smile.gif

[Cablepro]

ok, i will expound on my previously posted information...

i have an Antec 300W atx power supply in my case, but that's only to power my system...

i have a Pyramid PS36KX 15V 32A power supply to power my 156W peltier and 4 120mm fans for the 2 radiators...

i am very grateful of the way the new 1.2gig chips are manufactured...this chip is running at 1496mhz (11X136) with default voltage, heh...I am "burning in" the chip...I have not attempted higher fsb yet, although i believe it wont be a problem...I'm using the kt7a-raid with pc175 ram...but I want to let it run for a while at each step...

i'll keep ya posted

[Squishyface]

Cablepro.....11x136???

**WHINE**

hrm...i'm using an A7v board...from what i gather on this bbs, they don't overclock the fsb well....and the abit k7-raid...whether you use raid or not....just plain rocks at fsb overclocking. Is this true?
Right now i'm just runnign 12.5x100...i can get it to 105 fsb with air cooling..but no higher. I'll try knocking down the mult to 10 or so and going up on the fsb...but previously when i did this using a 1.0G chip...i hit 115 max http://www.procooling.com/ubb/frown.gif

------------------
1.2Tbird @1456 with winter air cooling...yeah baby!

[Cablepro]

nah...the kt7 sucks for fsb bumping...

the highest I could get with my kt7 was 12.5 X 104 = 1300mhz...I just couldn't get stable at any higher...I have heard of some people getting as high as 112fsb from the kt7...

the board I use now is the new kt7A-raid...it loves bumbed up fsb...but the chipset gets real hot...I am considering putting a new heatsink/fan on the north and south bridge...with this pc175 ram, I am hoping to get near 150FSB

[UnaClocker]

I use a Blorb on the chipset on my Asus A7V133.. Seems to work good, and looks great.. http://www.procooling.com/ubb/smile.gif

------------------
The UnaClocker
Overclock till it goes BOOM!
Watercooler extraordinaire!

[Alives]

yeah, external power supplies are the only way to go...i have a tripplite 10A for an 8A pelt...havent hooked it up yet though (waiting on another slotket cuz i broke a ****in clip...ridiculous)

[Freakyfrank]

Originally posted by mfpmax:
According to Overclockers.com article a 172w peliter at 12v will handle a 1 gigger.


if i'm correct... the 172W Pelt will pump 86W@12v and the 156W 115@12V




------------------
OC'ing is my middlename

[Cablepro]

As I have learned from RichW here, the power curve for peltiers is not linear, but quadratic...you cannot simply say, "ok, i'm giving the peltier half the power, and it should be able to pump half the heat"...

quadratic refers to exponential change...
to give you an example, for testing, I hooked up my 172W peltier to my 12V power supply(ATX), and it didn't even warm my hand...suggesting to me to have been pumping maybe 30-50W at maximum...

The 156W peltier, according to logic, would provide approximately 124W of pumping power, ASSUMING, that the power supply can provide the appropriate amperage(minimum 15A in this case, since initial draw is more than constant draw)...

having said that, I actually put my 156W to the test...I hooked it up to my new variable 12-15V power supply which can power up to 32A continuously, and 35A surge...at 12V, the peltier was drawing 12.5A consistantly (150W)...hrmm...interesting
at 15V, the peltier is drawing 15A consistently(225W)...wait a minute...hold the phone...how can this be???...

What this means, (and i've verified with Tedist) is that the advertised Qmax is wattage at "maximum efficiency", hence the "max" subscript...since it has been determined that peltiers run at maximum efficiency under 75-80% of full power, it stands to reason that I would be getting Qmax while only providing 12.5A, 12V...

Basically, because of diminishing returns, any further power I give this peltier above 12V 12.5A, reduces efficiency...yes, it can pump 200+ watts of heat, but it is itself generating so much more heat that would be operating with gross inefficiency...

I decided to test cold side temperatures...I held a bag of ice to one side of the peltier, and a rag to the cold side(once I held the cold side directly, and had to "peel" my hand off of it, because it froze)...I got below zero temps on the cold side at 12V, but when I pumped it up to 15V, the temperature reading began to rise...

keep in mind that my temperature readings were erroneous across the board, because the peltier was not sealed, so the probe was pulling heat from the surroundings...but since it's the same compromise at both voltage, the result is the same...higher than 12V on the 156W peltier reduces efficiency...BUT ONLY if you are providing the appropriate current...you must be aware that if you are running this off of your ATX supply that is running your computer, your 12V line probably only has 8-10A left to share for your peltier, if you are using something like the Enermax 430, which provides 12A of power...this will underpower your peltier, and make it mad at you, heh...it will try to draw the current it needs anyway, and your power supply will suffer from it...

bottom line...remember that Qmax is already advertised at maximum efficiency...I have detailed what that is for the 156W peltier from Tedist...but for the 172W and others, I am not sure, as I have not done proper testing with a 18-24V supply...but I am guessing max efficiency would be around 18V, 10A...



[This message has been edited by Cablepro (edited 02-16-2001).]

[Alives]

do you think a 10amp (11 amp surge) @ 13.8v power supply would cover an 8.5 amp @ 15.4V pelt?

[Squishyface]

Cablepro, you have a kt7a board?....don't those have a 266 fsb? is your tbird of the 133 clock variety or are you using a standard 100 one?
If you are using the 12x100 kind...can you lower the fsb on the kt7a boards to start off overclocking slowly? And i heard there was a problem with booting up...the computer initially tries to recognize your cpu at 133xclock mult regardless of the fsb you set it to? is this true? have you had any problems with this? This should probably be in a new topic :P hehe

------------------
1.2Tbird @1456 with winter air cooling...yeah baby!

[Cablepro]

I've heard of that initial startup stuff, but I've experienced no problems whatsoever...I'm using a 200mhz chip...I started out at 9X133, and began increasing from there, in order to "burn in" the chip...but...some people say it's not necessary...either way, i'm just being cautious

personally, I think spending on a 266mhz chip is a waste of money...all the 1.2 chips are made in Dresden, come unlocked, and most overclock insanely well...i'll use this chip until next year...I wont swap out until the next generation of chips are able to get me to 2G and beyond...and of course, i'm hoping new memory technology comes out...i'm still not sold on this DDR stuff

[Freakyfrank]

cablepro..

you're right about the Amps drawn at 12V.. http://www.procooling.com/ubb/smile.gif

but what if:
i don't care about the ineffeciency at Vmax (and the xtra heat generated at the hotside).. and want to get the highest Tdelta..

if u run the pelt at Vmax and Imax... and with no load the Tdelta will be the Tmax

this is for a TEC-55W@16V@6A@70°C

http://www.electronics-cooling.com/Resources/EC_Articles/SEP96/images/a4df2art.jpg
http://www.electronics-cooling.com/Resources/EC_Articles/SEP96/images/sym_068.gif T(°C)

as u can see the Max http://www.electronics-cooling.com/Resources/EC_Articles/SEP96/images/sym_068.gif T will be reached at Imax and Vmax..
and at 13.8V and 4.8A the Tdelta will be 67°C

http://www.electronics-cooling.com/Resources/EC_Articles/SEP96/images/a4df3art.jpg
http://www.electronics-cooling.com/Resources/EC_Articles/SEP96/images/sym_068.gif T(°C)

here u see that the power curve is quadratic.. the space between the 6A and 4.8A line is much smaller than between the 2.4A and 1.2A line


What this means, (and i've verified with Tedist) is that the advertised Qmax is wattage at "maximum efficiency", hence the "max" subscript...since it has been determined that peltiers run at maximum efficiency under 75-80% of full power, it stands to reason that I would be getting Qmax while only providing 12.5A, 12V...


the MaxTdelta is reached at Imax and Vmax..
normally the Qmax is taken at Vmax and Imax..(because over the Vmax the deltaT is reduced due the 'joule' effect)

joule vs peltier effect

the peltier effect draws the heat from one cold side to the hot side... the peltier effect is linear to the VA (volt x amp)

the joule effect causes the internal heating of the TEC (think about a light).. the joule effect is quadratic to the VA..

thats why the power vs Q curve is quadratic.. at a certain point the joule effect start being dominant to the peltier effect.. at that point the TEC draws Vmax @ Imax and can move Qmax and can have a MaxTdelta of Tmax with noload.


so now u understand why I dont understand ur little test :?

------------------
OC'ing is my middlename

[Cablepro]

http://www.kryotherm.spb.ru

nice program...

yes, you are technically correct...you WILL infact have a greater deltaT at 15V than at 12V...

but look at these numbers...at 25C ambient, with a heat exchanger system that is able to keep water temperature at 3C above ambient, with silicone thermal paste with thermal conductivity of 0.7W/m.K, copper coldplate 1/4" thick (48X48mm).....
yes, the peltier will produce a cold side temperature that is 3.5C lower at 15V than at 12V...however, the hotside temp is 22C higher at 15V than at 12V...that seems like a pretty rough trade-off(3.5C coldside for 22C hotside)...
of course, that is assuming that your heat exchanger setup will still cool that water to 3C above ambient with that increased hotside temperature...but if it does, you will be giving off alot of heat to the rest of the system...

[Freakyfrank]

i like my room hot and steamy http://www.procooling.com/ubb/smile.gif

------------------
OC'ing is my middlename

[Squishyface]

I like my sauna hot and steamy :P

------------------
1.2Tbird @1352 air cooled. Soon to be watercooled :P

[Cablepro]

lol...well I like to keep my video card cool, so I'd rather sacrifice the 3 degrees on the chip for the 22 degrees in the system...