The i7's high TDP maybe outside the efficient range of a 437w TEC.
TEC temperature controller
Hi, I'm new here to the forum
![]()
I was hoping for some help or input from you guys to get a good temperature control for a TEC. I'm thinking of cooling my future core i7 with a high cooling power TEC, somewhere in the 437W range, direct die of course, a little undervolted for efficiency. I'll be cooling the hot side with 2 triple TFC radiators and 3k rpm ultra kaze fans which for 24/7 use will be undervolted, only when benching they'll run at 12V, that should provide the necessary cooling for the hot side.
Now for my question, I would like a good temperature control for the TEC to keep condensation to a minimum for 24/7 use. I know it has been done before, but I can't seem to find anyone who explaned it decently or that had a good temperature control. I read that someone on the forum used a t-balancer to do it, but is it effective and accurate? Basically I need a temperature controller that controls the power provided to the TEC, according to the cold plate temp that is measured, around a setpoint that is set in the controller.
I have no idea how well or how "fast" TECs respond to temperature control when given a sudden CPU heat load for example, if anyone has an answer to that...
Wouldn't it be more accurate for example buying a temperature controller with build in functions like PWM, PID control to get a better automated temperature control?
For example here is a link to one of the many manufacturers:
Temperature Controllers and Thermal Management Solutions
They sell a pretty decent temp controller, but of course you need the H-bridge amplifier to be able to control high power TECs.
I'm just looking for a good solution for 24/7 use to keep condensation to a minimum... Any input is welcome on what instruments to use etc!
Thanks and hi to everyone since I'm new here
![]()
I was hoping for some help or input from you guys to get a good temperature control for a TEC. I'm thinking of cooling my future core i7 with a high cooling power TEC, somewhere in the 437W range, direct die of course, a little undervolted for efficiency. I'll be cooling the hot side with 2 triple TFC radiators and 3k rpm ultra kaze fans which for 24/7 use will be undervolted, only when benching they'll run at 12V, that should provide the necessary cooling for the hot side.
Now for my question, I would like a good temperature control for the TEC to keep condensation to a minimum for 24/7 use. I know it has been done before, but I can't seem to find anyone who explaned it decently or that had a good temperature control. I read that someone on the forum used a t-balancer to do it, but is it effective and accurate? Basically I need a temperature controller that controls the power provided to the TEC, according to the cold plate temp that is measured, around a setpoint that is set in the controller.
I have no idea how well or how "fast" TECs respond to temperature control when given a sudden CPU heat load for example, if anyone has an answer to that...
Wouldn't it be more accurate for example buying a temperature controller with build in functions like PWM, PID control to get a better automated temperature control?
For example here is a link to one of the many manufacturers:
Temperature Controllers and Thermal Management Solutions
They sell a pretty decent temp controller, but of course you need the H-bridge amplifier to be able to control high power TECs.
I'm just looking for a good solution for 24/7 use to keep condensation to a minimum... Any input is welcome on what instruments to use etc!
Thanks and hi to everyone since I'm new here
2
1 - 20 of 72 Posts
Well, I know it won't be easy to keep an i7 cool, this forum is filled with that kind of observations, you're right, but maybe I'll cool it in the less efficient range then
![]()
It should still be possible right, with very good watercooling present.
It should still be possible right, with very good watercooling present.
Quote:
Are there precedents of people achieving a decent overclock (+4.2ghz) on a core i7 using high power TECs? Let's say I run the TEC at 80% Umax for benching. I know every chip is different but still. If anything, when done right the achieved overclock should be better than on water only right.
For 24/7 use I won't be using an extreme overclock so that shouldn't be a problem. I just don't really know what overclock can be achieved with let's say a 437W TEC at 80% and 2 triple rads cooling the hot side and a D5 pump. And only for the lower overclock for 24/7 use I'll be using the temperature control, for benching I'll set a low temperature setpoint so that the TEC is constantly powered at for example 75-80%.
I actually read about someone using a 288W TEC and keeps his temps below 30°C under load with an i7 965 @4.1ghz, so that looks like a good indication, although it was probably run more likely at 90% Umax or something, so that seems to leaves some headroom for a higher cooling power TEC with high-end watercooling.
Originally Posted by flak-spammer  well it depends on what temps you are looking for. With what you have proposed with an overclock and under load you'd probably achieve 15-30C at 75-100% Umax (anything over 80% not recommended). |
For 24/7 use I won't be using an extreme overclock so that shouldn't be a problem. I just don't really know what overclock can be achieved with let's say a 437W TEC at 80% and 2 triple rads cooling the hot side and a D5 pump. And only for the lower overclock for 24/7 use I'll be using the temperature control, for benching I'll set a low temperature setpoint so that the TEC is constantly powered at for example 75-80%.
I actually read about someone using a 288W TEC and keeps his temps below 30°C under load with an i7 965 @4.1ghz, so that looks like a good indication, although it was probably run more likely at 90% Umax or something, so that seems to leaves some headroom for a higher cooling power TEC with high-end watercooling.
Quote:
That would be me IT works perfectly fine as a PWM controller for TEC's i wouldn't live without it SERIOUSLY . I just set it to "target temp" which increases and decreases the PWM
It keeps the cold plate at X temp which is just what you want. The sensitivity can be adjusted too which is a good thing.
I must admit i do it more or saving power than anything else. With my current target temp it requires only 18% to keep my cold plate at its target when the CPU is idling
The problem with ANY controller is that it is reactive so you will get some variance in temps no matter what you do
Originally Posted by Nighthawk82 Hi, I'm new here to the forum
Now for my question, I would like a good temperature control for the TEC to keep condensation to a minimum for 24/7 use. I know it has been done before, but I can't seem to find anyone who explaned it decently or that had a good temperature control. I read that someone on the forum used a t-balancer to do it, but is it effective and accurate? Basically I need a temperature controller that controls the power provided to the TEC, according to the cold plate temp that is measured, around a setpoint that is set in the controller. I have no idea how well or how "fast" TECs respond to temperature control when given a sudden CPU heat load for example, if anyone has an answer to that... Wouldn't it be more accurate for example buying a temperature controller with build in functions like PWM, PID control to get a better automated temperature control? |
It keeps the cold plate at X temp which is just what you want. The sensitivity can be adjusted too which is a good thing.
I must admit i do it more or saving power than anything else. With my current target temp it requires only 18% to keep my cold plate at its target when the CPU is idling
The problem with ANY controller is that it is reactive so you will get some variance in temps no matter what you do
2
Quote:
I knew it was you, I read your entire thread, actually I read couple of them
![]()
Now that I have your attention...
![]()
How did you actually achieve the PWM function? I know you used the t-balancer which sends a signal proportional to the temp difference that needs to be neutralized, but is that constant changing voltage interpreted by the mosfets to give a certain amount of power to the TEC? Or is the TEC just turned on/off during periods of time, powering it fully on/off. I don't think so since you say it's PWM and not on/off. I think I'm just confused as how the mosfets react to the signal coming from the t-balancer. Sorry if this is a stupid question, I got my engineering degree but too bad it's not in electronics, I just would like to give it a shot
![]()
I won't be using the t-balancer though, I think I found a nice stand-alone temp controller with LCD display and lots of functions (Omron), I just don't know how to make the connection controller-PSU, just buying and SSR would make it only on/off switching right and I'd like PWM for better accuracy.
Originally Posted by Ultrasonic2 (muffy) That would be me IT works perfectly fine as a PWM controller for TEC's i wouldn't live without it SERIOUSLY . I just set it to "target temp" which increases and decreases the PWM It keeps the cold plate at X temp which is just what you want. The sensitivity can be adjusted too which is a good thing. I must admit i do it more or saving power than anything else. With my current target temp it requires only 18% to keep my cold plate at its target when the CPU is idling The problem with ANY controller is that it is reactive so you will get some variance in temps no matter what you do |
Now that I have your attention...
How did you actually achieve the PWM function? I know you used the t-balancer which sends a signal proportional to the temp difference that needs to be neutralized, but is that constant changing voltage interpreted by the mosfets to give a certain amount of power to the TEC? Or is the TEC just turned on/off during periods of time, powering it fully on/off. I don't think so since you say it's PWM and not on/off. I think I'm just confused as how the mosfets react to the signal coming from the t-balancer. Sorry if this is a stupid question, I got my engineering degree but too bad it's not in electronics, I just would like to give it a shot
I won't be using the t-balancer though, I think I found a nice stand-alone temp controller with LCD display and lots of functions (Omron), I just don't know how to make the connection controller-PSU, just buying and SSR would make it only on/off switching right and I'd like PWM for better accuracy.
4
Quote:
Yes indeed, with SSR I meant a solid state relay. I actually have no experience with those, but since they are solid state and not mechanical, I thought they'd make less or no noise? Or am I wrong here
![]()
And yes, powering a TEC with a controller alone is nearly impossible, I've found a temperature controller that can handle quite strong TECs, but I bet those cost a small fortune. But as I said I won't be using a t-balancer or any other PC-related controller, I think I'll buy a controller used in automation, they are actually not that expensive and quite nice if you could integrate them in your system. I just need to get a clearer view of the middle circuit working between the controller and the PSU, so I can make an informed decision. Guess I'll need to read up on that
![]()
Btw, since PWM is actually on/off, isn't that gonna shorten the lifetime of the peltier element? I thought I read that somewhere, not sure anymore. I was looking for a way to supply the TEC with a constant changing voltage, not really on/off, the voltage depending on the heat load it has to carry of course. Or is that what you're doing... I'm a bit confused here
![]()
EDIT: I was thinking something in the direction of PID control, so that it reacts fast on the temperature differentials that arise because of a sudden heat load for example. That's also the reason I asked in my first post how fast TECs respond to a voltage change.
Originally Posted by Ultrasonic2 (muffy) You do realize that PWM Is only on and off thats what PWM is The amount of power sent to the TEC's is based one what % of time the power is on for. You could in theory use a relay but they are to SLOW and they'd make the most annoying noise all day long The PWM signal is created by the T-balancer (or any PWM fan controller). But NO one makes a 600 watts fan controller the T-balancer can only handle a dubious 80watts so you need a way to boost his limit from 80 to 600 which is what my mosfets are for. the other reason for them is i can uses the t balancer and it's 12 supply form the PC to control an external PSU of ANY voltage. in my case my t- balancer controls two 24 volts psu's An important factor to getting this to work is you can not have any caps running on the out puts of the fan controller otherwise your external Mosfets will catch on fire literally. This is because they will be trying to resists some of the power instead of switching it does SSR stand for a relay ? |
And yes, powering a TEC with a controller alone is nearly impossible, I've found a temperature controller that can handle quite strong TECs, but I bet those cost a small fortune. But as I said I won't be using a t-balancer or any other PC-related controller, I think I'll buy a controller used in automation, they are actually not that expensive and quite nice if you could integrate them in your system. I just need to get a clearer view of the middle circuit working between the controller and the PSU, so I can make an informed decision. Guess I'll need to read up on that
Btw, since PWM is actually on/off, isn't that gonna shorten the lifetime of the peltier element? I thought I read that somewhere, not sure anymore. I was looking for a way to supply the TEC with a constant changing voltage, not really on/off, the voltage depending on the heat load it has to carry of course. Or is that what you're doing... I'm a bit confused here
EDIT: I was thinking something in the direction of PID control, so that it reacts fast on the temperature differentials that arise because of a sudden heat load for example. That's also the reason I asked in my first post how fast TECs respond to a voltage change.
4
Ok thanks for the information, so PWM it'll be
![]()
With what program did you do those simulations, seems quite useful:
![]()
After some research, this is 5Spice?
With what program did you do those simulations, seems quite useful:
After some research, this is 5Spice?
2
Quote:
Analogue is 20 watts
the T-balancer indirectly controls the TEC's Via the booster"
the PWM booster directly powers the TEC's but the Booster is controlled by the t-balancer
the booster can handle 600 watts at 24 volts where the T-balancer can't
Originally Posted by ChielScape youre saying you drive the TEC with your T-balancer, but isnt it rated for just 80W? or is that just for analogue mode? |
the T-balancer indirectly controls the TEC's Via the booster"
the PWM booster directly powers the TEC's but the Booster is controlled by the t-balancer
the booster can handle 600 watts at 24 volts where the T-balancer can't
Quote:
Also with a decent controller used in the automation industry? That would be strange since those things are made specifically to control temps as accurate as possible...
Just to be sure on how exactly it works on your end: so when the temp goes above your setpoint, the t-balancer sends a signal ("x"mA or "x"V? and is this signal always the same or does it vary) to the mosfet which now will power the TEC with the value the PSU is set at, so always full power. And as soon as the temp drops below the setpoint the signal from the t-balancer stops and no power goes to the TEC from the PSU.
Somewhere i think I got it wrong
![]()
Originally Posted by Ultrasonic2 (muffy) yes . Remember no matter what controller you buy it's almost guaranteed to have cap on it's out put to smooth the switching out some what . you'll have to remove them from what what controller you buy. |
Just to be sure on how exactly it works on your end: so when the temp goes above your setpoint, the t-balancer sends a signal ("x"mA or "x"V? and is this signal always the same or does it vary) to the mosfet which now will power the TEC with the value the PSU is set at, so always full power. And as soon as the temp drops below the setpoint the signal from the t-balancer stops and no power goes to the TEC from the PSU.
Somewhere i think I got it wrong
2
Quote:
Also with a decent controller used in the automation industry? That would be strange since those things are made specifically to control temps as accurate as possible...
if your not going to worry about the "booster " because the automotive controller is powerful enough to handle the high wattage TEC with out it . then you dont need to worry about the caps on the out puts. But if you plan on adding the Booster then you WILL have to remove the caps.
Just to be sure on how exactly it works on your end: so when the temp goes above your setpoint, the t-balancer sends a signal ("x"mA or "x"V? and is this signal always the same or does it vary) to the mosfet which now will power the TEC with the value the PSU is set at, so always full power. And as soon as the temp drops below the setpoint the signal from the t-balancer stops and no power goes to the TEC from the PSU.
so when the temp goes above your setpoint, the t-balancer sends a signal ("x"mA or "x"V?
it sends a 12 volts signal
and is this signal always the same or does it vary
the duty cycle changes.
the mosfet which now will power the TEC with the value the PSU is set at
Yes (24volts)
so always full power. And as soon as the temp drops below the setpoint the signal from the t-balancer stops and no power goes to the TEC from the PSU
No
Originally Posted by Nighthawk82 Also with a decent controller used in the automation industry? That would be strange since those things are made specifically to control temps as accurate as possible... Just to be sure on how exactly it works on your end: so when the temp goes above your setpoint, the t-balancer sends a signal ("x"mA or "x"V? and is this signal always the same or does it vary) to the mosfet which now will power the TEC with the value the PSU is set at, so always full power. And as soon as the temp drops below the setpoint the signal from the t-balancer stops and no power goes to the TEC from the PSU. Somewhere i think I got it wrong
|
Also with a decent controller used in the automation industry? That would be strange since those things are made specifically to control temps as accurate as possible...
if your not going to worry about the "booster " because the automotive controller is powerful enough to handle the high wattage TEC with out it . then you dont need to worry about the caps on the out puts. But if you plan on adding the Booster then you WILL have to remove the caps.
Just to be sure on how exactly it works on your end: so when the temp goes above your setpoint, the t-balancer sends a signal ("x"mA or "x"V? and is this signal always the same or does it vary) to the mosfet which now will power the TEC with the value the PSU is set at, so always full power. And as soon as the temp drops below the setpoint the signal from the t-balancer stops and no power goes to the TEC from the PSU.
so when the temp goes above your setpoint, the t-balancer sends a signal ("x"mA or "x"V?
it sends a 12 volts signal
and is this signal always the same or does it vary
the duty cycle changes.
the mosfet which now will power the TEC with the value the PSU is set at
Yes (24volts)
so always full power. And as soon as the temp drops below the setpoint the signal from the t-balancer stops and no power goes to the TEC from the PSU
No
2
Quote:
I know that, but thanks anyway for the help, I'm thankful for the time you take to help out in any way you can
![]()
1 cycle is a very short period of time and in that period the power would be 50% on and 50% off if your duty cycle would be 50%, so if you would measure for lets say 1 second then the PSU would provide power for 0.5 seconds to the TEC in total, although it actually has switched on and off many many times. That's it right? lol
![]()
The current going to the TEC's is then the average of the on offs (50%)
Actually the "average" current going through the TEC's would be the average of the on offs, the actual current would still be the max current, but only at certain moments. But yes, I know what you meant
![]()
Originally Posted by Ultrasonic2 (muffy) im not sure you fully understand how PWM works ... if you do then sorry . 1 it is ether on or off. full power or nothing 2 However PWM is always at a frequency we'll has 1million times a second 3 this means the signal can ether be on 1 million times a second or off 1mmilion times 4 At 50% of duty cycle it's not on for 1/2 a second then off for 1/2 second in a row it is in fact no,off,on,off...... a million times. 5 The current going to the TEC's is then the average of the on offs (50%) So im my case if it requires 20% of duty cycle to reach x temp then that is what it will run at . if the load increases then the duty cycle increases to what ever is required to keep it at x temp |
1 cycle is a very short period of time and in that period the power would be 50% on and 50% off if your duty cycle would be 50%, so if you would measure for lets say 1 second then the PSU would provide power for 0.5 seconds to the TEC in total, although it actually has switched on and off many many times. That's it right? lol
The current going to the TEC's is then the average of the on offs (50%)
Actually the "average" current going through the TEC's would be the average of the on offs, the actual current would still be the max current, but only at certain moments. But yes, I know what you meant
4
1 - 20 of 72 Posts
