ebay 245/320W Peltier Cooling Rating

[Premmer]

Consider the following.

Qcmax = 226W @ Th = 27C
Vmax = 15.2V @ Th = 27C
Imax = 24A @ Th = 27C
Pcmax = 364.8W @ Th = 27C

Qcmax = 320W Th = 50C
Qcmax = 243.5W @ Th = 27C
Vmax = 15.4V @ Th = 27C
Imax = 26A @ Th = 27C
Pcmax = 400.4W @ Th = 27C

If one believes the 226W and 320W related parameters above are factual and you compare the differences between the 226W and alleged 320W device, then for a 35.6W increase in input power (400.4 - 364.8) you get a 94W increase in cooling power (320 - 226) from the 320W device.

How about this, for a 35.6W increase in input power you get an 18W increase in cooling power from a 244W device.

Which scenario is believable?

[ls7corvete]

Consider the following.

Qmax = 226W
Vmax = 15.2V
Imax = 24A
Pmax = 364.8W

Qmax = 320W Th = 50C
Qmax = 243.5W Th = 300K = 26.8C
Vmax = 15.4V
Imax = 26A
Pmax = 400.4

If one believes the 226W and 320W related parameters above are factual and you compare the differences between the 226W and alleged 320W device, then for a 35.6W increase in input power (400.4 - 364.8) you get a 94W increase in cooling power (320 - 226) from the 320W device.

How about this, for a 35.6W increase in input power you get an 18W increase in cooling power from a 244W device.

Which scenario is believable?


I beleive both, problem is that we have been assuming dT to be a constant but it varys with Th, I think we would need graphs of dT vs Th to be able to fairly compare both TECs.

I bet the graph is non-linear as well. :brick:

[Premmer]

I beleive both, problem is that we have been assuming dT to be a constant but it varys with Th, I think we would need graphs of dT vs Th to be able to fairly compare both TECs.

I bet the graph is non-linear as well. :brick:

Regardless of the dT vs. Th response characteristics, common sense tells me that a 35.6W increase in input power will not get you a 94W increase in cooling power unless this peltier has an efficiency rating much greater than 100%.

[ls7corvete]

Regardless of the dT vs. Th response characteristics, common sense tells me that a 35.6W increase in input power will not get you a 94W increase in cooling power unless this peltier has an efficiency rating much greater than 100%.

Indeed, but that doesnt really help us with calculations much. We need those graphs or some generalized graph we can use to apply to all TECs, that is of coarse if they are all basicly the same.

Its not is this one lieing or is this one bigger, its how much bigger, how much of a lie. Those Qs can tell us something.

[Revivalist]

Indeed, but that doesnt really help us with calculations much. We need those graphs or some generalized graph we can use to apply to all TECs, that is of coarse if they are all basicly the same.

Its not is this one lieing or is this one bigger, its how much bigger, how much of a lie. Those Qs can tell us something.I think to eliminate as many variables as possible, we would need to compare only the Q versus operating temp of one pelt at a time. It seems that the different manufacturers have different "sweet spots" for their pelts and thus there could be a variance in the properties of the different pelts. . .

By the way, along those lines, would it be right to compare the 226W and 245W/320W pelts and say that a "35.6W increase in input power will not get you a 94W increase in cooling power unless this peltier has an efficiency rating much greater than 100%." I think that statement would definately be true if we were talking about increasing the input power of the same pelt at a given operating temperature. But that statement is comparing two different pelts from two different manufacturers and possibly at two different operating temperatures. I'm not saying it's certainly wrong. I'm just asking if these factors are being taken into consideration.

[ls7corvete]

I think to eliminate as many variables as possible, we would need to compare only the Q versus operating temp of one pelt at a time. It seems that the different manufacturers have different "sweet spots" for their pelts and thus there could be a variance in the properties of the different pelts. . .

By the way, along those lines, would it be right to compare the 226W and 245W/320W pelts and say that a "35.6W increase in input power will not get you a 94W increase in cooling power unless this peltier has an efficiency rating much greater than 100%." I think that statement would definately be true if we were talking about increading the input power of the same pelt at a given operating temperature. But that statement is comparing two different pelts from two different manufacturers and possibly at two different operating temperatures. I'm not saying it's certainly wrong. I'm just asking if these factors are being taken into consideration.


what temperature though, all systems are gonna vary and Th is gonna vary with the size of the pelt as well, more heat more Th. the 50c ratting might be right for a big pet, the 27c rating might be closer for a smaller pelt. Most people are gonna end up in the middle and not get a correct figure at all.

Graphs, we need graphs.

[Premmer]

I emailed Thermal Enterprises and posed the following question.
"Your ebay advertisement refers to the device as 245-320w. Your specs clearly state that the maximum Qc of the device is 243.5w . Why do you use the 320w reference in your advertisement? It's misleading to the public".

Their response was as follows.
"We simply follow the same common practice as most other TEC manufactures, we establish the watt rating of our devices under two different operating conditions. The lower wattage rating is established with hot side temperature at 27 deg C. The higher wattage rating is established with hot side temperature at 50 deg C".

After researching the peltier rating subject, I have to agree with Thermal Enterprises response. It is up to the end-user to determine which set of peltier parameters would be most applicable for their operating conditions.

In summary, when peltier maximum cooling power wattage ratings (Qcmax) are expressed by two values as is the case with Thermal Enterprises devices, the lower rating equates to a Th = 27C ambient condition and the higher rating to a Th = 50C. In any well planned CPU peltier/water-cooled system, the Qcmax that would be most representative of their system's cooling performance would be the vaule at Th = 27C.

[Bloody_Sorcerer]

actually, the value most representative would be a point somewhere at least halfway between the 2 (if not closer to the 50C value), assuming you have an ambient of room temp (~25C) and a waterblock with a decent enough c/w and adequate radiatorage to keep water temps low. So basically, the number most applicable is somewhere in between the 2 provided values; if things were nice, it'd be the average, but i doubt it's a linear function.

[Premmer]

actually, the value most representative would be a point somewhere at least halfway between the 2 (if not closer to the 50C value), assuming you have an ambient of room temp (~25C) and a waterblock with a decent enough c/w and adequate radiatorage to keep water temps low. So basically, the number most applicable is somewhere in between the 2 provided values; if things were nice, it'd be the average, but i doubt it's a linear function.
If you operate a water-cooled TEC with a coolant temp much greater than 10C above room ambient under maximum load conditions, you probably have a poor cooling system.

[ls7corvete]

If you operate a water-cooled TEC with a coolant temp much greater than 10C above room ambient under maximum load conditions, you probably have a poor cooling system.

Dont forget the water block, say your putting 400w through a water block, you would need a C/W of .025 to keep the hot side at 10c above ambient. I would say that the 50c measurement is the best bet for this TEC at full power.

[Bloody_Sorcerer]

the 50C number is definitely the one we want; at a c/w of around .05 (blind shot in the dark, but I'd call it a fair estimate for a waterblock) and a heat load of 400 watts, thats a dT of 20C, plus 2-3C for water temps, plus ambient of 25C, you're at around 48-50C as it is. the 50C number is definitely the one of interest; we can safely call these 320watt TECs.
I wonder what the wattage of a 226watt TEC at Th=50 is... *goes lookin for info*

[Mr. Tinker]

Thanks, guys, for answering some questions I've had about that little guy. I do own the 320W from Thermal Enterprises.

[ls7corvete]

the 50C number is definitely the one we want; at a c/w of around .05 (blind shot in the dark, but I'd call it a fair estimate for a waterblock) and a heat load of 400 watts, thats a dT of 20C, plus 2-3C for water temps, plus ambient of 25C, you're at around 48-50C as it is. the 50C number is definitely the one of interest; we can safely call these 320watt TECs.
I wonder what the wattage of a 226watt TEC at Th=50 is... *goes lookin for info*

.05 is low, I have been woundering how the large size of a TEC affects dT of water blocks. Anyone who has been to procooling is familiar with the effects of a change in die size on water block testing. I wounder if someone where with good modeling skills can tell us how going from a 10x10mm heat source to a 50x50mm heat source affects dT of a block

[Bloody_Sorcerer]

i agree, .05 feels very low, but after lookin at swiftech's graphs of the storm, apogee, and MCW6000 (especially that last one, as it can/is used on TECs with modification), but those are all numbers from an unknown die size *cough TTV cough*

[Premmer]

Dont forget the water block, say your putting 400w through a water block, you would need a C/W of .025 to keep the hot side at 10c above ambient. I would say that the 50c measurement is the best bet for this TEC at full power.
Maze4-1 CPU Block, under-powered, 244W 50mm Sq.(Qcmax) at Th=27C
If:
Th = 33C - measured coolant temp
Ta = 25C - measured room ambient
V = 12.3V - measured
I = 22A - estimated from performance specs
Q = 150W - estimated, includes pump heat contribution

Then: RQ = (Th - Ta)/ (V x I +Q) = 0.019C/W

Is RQ a believable approximation?

OR

Dominator Pro CPU Block, under-powered 377W 62mm Sq.(Qcmax) at Th=27C
If:
Th = 30.4C - measured coolant temp
Ta = 20.3C - measured case ambient
V = 19.9V - measured
I = 24.6A - measured
Q = 143W - estimated, includes pumps heat contribution

Then: RQ = (Th - Ta)/ (V x I +Q) = 0.016C/W

Is RQ a believable approximation?

dT = 45C; 30.4C - (-14.6C), under substained maximum load conditions

Operating a TEC at the Th = 50C rating does yield a higher dT and it requires even more input power to accomplish it.
Case in point, a DP 377/437W:
Th = 27C
Qmax = 376.8W
I(Qmax) = 30.4A
Umax = 24.8V
dT = 66.0
Pcmax = 754W

Th = 50C
Qmax = 437.3W
I(Qmax) = 32.8A
Umax = 26.7V
dT = 74.5
Pcmax = 876W

[Bloody_Sorcerer]

.016 c/w? thats utterly insanely low... it may have to do with die size, but I'm disinclined to believe that, considering that many radiators have c/ws of around .01 and that .01 at the size of a die (though TECs are much much larger) is the C/W for TIMs... however, if these are measured numbers... *shrug*

[ls7corvete]

Maze4-1 CPU Block, under-powered, 244W 50mm Sq.(Qcmax) at Th=27C
If:
Th = 33C - measured coolant temp
Ta = 25C - measured room ambient
V = 12.3V - measured
I = 22A - estimated from performance specs
Q = 150W - estimated, includes pump heat contribution

Then: RQ = (Th - Ta)/ (V x I +Q) = 0.019C/W

Is RQ a believable approximation?

OR

Dominator Pro CPU Block, under-powered 377W 62mm Sq.(Qcmax) at Th=27C
If:
Th = 30.4C - measured coolant temp
Ta = 20.3C - measured case ambient
V = 19.9V - measured
I = 24.6A - measured
Q = 143W - estimated, includes pumps heat contribution

Then: RQ = (Th - Ta)/ (V x I +Q) = 0.016C/W

Is RQ a believable approximation?

dT = 45C; 30.4C - (-14.6C), under substained maximum load conditions

Operating a TEC at the Th = 50C rating does yield a higher dT and it requires even more input power to accomplish it.
Case in point, a DP 377/437W:
Th = 27C
Qmax = 376.8W
I(Qmax) = 30.4A
Umax = 24.8V
dT = 66.0
Pcmax = 754W

Th = 50C
Qmax = 437.3W
I(Qmax) = 32.8A
Umax = 26.7V
dT = 74.5
Pcmax = 876W


Why are you using coolant temps as Th?

Th will always be hotter than the coolant temp.... by the amount C/W(block)*heat across block.

The calculations you show give the C/W of the radiator system.

[Bloody_Sorcerer]

that explains why they look like radiator c/ws, then. I'm gonna bet a 120x3 sized rad, no?
though that does do an excellent job of showing that TEC systems need all the radiator oomph they can get.

[Premmer]

Why are you using coolant temps as Th?

Th will always be hotter than the coolant temp.... by the amount C/W(block)*heat across block.

The calculations you show give the C/W of the radiator system.

You may be right. How much hotter would you estimate Th is relative to coolant temperature? What method would you use to measure Th?

[ls7corvete]

You may be right. How much hotter would you estimate Th is relative to coolant temperature? What method would you use to measure Th?

C/W of the block times the heat put through it=Th.

The only problems is common C/W measurements are taken with something like a 10x10mm heat source, we have a 62mm heat source.

Measuring Th....well I am afraid it is not that simple to setup a test platform and get good results, try the water block testing forum for some info. Our best bet is to have someone test for us or run a computer model to show the performance difference when using a large heat source.

We should start asking around to see who has the parts to setup such a test/model.

[Holst]

I think that these ebay numbers may just be incorrect.

[Premmer]

C/W of the block times the heat put through it=Th.

The only problems is common C/W measurements are taken with something like a 10x10mm heat source, we have a 62mm heat source.

Measuring Th....well I am afraid it is not that simple to setup a test platform and get good results, try the water block testing forum for some info. Our best bet is to have someone test for us or run a computer model to show the performance difference when using a large heat source.

We should start asking around to see who has the parts to setup such a test/model.

I already have a temperature metal probe embedded in the cold-plate. That addresses the Tc value. If I install another probe in the hot-plate, would that produce a creditible Th value that could be used to not only establish dT but the block C/W?

[ls7corvete]

I already have a temperature metal probe embedded in the cold-plate. That addresses the Tc value. If I install another probe in the hot-plate, would that produce a creditible Th value that could be used to not only establish dT but the block C/W?

Crude measurement but IMHO, its worth trying.

[Premmer]

Crude measurement but IMHO, its worth trying.
How would you refine it, details please?

[Holst]

That would be ideal. If we can find the temperature of the hot side of the TEC this will help us work out how to improove things.

How would you refine it, details please?

The value we are most interested in is the hot side temperature.
Ideally the probe should be on the hot side of the TEC... but this is not very practical (the waterblock is in the way)
Placing a probe on the bottom of the waterblock as close to the TEC as you can will give a reasonable reading ... just be sure to insulate it from the cold side or the reading may be lower than it should be.