PDA

View Full Version : Modifying TEC coolers


GFORCE100
03-20-2007, 07:27 AM
Hi all,

Has anyone here thought about replcing the TEC on a Titan Amanda and similar products? Unless the TEC is overpowered with a hot CPU, it's actually quite effective i.e. QX6700 @ 3GHz @ 1.25V = 53C load (4x Prime95). It's only when you cross the TEC cooling capability threshold that temps go into the high 60's and 70's. If you keep within these limits, the heatsink isn't really hot, warm at most leading one to think a better TEC could be installed and work nicely (89W TEC for example).

The four screwns on the bottom of these units keep the TEC's in place so installing a different TEC is possible, and then feeding its wires into the molex type connector in place of the existing one's (with some tweezers lets say).

There's a lot of critism with air-cooled TEC coolers but I can't get why providing the above conditions are met. TEC's simply don't want to be overpowered on the cold side beyond a certain threshold else their cooling capability drops exponentially.

theelectic
03-20-2007, 07:49 AM
The TEC in those units is already undervolted. Hit it with 18-24V and see what happens.

GFORCE100
03-20-2007, 08:38 AM
The TEC in those units is already undervolted. Hit it with 18-24V and see what happens.\

Because it's rated for 15V and powered by 12V? Even so, a more powerful TEC would still posses greater cooling potential than the current one.

ColonelCain
03-21-2007, 01:30 AM
Only if the heatsink can cool it.

GFORCE100
03-21-2007, 06:22 AM
Only if the heatsink can cool it.

Right but the heatsink doesn't seem hot, warm at most, same for the heatpipes. The heatsink only gets hotter when I overload the TEC (i.e. overclock above 3GHz at 1.45V, 3.33GHz, QX6700).

As I said, at 3GHz, it's 54C most under load and the heatsink is moderately warm or cool/warm.

littleowl
03-21-2007, 04:45 PM
IMO The reason people(me included) look down on air cooled tecs is because look at your temp 54c! I can get that with just plain old water or maybe even an xp90 with the right air flow. but with the xp90 I may not get that much of an oc. look at the price to run that system. And the other thing that turns me away from them is my experance so far with my tec's. This is just my oppenion.

GFORCE100
03-22-2007, 07:39 AM
littleowl;2083465']IMO The reason people(me included) look down on air cooled tecs is because look at your temp 54c! I can get that with just plain old water or maybe even an xp90 with the right air flow. but with the xp90 I may not get that much of an oc. look at the price to run that system. And the other thing that turns me away from them is my experance so far with my tec's. This is just my oppenion.

I see, but on the other hand 54C isn't much when you hear other QX6700 owners running into the 70's at stock, let alone when overclocked.

serialk11r
03-22-2007, 03:26 PM
I see, but on the other hand 54C isn't much when you hear other QX6700 owners running into the 70's at stock, let alone when overclocked.

That's stock... you're comparing the stock heatsink (which isn't great) to a fat heatsink with a TEC. Because its got a huge heatsink and eats up tons of electricity, its better than stock.

GFORCE100
04-05-2007, 05:22 AM
That's stock... you're comparing the stock heatsink (which isn't great) to a fat heatsink with a TEC. Because its got a huge heatsink and eats up tons of electricity, its better than stock.

With stock I meant at 2.66GHz, not the stock cooler. Even with a 3rd party air cooler 70C is achieved when overclocking.

Cronos
04-05-2007, 05:40 AM
To be effective, TEC must have 2x cooling capacity of CPU thermal load.
For example,for 130W Kentsfield, TEC must be rated for ~260W cooling capacity. It will consume additionaly ~400W of power. So, you'll have to dissipate around 530W with air cooler. This is simply not possible, not saying it is highly unpractical and energy-inefficient.

GFORCE100
04-05-2007, 05:53 AM
To be effective, TEC must have 2x cooling capacity of CPU thermal load.
For example,for 130W Kentsfield, TEC must be rated for ~260W cooling capacity. It will consume additionaly ~400W of power. So, you'll have to dissipate around 530W with air cooler. This is simply not possible, not saying it is highly unpractical and energy-inefficient.

400W sounds too much, but anyway, you're assuming that whomever wants to use a TEC, does so to reach 0C or even lower temps. If one is happy with say up to 50C temps, then a lesser TEC coupled with a beefy heatsink, should do fine and keep temps low enough to maintain a stable overclock (below 70C).

Cronos
04-05-2007, 06:05 AM
400W sounds too much, but anyway, you're assuming that whomever wants to use a TEC, does so to reach 0C or even lower temps. If one is happy with say up to 50C temps, then a lesser TEC coupled with a beefy heatsink, should do fine and keep temps low enough to maintain a stable overclock (below 70C).

Nowhere near 0C. This will require much more powerful TEC and watercooling.

2x equation assumes 35C difference between cold and hot side of the TEC.
If you can keep a hot side around 60C, than cold side will be 25C. And CPU temperature will be much, much higher!

Holst
04-05-2007, 09:02 PM
As the other guys have allready said.

These aircooled TEC only work as the TEC isnt using much wattage and is running in as efficent a way as possible.

The hotside of the TEC MUST! be kept cool or any gains from a bigger TEC will be lost with increased hot side temps.

You may be able to further improove efficency by using a bigger 60x60mm TEC and runing it at an even lower wattage.. this method may show some modest gains.

If you really want to try tweaking this sort of heatsink then I suggest that you firstly concentrate on getting the heatsink side as cool as possible (172mm delta fan + shroud or something like this), then try ajusting the TEC voltage to optimum.
Im sure that you can get these coolers working a bit better, but probably not allot better.