TEC problem

[jens21]

Hi there,

I have some difficulties with my peltier setup. I hope you guys can help me out!

The peltier should have a deltaT of about 25 degrees. My setup only deltaT 14 degrees. I've added the specs and some pictures, I'll hope its enough.

Specs
Alphacool CPU-block
CPU-block: Alphacool
Radiator: Alphacool NexXxoS Xtreme III Radiator - Rev. 2
Reservoir: Alphacool Cape Coolplex Pro 10 external
Pump: Innovatek HPPS PLUS 12v Pump
Powersupply: Meanwell s-320-12 (12Volt, 26Amps)
Fans: 3x Yateloon SM 120mm
Thermostat: ETC-200B
Coldplate & hotplate: thick enough :P

Ambient temp: 23 degrees celcius
Water temp: 23 degrees
Hotside: 32,2 degrees
Coldside: 18,3 degrees


I have wired the pelts in serie, so each pelts gets 6volt. Following some graphs the COP should be very very nice, but it isnt.

If you need more information, please let me no!

(Sorry for my bad englisch, I'm from the Netherlands)

http://img843.imageshack.us/i/30072010027.jpg/
http://img294.imageshack.us/i/30072010028.jpg/
http://img299.imageshack.us/i/30072010029.jpg/
http://img842.imageshack.us/i/30072010030.jpg/
http://img295.imageshack.us/i/30072010031.jpg/
http://img834.imageshack.us/i/30072010032o.jpg/
http://img830.imageshack.us/i/30072010033.jpg/

Thanks!!!!

[TJ TRICHEESE]

what is the hot side temperature the deltaT has been shown to be 25c

[jens21]

what is the hot side temperature the deltaT has been shown to be 25c

Ambient temp: 23 degrees celcius
Water temp: 23 degrees
Hotside: 32,2 degrees
Coldside: 18,3 degrees

32,2 minus 18,3 = 14 degrees (DeltaT)

[zipdogso]

Why should your peltier have a Dt of 25ºC ???
Dt is one of the hardest things to predict/calculate. Generally one can only get a ballpark figure without actually running a TEC.

If you used that ridiculous formula for Dt that goes round and round your Dt will be calculated far too high. The formula does not work despite what most people will try and tell you. IT DOES NOT WORK.

You also havent specified what TECs you are using.

The main problems with you expecting a Dt of 25ºC are :-

1.) Graphs for TECs COP etc are at a specific hotside temp this is usually either 25ºC, 27ºC and occasionally 50ºC.
So basically any figures you get will not match any charts as your hotside is 32.2ºC

2.) Dt is affected by input power, load on the coldside and the cooling to the hotside.

3.) load on the coldside and cooling on the hotside are affected by amongst other things quality of TIM and amount of pressure applied to the TEC. Also a change in ambient temp can affect the hotside temp.

So you see, in running, there are quite a few things that can vary from a "calculated" value. and your Dt will vary with it during a run.

Without knowing more specifics your Dt is one that could be expected.

NOTE your COP will not be fantastic...I trust you are using 127 couple TECs undervolted to 6v any COP you can expect will probably be somewhere between 1 and 1.7. You will probably need to undervolt a little more to get anything "Great"

[jens21]

Why should your peltier have a Dt of 25ºC ???
Dt is one of the hardest things to predict/calculate. Generally one can only get a ballpark figure without actually running a TEC.

If you used that ridiculous formula for Dt that goes round and round your Dt will be calculated far too high. The formula does not work despite what most people will try and tell you. IT DOES NOT WORK.

You also havent specified what TECs you are using.

The main problems with you expecting a Dt of 25ºC are :-

1.) Graphs for TECs COP etc are at a specific hotside temp this is usually either 25ºC, 27ºC and occasionally 50ºC.
So basically any figures you get will not match any charts as your hotside is 32.2ºC

2.) Dt is affected by input power, load on the coldside and the cooling to the hotside.

3.) load on the coldside and cooling on the hotside are affected by amongst other things quality of TIM and amount of pressure applied to the TEC. Also a change in ambient temp can affect the hotside temp.

So you see, in running, there are quite a few things that can vary from a "calculated" value. and your Dt will vary with it during a run.

Without knowing more specifics your Dt is one that could be expected.

NOTE your COP will not be fantastic...I trust you are using 127 couple TECs undervolted to 6v any COP you can expect will probably be somewhere between 1 and 1.7. You will probably need to undervolt a little more to get anything "Great"

Thanks for your reply! A brief reply this time, later on I will add some more information. I'm using two of these (http://www.auspicious-e.com/index.ph...roducts_id=286 ) TECS. My meanwell powersupply can produce between 10 and 14 volts. So two tecs in series, get 5 volt each.

[zipdogso]

I remember your name but not the project you were working....

Refresh me or is it a different one ?

I should of pointed out that reducing the voltage slightly to get better COP will reduce the Dt slightly too...it's initial value is determined by input power (and it is NOT linear.) then it is affected further by the other bits but since high COP goes hand in hand with low Dt (you wont get COP much over 1 with a Dt over 20.) you will need to adjust the voltage variation of your PSU to determine the best setting you can get for your current setup.
I would guess from the Dt you have that the coldside loading (watts.) you have is close to the Q of the TECs at 6v which is about 75w each at 27ºC hotside, your 5ºC higher hotside temp wont make a massive difference probably only in the order of another 10w. That figure being your effective Qmax at 6v so in real world you will only achieve a lower figure perhaps only getting about 65w-75w each one...i'm not sure I cant tell, I have no idea what your thermal transfer inefficiencies are. Neither will you unless you know for sure the coldside loading, and of course there is ALWAYS unknown thermal transfer inefficiencies.

If after adjusting your PSU you still cant get cold enough you only recourse is to further cool the hotside somehow. Or run higher Qcmax TECs which will, considering your on 12726 size, mean TECs with more couples and the higher voltage that goes with that.

Efficient running of a well undervolted TEC needs A1 hotside cooling to work, often rad cooling is insufficient unless you have very cold ambients. It is precisely the fact that you havent much Dt that the hotside cooling is paramount.

[jens21]

I remember your name but not the project you were working....

Refresh me or is it a different one ?

Yes, you know me from another topic. I think a year ago I tried to use peltiers to cool a basin with water. My main target was to cool liquid drinks like beer or wine or some. During the project I had several things, like a studie, a job and some other things, which gave me no time to work on my project.

I should of pointed out that reducing the voltage slightly to get better COP will reduce the Dt slightly too...it's initial value is determined by input power (and it is NOT linear.) then it is affected further by the other bits but since high COP goes hand in hand with low Dt (you wont get COP much over 1 with a Dt over 20.) you will need to adjust the voltage variation of your PSU to determine the best setting you can get for your current setup.
I would guess from the Dt you have that the coldside loading (watts.) you have is close to the Q of the TECs at 6v which is about 75w each at 27ºC hotside, your 5ºC higher hotside temp wont make a massive difference probably only in the order of another 10w. That figure being your effective Qmax at 6v so in real world you will only achieve a lower figure perhaps only getting about 65w-75w each one...i'm not sure I cant tell, I have no idea what your thermal transfer inefficiencies are. Neither will you unless you know for sure the coldside loading, and of course there is ALWAYS unknown thermal transfer inefficiencies.

I know it’s not linear, I’ve used graphs to point out the most efficient voltage. At 5-6 volt the COP is huge and the transferred heat is also at a acceptable level. A high COP also means, less heat to be cooled by the radiator. *Looking at the graphs* At 6Volt the TEC draw 10A, at 5Volt only 7A. The graphs where send by someone is called ‘Simon’, I believe that’s you!  At this moment there is no coldside loading, so it ‘should’ drop below 0C in no time. It doesn’t get lower than 18,3 degrees C. Ambient and watertemp are both 23 degrees. My main goal is 5 degrees, so I need a DeltaT of 18 degrees. I believe the radiator can cool enough water to keep the hotside cool enough. Because: after an hour the watertemp (of the coolingcircuit) is one degree higher than ambient.

If after adjusting your PSU you still cant get cold enough you only recourse is to further cool the hotside somehow. Or run higher Qcmax TECs which will, considering your on 12726 size, mean TECs with more couples and the higher voltage that goes with that.Efficient running of a well undervolted TEC needs A1 hotside cooling to work, often rad cooling is insufficient unless you have very cold ambients. It is precisely the fact that you havent much Dt that the hotside cooling is paramount.

Hotside cooling should be sufficient because the watertemp of the coolingcircuit doesn’t rise a lot, just one degree. I ‘could’ add another CPU coolblock, so the heat doesn’t have to transfer a lot of ‘distance’. The hot- and coldplate are 12centimeter long, and the CPU coolblock just 5 centimer. Adding another could do the trick.

[zipdogso]

Your missing the point about the cooling....

Using a rad the cooling you use to cool the hotside will never go below your ambient. Your rad is pegging the coolant at the just above the ambient. This is normal.
To have your hotside roughly 10ºC above ambient is pretty average in a rad cooled system. Using undervolted TECs you have a low Dt since the Dt is taken from your hotside temp to arrive at your coldside temp using a rad to cool your hotside will never get you anywhere when using undervolting.

Your rad is holding back the TEC. that's why you only get to 18ºC with "No-load" (technically incorrect the only "No load" with a TEC is when it is not powered.) All the while your rad pegs the coolant temp at just above ambient your getting nowhere and never will.

With undervolting hotside cooling is paramount....rads wont do...not if you want low temps.

I will say I wondered if your Dt was on the low side I think your heat transfer ineficiencies are high so your Q is low (at your operating point the max Q in perfect conditions is only about 80w each anyway.) and your load is higher than you think (there is no "no-load") so your load is close to your Q - the effect lowers your Dt, If your load is significantly lower than the Q the Dt will rise but never to a figure more than determined by the operating point.
Perhaps you need to reseat the TECs - (badly seated TECs become large heat transfer inefficiencies) use a good TIM and the correct pressure (150-300 p.s.i.) apply the pressure slowly, tightening each bolt incrementally in rotation till the correct pressure is achieved on all the bolts. You need the pressure distributed as even as possible over the entire face of the TEC.
Note depending on how many bolts you have and their size you may only need 8-9 p.s.i. on each bolt. Ensure you insulate the bolt from the hotside using fibre washers or something so the heat doesn't go down the bolt to the coldside.

The only other thing you can do is use a more powerful TEC or forget the COP (it should only be just under 1) and just increase the voltage till you get your cooling level.

[Ultrasonic2]

i haven't sat down and done though the maths. in evolved yet but

i have a few comments on this though



1 the random copper you buy is NEVER perfectly and i mean perfectly flat . which impacts on performance greatly

2 your apply a load to the cold side so what ever the theoretical max delta is you wont be achieving it

3 your relying on alot of conduction though the hot and cold plates then to the water block then to the water so efficiency will be lower and delta low than in idea situations

[aussie-revhead]

What is that in your pic US2? Is the hot side there water cooled then the sink is submerged to make a chiller? Alot of clamping force there too.

:-)