Overclock.net › Forums › Cooling › Specialized Cooling › Peltiers / TEC › Peltier/TEC Hybrid Water Cooling vs 2.0
New Posts  All Forums:Forum Nav:

Peltier/TEC Hybrid Water Cooling vs 2.0 - Page 15

post #141 of 389
Quote:
Originally Posted by toeCUTTER78 View Post

How much heat do you think a 360 would remove in watts ?

Please explain the CTE400 calculation, if I am reading your message correctly you have calculated power draw of 12V * 12,55A = 150w
and a cold side cooling power of 140w - thus making the tec ~90% efficient ? - surely I misunderstood something here ? biggrin.gif


/Toe

I like to have no more then 150w per 360 rad. Keep in mine this isnt plain water cooling, so we are not looking for "good enough" , we want as close to ambient as possible since the hot side basically directly impacts the cold side. While a 360 can probably hold 500-600w of heat itself, its actual dissipation is usually 150-200w.

For the TEC, you are not factoring in any delta smile.gif. That roughly 140w of cooling at 12v is around where the delta drops to 0c(hot side and cold side same temp)...and the TEC starts hurting instead of helping beyond that. Performance on the CTE400 starts dropping dramatically after 120w or so. Due to that, they work better in multiples for chillers then a direct die on a modern CPU. The TN660 is a whole different animal and can handle ~75w more then the CTE400 when both at 12v.

Keep in mind all TEC datasheets are theoretical and seem to be extrapolated from the per couple performance...IE, they are pretty much all advertised optimistically.
Behemoth
(25 items)
 
  
CPUMotherboardGraphicsGraphics
Intel i7 4790K "Relidded" @ 5ghz <50c Asus Sabertooth Z97 MK1 Custom WC'd 7970 3GB @ 1100/1500 stock volts Custom WC'd 7970 3GB @ 1100/1500 stock volts 
RAMHard DriveCoolingCooling
2X4GB Patriot Viper Extreme 1TB Velociraptor 10k RPM w/ 64MB Cache Two Arqtik TEC Chillers in Series Archimedes TEC block w/TN660 TEC 
CoolingCoolingCoolingCooling
Two Ported/tapped MCP655 Pumps 2x Black Ice GTX 480 Gen2 Rads DD dual drive bay Res with LCD water temp monit... GPU Loop: Modded Enzotech Sapphire CPU blocks, ... 
CoolingOSOSMonitor
36x Globe 120mm 80cfm server fans Win 10 Pro Mint 15(KDE) 4K 28" Samsung U590D 
KeyboardPowerCaseMouse
Logitech G710+ Mechanical CM Silent Power M 1000watt Main, OCZ Z850W Slave Custom Mountain Mods Extended-Extended Ascension Logitech G502 Proteus 
Mouse PadAudioAudioAudio
Homemade 16ga 304 Stainless pad, Neoprene backed Aune T1 DAC Various Tubes: Mostly switching from 60's GE "S... AKG K7XX Headphones 
  hide details  
Reply
Behemoth
(25 items)
 
  
CPUMotherboardGraphicsGraphics
Intel i7 4790K "Relidded" @ 5ghz <50c Asus Sabertooth Z97 MK1 Custom WC'd 7970 3GB @ 1100/1500 stock volts Custom WC'd 7970 3GB @ 1100/1500 stock volts 
RAMHard DriveCoolingCooling
2X4GB Patriot Viper Extreme 1TB Velociraptor 10k RPM w/ 64MB Cache Two Arqtik TEC Chillers in Series Archimedes TEC block w/TN660 TEC 
CoolingCoolingCoolingCooling
Two Ported/tapped MCP655 Pumps 2x Black Ice GTX 480 Gen2 Rads DD dual drive bay Res with LCD water temp monit... GPU Loop: Modded Enzotech Sapphire CPU blocks, ... 
CoolingOSOSMonitor
36x Globe 120mm 80cfm server fans Win 10 Pro Mint 15(KDE) 4K 28" Samsung U590D 
KeyboardPowerCaseMouse
Logitech G710+ Mechanical CM Silent Power M 1000watt Main, OCZ Z850W Slave Custom Mountain Mods Extended-Extended Ascension Logitech G502 Proteus 
Mouse PadAudioAudioAudio
Homemade 16ga 304 Stainless pad, Neoprene backed Aune T1 DAC Various Tubes: Mostly switching from 60's GE "S... AKG K7XX Headphones 
  hide details  
Reply
post #142 of 389
Thread Starter 
Quote:
Originally Posted by Puck View Post

Keep in mine this isn't plain water cooling, so we are not looking for "good enough" , we want as close to ambient as possible since the hot side basically directly impacts the cold side.

Puck, You should add that sentence to your sig, It is priceless! Ryan
post #143 of 389
Originally Posted by Puck View Post

Keep in mine this isn't plain water cooling, so we are not looking for "good enough" , we want as close to ambient as possible since the hot side basically directly impacts the cold side.

+1
post #144 of 389
Quote:
Originally Posted by toeCUTTER78 View Post

How much heat do you think a 360 would remove in watts ?

Please explain the CTE400 calculation, if I am reading your message correctly you have calculated power draw of 12V * 12,55A = 150w
and a cold side cooling power of 140w - thus making the tec ~90% efficient ? - surely I misunderstood something here ? biggrin.gif


/Toe

If you study the C.O.P charts on the third page of the TN669 TEC data sheet, it will help you (and others) understand just how efficient TEC's can be when undervolted.

http://www.thermonamic.com/TEHC1-19940-English.PDF

For instance, at 12v with a target of 40*c dT, the TEC can be 50% efficient, ie has a C.O.P of 0.5. This is a 199 couple TEC, so 24v, meaning 12v is a 50% undervolt as well.

At 3v with a dT of 10*c it is 400% efficient, ie C.O.P of 4 but it would only move about 25-50w of Qc.

You need to factor in that at lower voltages the max delta is reduced, and the Qc is reduced, ie at 3v it can only reach a delta of 10-15*c, and only move about 0-50w, but if you find the balance at about 50-70% of Umax (voltage) you have reasonable efficiency, in terms of power used to heat moved and you can still get a pretty good dT of 20-40*c.

The data sheets are not accurate so I usually derate the Qc wattage a bit or reduce the dT a bit to what is shown on the data sheet.
Edited by LiamG6 - 4/10/17 at 7:38am
post #145 of 389
Quote:
Originally Posted by toeCUTTER78 View Post

How much heat do you think a 360 would remove in watts ?

Please explain the CTE400 calculation, if I am reading your message correctly you have calculated power draw of 12V * 12,55A = 150w
and a cold side cooling power of 140w - thus making the tec ~90% efficient ? - surely I misunderstood something here ? biggrin.gif


/Toe

Toe,

Puck has given a pretty good answer so far; but I'll chime in. I'm not sure how much you know about TECs so I'll just give a full response.

The hot side heat load should be calculated by:
Cold side heat load(W) + V * A

As you say, if you are supplying 12v, the peltier would draw above 12.5a, which gives you (as you pointed out) 150w.
Now if your cold side is managing a heat load of 80w, then you have a total heat load:

80w + 150w = 230w

Now typically you can make a cold side estimate of the pelt by using a Coefficient of Performance (COP) chart. The COP is calculated by dividing the watts of cooling (Qc) by the watts required to cool (W or Work). You'll typically see charts with COP vs Amps. CTE, who makes the 400w pelt Puck referenced, does it a little different. They have a Volts vs Amps chart and a Qc vs Amps chart.
The charts are here:
http://customthermoelectric.com/tecs/pdf/19911-5M31-28CZ_spec_sht.pdf




In the above picture, I've drawn lines on to the chart to clearly show how you would go from knowing that you will supply 12v to finding the max amp draw you could expect (right chart).
Then you can use that amp draw to find the delta temperature (dT) at a given cold side heatload.

Now, the actual cold side temperature (Tc) of the peltier is calculated by Th - dT. What this tells us is the delta temperature is relative to the hot side. So if the hot side is running at 60c because you only have one 360mm radiator on it, then assuming the 80w of load and perfect efficiency, you would only have a cold side temperature of 30c.



I wonder if it wouldn't be a good idea for some of the guys here who know all the math and have experience in building TEC setups to write a general information sticky that goes over things like:
  • Heat pumping capability at a given voltage
  • Current draw at a given voltage and included considerations (such as wire gauge and power source)
  • Interpreting performance charts that typically accompany peltiers
  • Heat output and methods of dissipation
  • PWM control and voltage control
  • etc.

I have not yet found a single source to find all of this information. And I will be the first to admit that it can require quite a bit of searching to find out some of the more basic stuff. In addition, there's a lot of misinformation that goes around which can be very confusing to sort through for newer people.
There's also the factor that if you don't know what you need to know it will be awfully hard to search for it.


EDIT:
I see LiamG6 got there first. I'm still posting it as I spent too much time typing biggrin.gif
Edited by Skyl3r - 4/10/17 at 7:41am
post #146 of 389
Quote:
Originally Posted by toeCUTTER78 View Post

How much heat do you think a 360 would remove in watts ?

Please explain the CTE400 calculation, if I am reading your message correctly you have calculated power draw of 12V * 12,55A = 150w
and a cold side cooling power of 140w - thus making the tec ~90% efficient ? - surely I misunderstood something here ? biggrin.gif


/Toe



I don't know the math yet, so I am not going to give you some technical explanation, but I do know the real world outcome, the hot side of the TEC, seems to produce about twice the amount of heat, as the cold side is producing cold.

so a 250Watt TEC cold side, seems to produce about 500Watts worth of heat to the hot side.

edit IMO it seems like when ebay lists the wattage for there TEC's, they are listing what the hot side heat created is. rolleyes.gif

example: ebay listing for a 430Watt TEC is a 12726 or a 16volt 430Watt 26AMP TEC. (hot side heat created)
CTE listing for the same TEC is a 12711-5P31-26CW or a 16volt 245Watt 26AMP TEC. (cold side watts measured)
Edited by toolmaker03 - 4/10/17 at 7:45am
post #147 of 389
Quote:
Originally Posted by Skyl3r View Post

Toe,

Puck has given a pretty good answer so far; but I'll chime in. I'm not sure how much you know about TECs so I'll just give a full response.

The hot side heat load should be calculated by:
Cold side heat load(W) + V * A

As you say, if you are supplying 12v, the peltier would draw above 12.5a, which gives you (as you pointed out) 150w.
Now if your cold side is managing a heat load of 80w, then you have a total heat load:

80w + 150w = 230w

Now typically you can make a cold side estimate of the pelt by using a Coefficient of Performance (COP) chart. The COP is calculated by dividing the watts of cooling (Qc) by the watts required to cool (W or Work). You'll typically see charts with COP vs Amps. CTE, who makes the 400w pelt Puck referenced, does it a little different. They have a Volts vs Amps chart and a Qc vs Amps chart.
The charts are here:
http://customthermoelectric.com/tecs/pdf/19911-5M31-28CZ_spec_sht.pdf




In the above picture, I've drawn lines on to the chart to clearly show how you would go from knowing that you will supply 12v to finding the max amp draw you could expect (right chart).
Then you can use that amp draw to find the delta temperature (dT) at a given cold side heatload.

Now, the actual cold side temperature (Tc) of the peltier is calculated by Th - dT. What this tells us is the delta temperature is relative to the hot side. So if the hot side is running at 60c because you only have one 360mm radiator on it, then assuming the 80w of load and perfect efficiency, you would only have a cold side temperature of 30c.



I wonder if it wouldn't be a good idea for some of the guys here who know all the math and have experience in building TEC setups to write a general information sticky that goes over things like:
  • Heat pumping capability at a given voltage
  • Current draw at a given voltage and included considerations (such as wire gauge and power source)
  • Interpreting performance charts that typically accompany peltiers
  • Heat output and methods of dissipation
  • PWM control and voltage control
  • etc.

I have not yet found a single source to find all of this information. And I will be the first to admit that it can require quite a bit of searching to find out some of the more basic stuff. In addition, there's a lot of misinformation that goes around which can be very confusing to sort through for newer people.
There's also the factor that if you don't know what you need to know it will be awfully hard to search for it.


EDIT:
I see LiamG6 got there first. I'm still posting it as I spent too much time typing biggrin.gif

Your explanation expands on mine so both are relelvant. I was going to post the same little overlay of the CTE charts with the lines you;ve drawn in haha. But the COP charts explain the possible efficiency perfectly.

As to the sticky idea, i believe over at XS (xtremesystems) they used to have a sticky, that is where I cut my teeth with guys like Ultrasonic 2, mindchill etc. You could always do one up as you seem to have a decent grasp of TEC's wink.gif
post #148 of 389
Quote:
Originally Posted by toolmaker03 View Post

I don't know the math yet, so I am not going to give you some technical explanation, but I do know the real world outcome, the hot side of the TEC, seems to produce about twice the amount of heat, as the cold side is producing cold.

so a 250Watt TEC cold side, seems to produce about 500Watts worth of heat to the hot side. thumb.gif

edit IMO it seems like when ebay lists the wattage for there TEC's, they are listing what the hot side heat created is. rolleyes.gif

example: ebay listing for a 430Watt TEC is a 12726 or a 16volt 430Watt 26AMP TEC. (hot side heat created)
CTE listing for the same TEC is a 12711-5P31-26CW or a 16volt 245Watt 26AMP TEC. (cold side watts measured)

The hot side heat load is the sum of the cold side heat load plus the power consumption of the TEC, this is why I always advocate undervolting TEC's to 50-70% and using the most efficient TEC's you can find for the voltage of your PSU that still achieves the dT you want with the Qc you need to handle, that way your total heat load is reduced.

There is no set figure for how much power is used to pump a certain amount of heat from the cold side, it changes with voltage, with dT, with Qc and every TEC is different and because of that there is no rule of thumb, every TEC must be evaluated based on Qc, dT, voltage, Th etc.

everyone who never uses TEC's saying "TEC's aren't efficient, don't use them", just don't understand the possibilities of how efficient TEC's can be when undervolted
Edited by LiamG6 - 4/10/17 at 7:51am
post #149 of 389
Quote:
Originally Posted by LiamG6 View Post

If you study the C.O.P charts on the third page of the TN669 TEC data sheet, it will help you (and others) understand just how efficient TEC's can be when undervolted.

http://www.thermonamic.com/TEHC1-19940-English.PDF

For instance, at 12v with a target of 40*c dT, the TEC can be 50% efficient, ie has a C.O.P of 0.5. This is a 199 couple TEC, so 24v, meaning 12v is a 50% undervolt as well.

At 3v with a dT of 10*c it is 400% efficient, ie C.O.P of 4.

You need to factor in that at lower voltages the max delta is reduced, and the Qc is reduced, but if you find the balance at about 50-70% of Umax (voltage) you have reasonable efficiency, in terms of power used to heat moved and you can still get a pretty good dT of 20-40*c.

The data sheets are not accurate so I usually derate the Qc wattage a bit or reduce the dT a bit to what is shown on the data sheet.
Well said.

Lots of rough numbers here but this is one reason why a lot of new users experimenting with TECs run into issues...common train of thought is to grab the highest rated TEC they can find on ebay for the size they need, then run it at its rated voltage in hope to reach those big (fake) numbers. What ends up happening in reality, is hundreds of watts of heat dumped into too small of a cooling loop, leading to hot side temps that can creep into the 40s. The TECs are working off a delta, and if your application is chugging along with a 20c delta...congrats, you just basically tied a plain water cooling loop since your cold side is now 20* vs a potential ~23* on a WC loop with multiple rads. A lot of money, power usage, and heat output just to drop 3* from an Evo block and 480 rad redface.gif. Same scenario with a 2nd or third rad or bigger undervolted TEC? Now you have a 28-30c hot side and a chilly 10c cold plate with the same 20c delta, and a potential ~15c colder then water thumb.gif.

I'm sure this has happened to most of us in one way or another here...my first setup I tried to get away with just a low profile Swiftech 320 rad using the "at least 120mm per device" rule that normal WC users quote and it was terrible. Coolant temps were high enough that my big dual bay res was warm to the touch through the plexi, think they hit 37c LOL.
Behemoth
(25 items)
 
  
CPUMotherboardGraphicsGraphics
Intel i7 4790K "Relidded" @ 5ghz <50c Asus Sabertooth Z97 MK1 Custom WC'd 7970 3GB @ 1100/1500 stock volts Custom WC'd 7970 3GB @ 1100/1500 stock volts 
RAMHard DriveCoolingCooling
2X4GB Patriot Viper Extreme 1TB Velociraptor 10k RPM w/ 64MB Cache Two Arqtik TEC Chillers in Series Archimedes TEC block w/TN660 TEC 
CoolingCoolingCoolingCooling
Two Ported/tapped MCP655 Pumps 2x Black Ice GTX 480 Gen2 Rads DD dual drive bay Res with LCD water temp monit... GPU Loop: Modded Enzotech Sapphire CPU blocks, ... 
CoolingOSOSMonitor
36x Globe 120mm 80cfm server fans Win 10 Pro Mint 15(KDE) 4K 28" Samsung U590D 
KeyboardPowerCaseMouse
Logitech G710+ Mechanical CM Silent Power M 1000watt Main, OCZ Z850W Slave Custom Mountain Mods Extended-Extended Ascension Logitech G502 Proteus 
Mouse PadAudioAudioAudio
Homemade 16ga 304 Stainless pad, Neoprene backed Aune T1 DAC Various Tubes: Mostly switching from 60's GE "S... AKG K7XX Headphones 
  hide details  
Reply
Behemoth
(25 items)
 
  
CPUMotherboardGraphicsGraphics
Intel i7 4790K "Relidded" @ 5ghz <50c Asus Sabertooth Z97 MK1 Custom WC'd 7970 3GB @ 1100/1500 stock volts Custom WC'd 7970 3GB @ 1100/1500 stock volts 
RAMHard DriveCoolingCooling
2X4GB Patriot Viper Extreme 1TB Velociraptor 10k RPM w/ 64MB Cache Two Arqtik TEC Chillers in Series Archimedes TEC block w/TN660 TEC 
CoolingCoolingCoolingCooling
Two Ported/tapped MCP655 Pumps 2x Black Ice GTX 480 Gen2 Rads DD dual drive bay Res with LCD water temp monit... GPU Loop: Modded Enzotech Sapphire CPU blocks, ... 
CoolingOSOSMonitor
36x Globe 120mm 80cfm server fans Win 10 Pro Mint 15(KDE) 4K 28" Samsung U590D 
KeyboardPowerCaseMouse
Logitech G710+ Mechanical CM Silent Power M 1000watt Main, OCZ Z850W Slave Custom Mountain Mods Extended-Extended Ascension Logitech G502 Proteus 
Mouse PadAudioAudioAudio
Homemade 16ga 304 Stainless pad, Neoprene backed Aune T1 DAC Various Tubes: Mostly switching from 60's GE "S... AKG K7XX Headphones 
  hide details  
Reply
post #150 of 389
Quote:
Originally Posted by Puck View Post

Well said.

Lots of rough numbers here but this is one reason why a lot of new users experimenting with TECs run into issues...common train of thought is to grab the highest rated TEC they can find on ebay for the size they need, then run it at its rated voltage in hope to reach those big (fake) numbers. What ends up happening in reality, is hundreds of watts of heat dumped into too small of a cooling loop, leading to hot side temps that can creep into the 40s. The TECs are working off a delta, and if your application is chugging along with a 20c delta...congrats, you just basically tied a plain water cooling loop since your cold side is now 20* vs a potential ~23* on a WC loop with multiple rads. A lot of money, power usage, and heat output just to drop 3* from an Evo block and 480 rad redface.gif. Same scenario with a 2nd or third rad or bigger undervolted TEC? Now you have a 28-30c hot side and a chilly 10c cold plate with the same 20c delta, and a potential ~15c colder then water thumb.gif.

I'm sure this has happened to most of us in one way or another here...my first setup I tried to get away with just a low profile Swiftech 320 rad using the "at least 120mm per device" rule that normal WC users quote and it was terrible. Coolant temps were high enough that my big dual bay res was warm to the touch through the plexi, think they hit 37c LOL.

If there was ever a rule of thumb when it comes to TEC newbies, it should be undervolt to 50-70% of Umax haha, that would saves us a lot of headaches.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Peltiers / TEC
Overclock.net › Forums › Cooling › Specialized Cooling › Peltiers / TEC › Peltier/TEC Hybrid Water Cooling vs 2.0