15V or 12V PSU for 15.2V max TEC?

[freecableguy]

The Swiftech MCW5002-PT I have is currently being power with a Meanwell SP-320-12 along with 4 fans each drawing just over 1A. I am running the PSU at 13.8V because it gives be about 2C lower temps both at idle and load due to the increased power to both the TEC and the cooling fans.

I am thinking I am overworking this PSU and might not be getting a full return on my work. I am now looking at a couple of the bigger Meanwell power supplies. The 480W slim line series.

The TEC is rated at a maximum power of 15.2V and I was wondering if it would be better if I ran it at 15V or my current 13.8V. I should be able to adjust the voltage down to about 14V if I bought the 15V unit. I know my watercooling circuit can handle the load. I am pumping more than 2GPM through the loop and I have a dual 120mm Thermochill rad with 4 x 120x38mm fans (each rated at about 130CFM) in a push/pull config.

I am looking for some real advice here as I will most likely be ordering tomorrow. Should I go for the SP-480-12 or SP-480-15? Thanks.

-kris

[freecableguy]

*faint sound of crickets chirping in the distance*

Anyone have any TEC experience?

[JetForcegeminiX]

get the 15v one since you can always turn it down to 13.8 anyways

[freecableguy]

Excellent. That's what I was thinking. +/-10% adjustment with the 15VDC PSU allows 13.5 to 16.5V....meaning I can run anywhere from my nominal 13.8V to the max for the peltier of 15.2V. Any idea how long 4 12V fans will last at 15.2V??? :D

[saaya]

not as long as at 12v, but they should last some time... ive run fans at 25v for a few days without a problem, one of them died inmediatly though... so it depends...

its best to run 15v tecs at 12v i heard, its their sweet spot.

[freecableguy]

When people talk about the "sweet spot" they are talking about maximum efficiency. Maximum cooling ALWAYS occures at Vmax/Imax when Umax is produced. Now, whether or not someone's cooling system can keep up with all of those watts is a different story. At Umax the unit is putting out 360W + 130W for the CPU + ~45W from my pump. That's over 500W! Now you can see why I need A LOT of airflow across my rad...

[saaya]

no, afaik the highest efficiency of a tec is NOT at qmax, its slightly lower, around 20% i think. at more the tec will produce a lot more heat but will pump only slightly more watts.

thats what several people told me, i never tested this myself! i dont know if maybe their cooling of the hot side was just not good enough to cope with the few extra watts the hot side dissipates when running at full qmax and the pumping power is actually linear or even increases steadily. but since ive heard this from several people with a lot of expirience in tec cooling and powerfull rads to cool the tecs hot sides, i think a tec really works most efficiently at qmax minus around 20%.

so if you have a REALLY huge rad and cooling the hot side is not an issue, then yes, qmax would result in the be best temps on the cold side.

what tecs are you running? a 227W tec always puts out around 400W of heat on its hot side, no matter how many watts the cpu has ypu cool it with, the only difference you have when your cpu produces less heat is that more cooling power will be used to reach a lower temp.

a 227W tec sucks 227W of energy and dissipates 99% of it as heat, so 227W. plus it is able to pump around 175W of heat from its cold side to its hot side. so you will have to remove around 400W of heat on the hot side.

if you cpu puts out 80W of heat the tec will remove the 80W of heat wich in theory means the cpu now has the same temperature as the ambient, and the other 95W of pumping power will be used to pump even more heat from the cold side to the hot side wich means its used to lower the temp of the cold side to below ambient.

if your cpu produces 130W of heat you will only have 45W of pumping power left wich means the cold side will not be that much below ambient.

thats all theory ignoring the temperature of the hot side and assuming its the same as ambient temp of course.

with 130W of heatload and a 227W tec to cool it you will only get a delta T of 30C. so if you manage to get the hot side of the tec to 30C the cold side will be 0C. as the waterblock also has a thermal resistance i think you will need 20-25C cold water to reach 0C on the cold side.

again , thats theory with forumlas etc... what are your real temps?

[freecableguy]

Pretty damn close to that. I agree with everything you have typed. Yes, maximum EFFICIENCY (as I posted above) does occur at about 80% of Vmax....I want maximum cooling ability and therefore I want to run Vmax. The Vmax is actually the voltage at which any more pumping power provided by the TEC is overshadowed by increased power requirements which actually cause the TEC to self-heat both sides. Hence, Vmax/Imax. But, assuming I CAN keep the TEC cool on the hot side I will see maximum performance at Vmax....highest load on my watercooling setup....lowest efficiency....highest power consumption...and lowest temperatures. It is a diminishing returns argument at voltages > 0.8Vmax....I don't really care much about efficiency though...I want maximum cold!


Oh, yeah, temps:
As reported by Abit IC7-MAX3 BIOS:
Idle - 32C under ambient (or -2C)
Full load - ambient (or 30C)
Ambient - 30C (hot!)

-kris

[saaya]

Oh, yeah, temps:
As reported by Abit IC7-MAX3 BIOS:
Idle - 32C under ambient (or -2C)
Full load - ambient (or 30C)
Ambient - 30C (hot!)

-kris [/B][/QUOTE]

whats your water temp and what block are you using?

[freecableguy]

Don't know water temp....i will buy a standard thermometer today and put it in my t-line connection to find out. Block is a Swiftech MCW5002-PT (read sig....).