[lichking]

OK boys and girls,

Lets talk tecs realistically,

Tecs are a funny thing, and I hear this back and forth about tecs in theory and in practice. here are some facts that I know about tecs,

You can't cool a tec at its qmax, and this is why alot of people push the 50% to 60% range of a tec.. after that, there is just too much internal thermal resistance between the semiconductor, ceramic, hotplate, and air-water.

So that said, if we cool a tec at 60% then we really can't push the 69º delta, we can really hope for 60% of that delta being about 40º dt.

so realistically we can only hope for 40º off the cpu @ 60% tec utilization.

so in practice, shoot for about 30-40º of your cpu in normal water cooling, and then fill in the blanks with your tecs.

This nonsense of cooling tecs at -20 is ridiculous. the best you can acheive with tecs it about 30-40ºs off your normal water cooling setup.. so if your pushing 60º, then shoot for about 30º with tec cooling.. anything more than that is a pipe dream..

As for ultra's calculator, its nonsense, because you can't cool a tec at its qmax and the mere fact that there is too much thermal resistance in the ceramic, copper plate on each side.

as for cooling anything more than about 30 - 40º off the cpu from normal watercooling.. use phasechange, or don't do it at all.

anyone that says they can push subzero with a tec at full load is full of it.. shoot for about 30-40º at load.. then figure your tecs based on that..

regards
the chilking.. I mean lichking... lol

 

[lichking]

bump.. I want this on top!

 

[flak-spammer]

Wow, before you make a post like this you should know what you are talking about. At Q 0 or 0 moved watts you're at Dtmax when you reach Qmax you're at Dt 0. That's the relationship it doesn't matter if you can or can't cool at Qmax because at qmax it means you're moving the whole load to a delta of 0 it doesn't mean you aren't moving the load at all. It just means that in the theoretical world the hotside and cold side should be the same temperature. Perhaps before making a post that attacks everyone you should know what you are talking about.

 

[lichking]

Quote:

Originally Posted by flak-spammer Wow, before you make a post like this you should know what you are talking about. At Q 0 or 0 moved watts you're at Dtmax when you reach Qmax you're at Dt 0. That's the relationship it doesn't matter if you can or can't cool at Qmax because at qmax it means you're moving the whole load to a delta of 0 it doesn't mean you aren't moving the load at all. It just means that in the theoretical world the hotside and cold side should be the same temperature. Perhaps before making a post that attacks everyone you should know what you are talking about.

omg really

Ok.. so listen up!!!

I wasn't talking about qmax!!

I was saying that if your going to talk about running a tec an 50-60% imax then your dt only going to be 60% dt max!

so

that means if your running a tec at 60% of imax then your dtmax also be only be 60% and that equates to 69*.60 = 41.4dt!!!

you can't have it both ways!! you cant have 60% imax and 100% dt.. sorry!!

If your run a tec at 60% imax you only get 41ºdt.. sorry.. thats the truth!!!

 

[lichking]

Quote:

Originally Posted by lichking omg really Ok.. so listen up!!! I wasn't talking about qmax!! I was saying that if your going to talk about running a tec an 50-60% imax then your dt only going to be 60% dt max! so that means if your running a tec at 60% of imax then your dtmax also be only be 60% and that equates to 69*.60 = 41.4dt!!! you can't have it both ways!! you cant have 60% imax and 100% dt.. sorry!! If your run a tec at 60% imax you only get 41ºdt.. sorry.. thats the truth!!!

oh btw if you don't know most tecs run at 69ºdt!

thats the max difference between hot side and cold side at 0 q

 

[flak-spammer]

Uhh yeah most of us know all of this already. That doesn't mean you can't get negatives. It just means that Qmax, Dtmax and Imax are all related to Vmax. So...really the higher voltage you give the TEC the better things are from a Dt perspective. Even with 40 Dt at 20C Thot that means that negatives are still plausible you just have to have a TEC with a higher Qmax than the load you plan on moving. These are all accounted for in Ultra's calculator.

 

[lichking]

Quote:

Originally Posted by flak-spammer Uhh yeah most of us know all of this already. That doesn't mean you can't get negatives. It just means that Qmax, Dtmax and Imax are all related to Vmax. So...really the higher voltage you give the TEC the better things are from a Dt perspective. Even with 40 Dt at 20C Thot that means that negatives are still plausible you just have to have a TEC with a higher Qmax than the load you plan on moving. These are all accounted for in Ultra's calculator.

the point is that you can't run a tec at 60% load and expect a 60ºdt.

if your running a freezer that has no load sure!!.. but if your pushing a load on the tecs.. don't expect more than 30-40ºdt because you can't push a load on a tech @ full imax with the thermal resistance of the ceramic and substrates associated with tecs in a normal dyi cooling system.

This is the reason anyone that has run tecs (ie elloquin) states that you can't run a tec feasably over about 60% imax.

that said you can't expect a tec to run the full dtº at 60% imax!!..

this is why ultrasonic's calculator doesn't work with a load.. sorry!!.. if I wanted to cool a beer.. it would work!

 

[flak-spammer]

Why do you keep saying 60% Imax. On TECs you never use a current regulated power supply. The voltage you run it at determines the Qmax, Dt, and Imax. most manufacturers suggest 60-80% Vmax. I still don't get how you think ultrasonics calculator doesn't work. It does actually calculate the differences in Dt, Qmax, and Imax at the select percentage of Vmax.

 

[lichking]

Quote:

Originally Posted by flak-spammer Why do you keep saying 60% Imax. On TECs you never use a current regulated power supply. The voltage you run it at determines the Qmax, Dt, and Imax. most manufacturers suggest 60-80% Vmax. I still don't get how you think ultrasonics calculator doesn't work. It does actually calculate the differences in Dt, Qmax, and Imax at the select percentage of Vmax.

ok.. obviously you have never run tecs before.. this is the exact problem this forum has had for years..

You cannot... I repeat you cannot run a tec at full imax because of the thermal resistance in the ceramics and substrates.. the thermal resistance is just like a cpu, you end up with water or air temps at 10º higher and the actual tec substrate ends up at 50º+ at full imax.. pushing any gain into a - gain because of thermal resistance.

so in regards your tec @ 50c + will acheive 50dt.. you have lost all cooling and now your heating up your cold side instead of cooling it.. but yest you have your 50 dt.. but end result is a negative on the load

again I will say that your better off cooling a 60% imax load then a 100% load that you can't cool in practice..

I can show this in EVERY example on the forums TODATE!!!!

again muffies calculator is bull

regards
lichking

 

[xtremetechuk]

I will try to make this post as simple as possible.

TEC manufactures rate TEC's at their maximum operating performance according to a specific Th temperature, and so In theory any TEC can be run at Imax, they are just far more efficient when undervolted. This is why many of us recommend running TEC's at 40-60% of Umax. You keep insisting that because of thermal resistance you can't run TEC's at Imax, which is complete non-sense.

You're also making linear calculations on DT, by saying that 60% of Imax = 60% of DTmax, which is certainly not the case. TEC's become more efficient when undervolted, and so performance wise there's no linear relation between DTmax and 60% of Imax. Temperature and Qc also greatly affects DT, which makes linear calculations even more absurd.

You have also stated that you can't achieve Tc=Subzero while running TEC's at "full load", you will need to elaborate on that. Full load could mean Qc=Qmax or Vin=Vmax.

You obviously have no clue on how TEC's operate. I gather you don't trust Ultrasonics information, so I will suggest reading ferrotecs thermoelectric reference guide.

 

[zipdogso]

I think you need to stop being so high and mighty.

It is volts (Vmax) that determines the operating point, at that operating point the current draw depends on the Dt (and incidently the hotside temp.). Just look at any chart of any TEC on a manufacturers site.

Your right about lower voltage = lower Dt one of the problems is that TECs are semiconductors and therefore non-linear and the decay of Dt is not linear relative to V. So 60% Vmax does not = 60% Dt, that is also a generally known fact and certainly Flak-spammer ( who has run TEC's.) knows this.

The other problem is that Dt depends not only on V but also on the cooling of the hotside and the load applied to the coldside. It can vary constantly in operation especially if the TEC is mounted directly on the CPU (which also varies in temp with operation.)

Since you are generally getting a lower Dt than you probably hope for Hotside cooling is paramount. Another well known fact.
If you have good cooling of the hotside and a relatively low load you can get subzero on the coldside with only a 25-30 degree Dt.
This subzero on the coldside not subzero on the load or not likely anyway. It will depend on the heat transfer inefficiencies.

Low temps are obtained with a well designed and made chiller using multiple TECs. The problems with multiple TECs are the current draw but by arranging the TECs in series/parallel groups it can help also the TECs are undervolted quite a lot perhaps down to 35% Vmax. Hotside cooling must be A1 and cooling of the coolant is incremantal as it passes the TECs. As the load (the coolant) cools the Dt rises slightly and current draw lowers slightly. And temperatures approaching -20 degrees are feasible.....not easy but feasible.

One fact that people do tend to forget that every attribute of the TEC depends generally on 2 or more other attributes so when one thing changes - it could just be the ambient temp which alter the hotside temp - it will alter other things... if the hotside temp changes by a couple of degrees it will affect the Dt and the current draw. As Elloquin mentioned in his long post just adjusting the voltage 0.1v changed everything.. The only one thing you can be sure of with a TEC is it's physical size and even that is variable enough from one TEC to another to affect your expected performance.

Before you start accusing people of not having run TECs etc i really think you need to get your own house in order first. From the look of the post list it looks like you have installed yourself as the resident TEC champion......shame you don't actually understand how they work yet !!!

 

[flak-spammer]

As I did say I wasn't going to post again in this thread, this is actually an informative post to show the basis of accuracy for Ultrasonics calculator.

In response to the accuracy of ultrasonics calculator, Ultrasonic has provided this information:

On some forms the accuracy of my TEC calculator has been in question.

So it seems the best way to prove the calculator is pretty much right, is to put it against other manufactures numbers.

First my app will go head to head with another TEC simulator then real test numbers performed by a different TEC manufacture.. Seems a fair test to me

Here is the Kryotherm TEC calculator in comparison to mine it's Pretty Pretty close including all the random stuff like COP!

Above is the Kryotherm pic
At delta of 4.4c (TH-TC)
while moving a load of 124 watts (QC(W)
which requires 7.14 amps

Below is my app
it gets
a delta of 4.43c
while moving a load of 124 watts
which requires 7.13 amps

Above is the Kryotherm pic
At delta of 54.2c (TH-TC)
while moving a load of 30 watts (QC(W)
which requires 9.11 amps

Below is my app
it gets
a delta of 54.69c
while moving a load of 30 watts
which requires 9.08 amps

Above is the Kryotherm pic
At delta of 13.5c (TH-TC)
while moving a load of 10 watts (QC(W)
which requires 1.39 amps

Below is my app
it gets
a delta of 13.47c
while moving a load of 10 watts
which requires 1.39 amps

Above is the Kryotherm pic

NOW WHAT HAPPENS IF WE CHANGE THE HOT SIDE TEMP TO 50C
At delta of 33.3c (TH-TC)
while moving a load of 20 watts (QC(W)
which requires 3.44 amps

Below is my app
it gets
a delta of 33.35c
while moving a load of 20 watts
which requires 3.45 amps

http://www.customthermoelectric.com/tecs/pdf/19911-5M31-28CZ_spec_sht.pdf

Here we have CTE's 400Qmax TEC with REAL test data so i've put a few points on their graph then i'll bang those numbers into my app and see how close i get. Now remember this is REAL test DATA V my theoretical data

Now at point A they get
At delta of 20c
while moving a load of 150 watts
which requires just over 12 amps

Below is my app
it gets
a delta of 20c
while moving a load of 150 watts
which requires 13.3 amps (within 10%)

Now at point B they get
a delta of 30c
while moving a load of 175 watts
which requires 20 amps

Below is my app
it gets
a delta of 28.8c (within 4%)
while moving a load of 175 watts
which requires 20.13 amps

Now at point C they get
a delta of 60c
while moving a load of 0 watts
which requires 28 amps

Below is my app
it gets
a delta of 62.0c ( i based my numbers on a delta of 62 which it's advertised as )
while moving a load of 0 watts
which requires 28.0 amps

Now at point D they get
a delta of 0c
while moving a load of 400 watts
which requires 28 amps

Below is my app
it gets
a delta of 0.0c
while moving a load of 0 watts
which requires 28.0 amps

Now obviously these is a slight variance in some of the numbers this is partly because im basing the DELTA on 62c while in there test it only achieves 60c and at the end of the day this is theoretical verses real data it's not going to be 100% right but it's DARN CLOSE !

The calculator is only calculating the TEC part of the equation and is not including the thermal resistance of the radiator, water, flow rate, copper water block, TIM. Then TEC then TIM again then copper then TIM again and lastly the Thermal resistance of the IHS and Silicone .

This is how people seem to use calculators.
Ambient is 22 so the hot side of the TEC will be 22c
I'll apply 12volts and that gives me a TEC cold side of -10
And therefore the CPU temps read from apps like core temp will be -10
Yeah Right!

At the end of the day it's just an estimator

While someone maybe unable to make a block that replicates their understanding of the widely available data, That does not make the mountain of data is wrong. In allot of cases they have miss estimated in the first place though a lack of understanding.

 

[zipdogso]

Nice one Flak.....havent seen that before.... it is all good.

A point to remember as Ultrasonic said ALL calculators, charts etc even manufacturers charts are Ballpark figures. They dont take into acount any heat transfer inefficiencies,incorrect mounting, wrong TIM etc.

The real results from a TEC will always be different. In a multiple TEC setup each TEC will be different.

The only way to be sure how a TEC operates is to run it...........

 

[xtremetechuk]

Its a shame we can't simulate the complete TEC setup.

 

[un-nefer]

Quote:

Originally Posted by flak-spammer;12994090
Here is the Kryotherm TEC calculator in comparison to mine it's Pretty Pretty close including all the random stuff like COP!

lol, that is one of the other TEC calculators I used to compare with a while back

For anyone interested, it can be downloaded from HERE.

The only problem with it, is just like Ultrasonic's TEC calculators, you can not add more TECs to it - you are stuck with what they include.

I have pretty much moved to THIS one now, because it allows me to enter the TEC specs myself, and a good feature it includes is the ability to estimate your CPU heatload at a specific overclock

And like the Kryotherm calculator, it actually tries to calculate the actual heatsource temp (ie. the CPU temp) by allowing you to enter the coldplate and cooler thermal resistance (only really usefull if you go direct die TEC cooling though).

It is interesting to compare all three actually, as some results from each TEC calculator are almost exactly the same, but at other times this one gives slightly different results - sometimes a higher and sometimes lower.

30W results:

10W results:

I guess the only real way to see how accurate they are would be to buy and test a Kryotherm TEC listed in Ultrasonic's TEC calculator and the Kryotherm calculator.

 

[zipdogso]

Quote:

Originally Posted by un-nefer;12996558
I guess the only real way to see how accurate they are would be to buy and test a Kryotherm TEC listed in Ultrasonic's TEC calculator and the Kryotherm calculator.

Personally I wouldn't bother, it wont achieve anything worthwhile.

You cannot run a TEC in real tests then compare the results back to any calculator, chart or what have you....Real testing will have loads of inefficiencies built in while the calculators and charts have absolutely 0% inefficiencies built in. It would be like comparing Chalk and Cheese.

ALL CALCULATORS, CHARTS ETC GIVE BALLPARK FIGURES !!!!

Just choose the calculator your happy with, check to see if you can achieve what you want, set your operating point and run it.

Manufacturers charts etc are only to give people a starting point so you have an idea if the TEC you've chosen will be able to do what your hoping....thats it.... nothing more. If they didn't provide at least something to start from no-one would have any idea if a TEC is suitable for their use or not.

 

[lichking]

Quote:

Originally Posted by zipdogso;12994002
i think you need to stop being so high and mighty.

It is volts (vmax) that determines the operating point, at that operating point the current draw depends on the dt (and incidently the hotside temp.). Just look at any chart of any tec on a manufacturers site.

its amps that determines dt, and volts that determines amps.

your right about lower voltage = lower dt one of the problems is that tecs are semiconductors and therefore non-linear and the decay of dt is not linear relative to v. So 60% vmax does not = 60% dt, that is also a generally known fact and certainly flak-spammer ( who has run tec's.) knows this.

really not sure that flack does know this

Before you start accusing people of not having run tecs etc i really think you need to get your own house in order first. From the look of the post list it looks like you have installed yourself as the resident tec champion......shame you don't actually understand how they work yet !!!
ROFL

^^^