PDA

View Full Version : TEC's and P4's



Roman79
01-01-2003, 10:21 PM
I'm thinking about adding a TEC to my water cooling setup. I've got a friend in an Electronic's program that can make me a 15 Amp 12v PS for pretty cheap, about $50US. I'll be upgrading to a P4 system in the next couple weeks, but I've got a couple questions that maybe the experts here can answer...

1. The P4 has a metal heat spreader which has small hole in it to let air escape during the manufacturing process (or so I've heard that's the reason for the hole). Won't there be a condensation problem on the inside of the heat spreader?

2. I haven't heard of many people using a TEC on a P4. Is question #1 part of the reason?

3. I've heard that a TEC system requires more maintenance. Right now I have an open system with needs to be topped up every day or two (goes through about 2 litres/day). Will I be looking at any more maintenance than this after the initial setup?

Thanks :)

Roman79
01-03-2003, 04:54 PM
Wow....almost 2 days and not even an "I don't know" response :confused:

dmitriyaz
01-03-2003, 05:09 PM
I don't know :p
i subscribed to the thread though, right after you posted it 2 days ago.
come on, people! :stick:

KnightElite
01-04-2003, 10:11 AM
1. No. In most cases, that hole will be filled up with thermal paste anyway.
2. May people use TECs on a P4. Swiftech sells several, such as the MCW5000-UT and MCW4000-UT that are for P4 only.
3. Yes. More heat into the water (from the TEC) = more evaporation = more water that needs to be added. If you close the system, it should be fine though. Also, swiftech includes all the necessary insulation with their package.

Sorry I didn't reply sooner, I've been slacking off the past couple days :cool:.

Roman79
01-04-2003, 12:08 PM
Thanks for the reply, that answers my questions. Do you find much advantage with TEC's over normal watercooling?

Oh, and I've got to ask.......what did you do to that poor Maze 3? :eek:

KnightElite
01-06-2003, 05:16 PM
Originally posted by Roman79
Thanks for the reply, that answers my questions. Do you find much advantage with TEC's over normal watercooling?

Oh, and I've got to ask.......what did you do to that poor Maze 3? :eek:

Well... the CPU definitely stays cooler :D. I've never actually done non-TEC watercooling, so I couldn't tell you :D.

As to the maze 3.... I had a TEC accident and it melted a little bit :D.

PimpJack
01-10-2003, 09:36 PM
if u buy the swiftech u will need dialectric grease as that pelt should give u ~0c or maybe even lower temps with regular water. The heatspreader hole will be filled with thermal paste like knight said. U can also remove the Heat S by simply sliding a rasor blade around the edges. Its only glued to the chip and there is nothing on the chip besides the core in the center.

Roman79
01-11-2003, 06:08 AM
Do you think it's worth removing the Heat Spreader? Will that improve the temps much?

PimpJack
01-11-2003, 07:37 AM
Originally posted by Roman79
Do you think it's worth removing the Heat Spreader? Will that improve the temps much?

I removed my HS after a bad mobo fried my magical 2.26 :mad: I was then curious as how the hs works and I noticed how easyli the thing can be cut off. I would imagine it would improve temps couse ur cooling the die itself without the hs being in the way and cousing another median. Its a good thing that HS but for xtreme cooling i think the p4 will benefit from direct die cooling. SO do as u please but if u do take it off make sure u post.

g.l.amour
01-22-2003, 02:45 AM
any concensus on which is the more efficient tec to use?

172 @ 24
226 @ 15

KnightElite
01-22-2003, 05:45 AM
Originally posted by g.l.amour
any concensus on which is the more efficient tec to use?

172 @ 24
226 @ 15 They'll both be most efficient at somewhat less than their max voltage, around 18 or 19V for the 172W and around 12V for the 226W. The 226W TEC is the more powerful cooler. The only real advantage to the 172W is that it is 40mmx40mm, as opposed to 50mmx50mm, if your block can't take larger TECs.

g.l.amour
01-22-2003, 06:25 AM
thk u very much for this straightforward answer.

ideal coldplate thickness for a tbred?

Tedinde
01-22-2003, 08:13 PM
I've been using tec's since my P2 rig. My last tec waterblock was the swiftech mcw462-ut 226 watt, P4/AMD block universal. Best one i ever had, mostly for the great gaskets on it. Easy, fast and no worries. It would get just as cold as my Vapo PE. Colder with a little ice in the water

Downside. It pulled a disgusting amount of electric. I was using a external variable 36 amp 5-18 volt supply. @ 13 volts it was pulling 24 amps steady. 1.6a @ 2.9ghz+ And it would cool my 2.8 to 3700mhz no problem.

I used it for 4 months straight, no problems at all. Cept maybe worrying about the pump or power supply dying. I still have it here, havnt figured if I should sell it or put in away for future use. But then i still have 2 slot one waterblocks in my junk drawer.


It used to do the best @ 13 volts, any more and CPU temps would rise.

That 15 amp PS your friend is going to make for you wont cut it. My 36 amp PS would heat up the room. you would fry the 15 amp.

g.l.amour
01-22-2003, 11:41 PM
if you'd have to make a guess. how much more electricity would the tec consume, as opposed to a vapochill?

KnightElite
01-23-2003, 08:36 AM
Originally posted by g.l.amour
if you'd have to make a guess. how much more electricity would the tec consume, as opposed to a vapochill?

Quite possibly twice as much. At 12V, a 226W Peltier cooler will actually use 230W, plus minor inefficiencies in the PSU, so probably about 250 or so. The vapochill is definitely less than that, though I don't know how much offhand, but I doubt it's more than 100W since it runs off the same power supply as the computer.

dmitriyaz
01-23-2003, 09:18 AM
for a 226w pelt, wouldn't it be (12v) x (24a) = 288w...?

KnightElite
01-23-2003, 01:16 PM
Originally posted by dmitriyaz
for a 226w pelt, wouldn't it be (12v) x (24a) = 288w...?

No. The reasoning behind this is simple:
Max Voltage = 15.2V for a 226W TEC from DangerDen
Max Current = 24A

From Ohm's law, we know that R = V/I
Therefore R = 15.2V / 24A = 0.633333Ù

So, now if we run the TEC at 12V, we know that:
I = V/R = 12V / 0.633333Ù = ~19A

So, now P = V*I = 12V * 19A = 228W

This is a very common mistake that is made all the time with TECs. Just use the formulae above to compute how much current your supply will have to be able to source to power the TEC that you want to use.

dmitriyaz
01-23-2003, 01:47 PM
i am aware of the Ohm's law,
however it doesn't always apply to semiconductors, such as TECs;
so without looking at the volt-ampere characteristic graph of that 226w TEC (which i've been desparately trying to find),
we can not know what current it will draw with a given voltage.
well then, if we can not know, where did i pull the 24a @ 12v out of?

Originally posted by Tedinde
@ 13 volts it was pulling 24 amps steady.
if it is consuming 24a at the Vmax (15.2v), and its consuming the same amount at 13v, it would be fairly safe to assume that the volt-ampere characteristic graph looks something like a straight horisontal line, at least in the voltage range we are interested in.

or am i wrong? :)

KnightElite
01-23-2003, 04:20 PM
You're wrong, because TECs are NOT semi-conductors. You can make your own TEC with two pieces of copper wire, some bismuth wire, and a 9V battery. Attach one piece of copper wire to each battery terminal, then one end of the bismuth wire to each copper wire. One end of the bismuth will get cold, the other get hot. I can't remember why offhand, but that's the Peltier effect. It was discovered in the late 19th century, I think.

dmitriyaz
01-23-2003, 04:39 PM
yes, TECs ARE semiconductors.

Thermoelectric (Peltier) modules package semiconductor pellets in a special ceramic substrate array, which is uniquely configured to pump heat
the quote is taken from this paper (http://tonio95.free.fr/download/download/peltiers.pdf).

KnightElite
01-23-2003, 08:01 PM
Interesting. I stand corrected. But it does not have to be semi conductor pellets, as shown here:
http://www.kryotherm.ru/what_is.htm

In any case, there was a program that someone had that told you the power requirements for a given TEC, and I think it confirmed my equations. I should use an ammeter and measure the current in my setup, I guess....

Roman79
01-24-2003, 01:51 PM
Wow, quite an interesting discussion going on here....

I would have guessed that a TEC was not a semiconductor as well. Since it's just two different metals sandwiched together, both of which are both good conductors...maybe it's the junction in between that makes it a semiconductor?

Tedinde
01-24-2003, 02:18 PM
The tec comsumes over 3 to 4 times the power of a vapochill, They are very inefficent, kinda like my girlfriend.

Xenogias
01-24-2003, 06:04 PM
Roman,
You may want to add another radiator to your water loop to dissipate some of that heat, especially if you decide to close ff your system. I've been TEC cooling my 2.26 since August and at first went with a Black Ice Extreme and a TT 120mm fan, but have since upgraded to another BIX and added a couple 190cfm deltas to blow them good ;] The noise is hell, but i have a custom high-wattage rheobus for when i'm not benching to quiet them down.

ZubZero
01-25-2003, 01:53 PM
As the VapoChill consumes 72W as a maximum, the TEc consumes almost 4 times as much, plus the power needed to cool the TEC itself :p