PDA

View Full Version : TEC not cold anymore



Torinalth
11-22-2002, 04:48 AM
ok... I have an it7 and a 2.26... runninug on it is a maze3-1 and a 226w tec with a meanwekk PSU. I used to get temps at about 12-13c idle at 3.2gig.... thing is it is now starting to overheat.... last night was benching it for one finl time before sending it off to a client, and the damn think spiked to 66c and crashed 3dmark.

the water was in a large metal pot with ice.... not too hot of water... the block was tight and secure to the motherboard.... I was clueles... removed the block and pelt and turned it on off the motherboard.... it got cool.... NOT cold.... maybe 50c? maybe 45c the water was still going so it wasnt impropper cooling.....

any ideas?

Torinalth

wymjym
11-22-2002, 07:11 AM
initial bad contact of pelt to hs then the pelt overheated and shorted out some of the elements, now it is working at less than it's rated capacity.....
wj

Torinalth
11-22-2002, 07:34 AM
what? didnt follow that at all.

T

wymjym
11-22-2002, 07:49 AM
Peltier to waterblock interface = not good (:banana::banana::banana::banana:ed, bad contact)
Hot side of peltier got too hot and shorted out (internally) some of the elements that make up the peltier unit, this didn’t kill the peltier but rather lessened it’s ability to cool/heat…maybe it is now a 126 watt element…..
If you removed it and tried to measure the temps I assume (hope) you had some sort of hs on the hot side? Right?
wj

Torinalth
11-22-2002, 08:13 AM
yes, water block was still atached and running. it has perfect flat contact as the compound showed..... it was as tight as it could go without putting a large amount of force behind it to crack it.... cant fathom what happened.....

any ways to fix it?

Torinalth

KnightElite
11-22-2002, 09:20 AM
I can't really think of any. You've more or less got to buy a new TEC if it's not working as well as it should.

wymjym
11-22-2002, 09:37 AM
yeh, if it has been damaged then it's not one to be used. I have had them fail but they were generally unpotted and the condensation killed them.
wj

will_perdikakis
11-22-2002, 09:55 AM
Also consider the stupid connections.

The wires of the pelt may be starting to detach. Pelts have a high current draw so there may be a lot of resistance in that lead.

I haTe pelt leads, I have destroyed 2 pelts by breaking of the leads. Even after soldering them back on, it was clearly worse than before.

Basically what I am trying to say is, if the leads were tugged or bent, the electrical resistance through the lead with decrease the potential across the pelt itself. You will lose voltage because of a shoody connection. Less voltage means less peltier effect.

Torinalth
11-22-2002, 10:16 AM
sigh.... damn things rae not cheap....

megatron
11-30-2002, 02:25 PM
Originally posted by will_perdikakis
I haTe pelt leads, I have destroyed 2 pelts by breaking of the leads. Even after soldering them back on, it was clearly worse than before.

I am not going to start saying you are not soldering properly since you may have been soldering since ... whenever, but I have a suggestion. :D The proper way to solder is to first heat the leads of the tec by putting them in place together how you want and holding them there with the soldering iron. Say 5+ seconds and then quickly feed a length straight at the hot part say 5 or 10 cm length (depending on how thick the wire is) in like a second or two. Then try not to jog the two wires apart when you release then. This is to get a good ammount of flux into the join which is the conductive part of the solder. You can in fact burn away all the flux if you linger for too long. Get the solder bent into a 90 degree bend before you start to make it easier to handle.

If all this fails to give the TEC back its power then sue me. :D

Also your TEC may be sensitive to the heat of the soldering iron and there is not alot I can do about that!

wymjym
11-30-2002, 03:29 PM
TECs are soldered with a lower temp solder from the manufacturer in order to avoid temp induced failure.
I have has a couple of leads come loose myself...they had been over heated, I resoldered them with typical rosin core solder (of a very small diameter) and everything was fine.
wj

megatron
11-30-2002, 03:34 PM
I heard that you should tape down your leads so they dont move to much and become brittle. So you must have to tape them to the cold plate? To stop them overheating? (Don't you hate it when people use to many question marks?)

Torinalth
12-02-2002, 08:46 AM
I just said screw it and ripped the dan thing out.... going chilled water in a month or two.... then might revisit the peltier thing.

appreciate all the assistance.... as for my soldering methods.... you mentioned nothing about a blow torch..... tight might be my problem... (joking)... yeah I do it the same way you mentioned, though dont use flux unless I'm doing pipe.

T

megatron
12-02-2002, 10:09 AM
Originally posted by Torinalth
dont use flux unless I'm doing pipe.
T
These no need to tell us about your drug taking habits (j/k).

Torinalth
12-03-2002, 07:07 AM
needed a good laugh this morning

Vingen
12-08-2002, 01:55 PM
Can someone tell me if my TEC is running right?
my system is

2400+ XP
AT7 MAX2
cosair pc3500

swiftec MCW462-UHT with 226W TEC
Swiftech S-320-12 power supplier
Dual water pump (1 Eheim 1250 + 1 sicce idra)
Dual Black ice extreme
6L water tank with 3% water wetter.

Water temp. is 29-31 celsuis
and my CPU temp is 29-31 celsius (NO OC) this reading is from Winbond HardwareDoctor

i have even tried it with 8 celsius chilled water but the temp are not lower then 17 celsius.
and there is good contact with the waterblock and tec.
so what is my problems? TEC=dead??

wymjym
12-08-2002, 03:47 PM
are you certain that the winbond reading is correct? (selected the correct one in the software)
I believe you should have cooler readings on the cpu. I use approximately the same amount of pelt power (75+156) to chill some water and I see 27~28 degrees with 21.5 degree water. I can pull my water down quite a bit lower and the ratio remains similar. Several weeks ago I was playing around and saw cpu temps of 12~13. This was with a 1600 oc'd to 1900 mhz @ 1.92V.
Your 2400 at default speed and voltage would require less cooling so it would run even cooler.
Bottom line, if you are certain that your reading are correct I would check the pelt to block to cpu interface..if they are good then I'd look at the psu...if it is working properly then I would look at the water cooling aspect.
wj

wymjym
12-08-2002, 03:52 PM
you edited while I was posting.....
the 9 degree difference is typical in a water cooled system (water temp to cpu temp) but when you add pelts everything changes.
Maybe the info on some of my pages (http://westech.home.mindspring.com/chiller/mychiller.htm) will help out.
wj