Overpowering a peltier

[It_The_Cow]

Would it be smart to overpower a 15.8 volt peltier with a 17 volt supply? Also, how can you figure out how many amps a line is making?

 

[LimeyGreg]

No.

A peltier needs to be run at the correct voltage for it to work at it's most efficient. If you under voltage it then it will not draw the correct current and so will not cool as it is designed.

Similarly, if you overvolt it it will not be functioning at it's most efficient point and will in fact be producing excess heat so reducing the amount of heat being pumped from the cold plate.

Might I also take this time to point out that a power supply that is incapable of supplying the necessary current at the specified voltage will also lead to reduced output voltage. When too much current is drawn from an underrated supply the supply will only provide it's rated power. Simple power/voltage/current calculations will show this to be true.

Finally, for all those that may think a 300 watt computer psu will power a 150 watt peltier. The rated power of the 12v rail is not 300 watts, it is only a portion. Also, you need to allow about 30% minimum safety margin. Also, be sure that the power consumption you are looking at is the total power and not just the amount pumped from the cold side, otherwise you will end up buying a power supply that will be seriously underated.

 

[William]

hmm, I always went by the amperage needed, and not the voltage, but am I wrong. You can change that 15volt power supply into a 15.8 with some simple physics calcs and some resistors.

 

[LimeyGreg]

Any good psu will usually have a pot for voltage adjustment and +- 5% is usual, so that shouldn't be a problem.

I've read posts where people think that the Peltier will draw more current if you use a lower voltage, I can only assume they think it will try to maintain the correct power consumption. Lower EMF means lower current, assuming the resistance remains constant, and hence lower power.

I've lost the link for the best explanation I ever read on a Peltiers efficiency, but the guy who wrote it showed a much larger graph of the performance with respect to power (V & I) than the normal graphs you see on the spec sheets. It clearly showed the u-shape and the lowest point being the most efficient correlated with the Qmax and the voltage and current required to produce optimal cooling. I will try and find it and post it here.

Okay, just searched my old posts and found it. http://www.procooling.com/html/pelt-tech.shtml

 

[MAV]

LimeyGreg (May 14, 2001 07:29 p.m.):
No.

A peltier needs to be run at the correct voltage for it to work at it's most efficient. If you under voltage it then it will not draw the correct current and so will not cool as it is designed.

Similarly, if you overvolt it it will not be functioning at it's most efficient point and will in fact be producing excess heat so reducing the amount of heat being pumped from the cold plate.

Might I also take this time to point out that a power supply that is incapable of supplying the necessary current at the specified voltage will also lead to reduced output voltage. When too much current is drawn from an underrated supply the supply will only provide it's rated power. Simple power/voltage/current calculations will show this to be true.

Finally, for all those that may think a 300 watt computer psu will power a 150 watt peltier. The rated power of the 12v rail is not 300 watts, it is only a portion. Also, you need to allow about 30% minimum safety margin. Also, be sure that the power consumption you are looking at is the total power and not just the amount pumped from the cold side, otherwise you will end up buying a power supply that will be seriously underated.

Regarding the 300watt PSU, am i safe in assuming that the 300watt should be sufficient to run a 54watt peltier ?
Regards

 

[LimeyGreg]

If that is a stand alone psu then yes it should do just fine. You can assume that the peltier will probably need about twice the power quoted due to it's efficiency (or lack of). Say your pelt consumes 2x54 watts, your current requirement at 15v will be approx 7 amps.

If it's a computer psu then check the current rating of the rail you are using and find it's power output for that output line. Bear in mind that if you run it at the 12v supplied by a computer psu the heat pumped will be considerably less, and if you try using it on an AMD with air cooling you will probably end up with cpu temps greater than just the HSF alone.

What cpu and method of cooling will you be using ?

 

[It_The_Cow]

Hmmm... Maybe I was using the voltmeter on the wrong setting. I'll try again. The problem is that I blow the fuse every time I try to get the readings for the amps being drawn

 

[LimeyGreg]

It doesn't matter what range you have the current setting on, it is still a shunt and all the current will still pass through the meters fuse, and your meter is probably not capable of reading more than 2 amps, at least my Fluke isn't.

You need to get a 1 ohm high current precision resistor and put it in series with the supply to the peltier. Then measure the voltage across it, whatever it is the current is the same value - except in amps of course.

You could measure the resistance of the peltier first and then use that along with the voltage across it to calculate the current and the power, but that will only hold true if the resistance doesn't change with temperature. I suspect the resistance will change as the performance graphs for peltiers are far from linear.

 

[MAV]

LimeyGreg (May 15, 2001 08:46 a.m.):
If that is a stand alone psu then yes it should do just fine. You can assume that the peltier will probably need about twice the power quoted due to it's efficiency (or lack of). Say your pelt consumes 2x54 watts, your current requirement at 15v will be approx 7 amps.

If it's a computer psu then check the current rating of the rail you are using and find it's power output for that output line. Bear in mind that if you run it at the 12v supplied by a computer psu the heat pumped will be considerably less, and if you try using it on an AMD with air cooling you will probably end up with cpu temps greater than just the HSF alone.

What cpu and method of cooling will you be using ?

Thanks for the reply Limeygreg, I've got a PIII 800E and using air cooling to cool the TEC, i've tried the TEC with air cooling on the coolermaster heatsink, the heatsink barely gets warm, the cold side of the TEC really gets cold and condensation starts to form with a bit of ice, though i did not have it connected for too long...just held it against the heatsink with some Dow Corning compound for a few minutes to see how the thing worked. But my next question is this : How do i mount the whole lot together ? The dimensions of the TEC are 40 x 44 x 4mm. This raises the heatsink obviously 4mm from the hold down on the socket 370 motherboard and then i still need to install a cold plate as well which will raise the heatsink even more. Any ideas or suggestions most welcome.
P.S. the amp rating on the TEC is 5.6amps max.
Regards

 

[LimeyGreg]

Okay, well you are going to have to either extend your present hold-down device or change it completely. It will depend on your type of chip, is it Slot 1 or FCPGA ?

Your best results will be realized if you make the cold plate from copper, it needs to be larger than the peltier but smaller than the HSF. For optimum cooling the peltier needs to be mounted with 150~300 psi clamping pressure depending on the device. You obviously can't achieve this with the 370 socket tabs, so drill and countersink the cold plate and drill/tap the HSF accordingly - now you can clamp the peltier to the HSF with the cold plate.

There are many different hold down designs and a search the web will yield several ideas. You may not to want to go to the trouble of mounting the cold plate in the above manner but an adjustable clamp is your best option, preferably with spring tensioned screws that will prevent overtightening.

Don't forget the insulation, closed cell neoprene (an old mouse pad ), dielectric silicone grease (automotive tune up grease) for the socket and conformal silicone RTV (more runny than regular silicone) for the back of the socket.

Is your peltier "potted", if not you may want to seal it around the edges to keep out condensation. The main problem here is the material used, if you use regular silicone RTV it usually gives off acetic acid fumes which when trapped in the peltier can cause long term corrosion. A good hi temp epoxy such as say JB Weld may be a better option. There are medical grade RTV's that do not give off acetic acid, but they are not as easily obtained, here is one such company http://www.nusil.com/

I hope that gives you a basis for your installation.

 

[It_The_Cow]

I've got a slot 1 PIII 750 mhz cpu. The task i'm trying to accomplish is to find how many amps a line on my PSU makes. I plan on making a modification by switching some of the mainboard wires and the 12 volt line wires. The only thing I'm worried about is whether or not enough amps will be drawn. The PSU is a 250 watt PSU which is supposed to push 12 amps on the 12 volt line

 

[MAV]

LimeyGreg (May 16, 2001 02:19 p.m.):
Okay, well you are going to have to either extend your present hold-down device or change it completely. It will depend on your type of chip, is it Slot 1 or FCPGA ?

Your best results will be realized if you make the cold plate from copper, it needs to be larger than the peltier but smaller than the HSF. For optimum cooling the peltier needs to be mounted with 150~300 psi clamping pressure depending on the device. You obviously can't achieve this with the 370 socket tabs, so drill and countersink the cold plate and drill/tap the HSF accordingly - now you can clamp the peltier to the HSF with the cold plate.

There are many different hold down designs and a search the web will yield several ideas. You may not to want to go to the trouble of mounting the cold plate in the above manner but an adjustable clamp is your best option, preferably with spring tensioned screws that will prevent overtightening.

Don't forget the insulation, closed cell neoprene (an old mouse pad ), dielectric silicone grease (automotive tune up grease) for the socket and conformal silicone RTV (more runny than regular silicone) for the back of the socket.

Is your peltier "potted", if not you may want to seal it around the edges to keep out condensation. The main problem here is the material used, if you use regular silicone RTV it usually gives off acetic acid fumes which when trapped in the peltier can cause long term corrosion. A good hi temp epoxy such as say JB Weld may be a better option. There are medical grade RTV's that do not give off acetic acid, but they are not as easily obtained, here is one such company http://www.nusil.com/

I hope that gives you a basis for your installation.

Thanks for the info and help LimeyGreg,much appreciated, the CPU is an PIII 800E FCPGA @ 100fsb,btw, you mentioned the word "Potted" i assume you mean is it sealed around the edges, no it is'nt....i have got some Dielectric tune - up grease, i suppose that will do....as for the silicone, the silicone i have at the moment is for marine fish tanks, i'm into tropical fish as well...i'll see if i can locate an old mouse pad as well, if not, i'll go and buy one thy're cheap....
Regards

 

[LimeyGreg]

Okay, now we are down to the brass tacks. What are the peltier specs? Qmax, Vmax and Imax. That 12v rail will supply approx 144 watts. If you try and draw more, the voltage will start to fall until the psu shuts itself down. !44 watts means it should power a Qmax 70 watt peltier, but unless it's a 12v peltier the actual heat pumped will be less and so the power consumption will be less.

For mounting on a slot one take a look at the article on the front page titled "Double Peltier CPU Cooling" by Joe Citrarella, this is the type of set up I was trying to explain except you subsitute the HSF for the waterblock.

A good explanation of Peltier calculations and properties can be found here http://www.procooling.com/html/pelt-tech.shtml

 

[It_The_Cow]

I plan on getting a FROST-74 from Tedist.com. It has an Imax of 6.1 amps, Qmax of 65.0 watts, and a Vmax of 16.7 volts, and a DeltaT of 74. And about the volts, it was supposed to be 16.7 volts contrary to my previous statement of 15.8 volts. Sorry

 

[It_The_Cow]

By the way, the bastards at KommaxComponents.com wouldn't replace my PSU. So now, it can only push 10 amps on the 12 volt line

 

[LimeyGreg]

The peltier you want will need 101 watts if it is operated at the requisite Vmax Imax, so your psu will power it, but as you are only going to be running it at 12v your power requirements will be less.

Unfortunately as you are only going to be running it at 12v you will only realize about 2/3 the cooling, so bear that in mind when you are looking at the results.

 

[Badger]

It_The_Cow (May 17, 2001 03:36 p.m.):
I plan on getting a FROST-74 from Tedist.com. It has an Imax of 6.1 amps, Qmax of 65.0 watts, and a Vmax of 16.7 volts, and a DeltaT of 74. And about the volts, it was supposed to be 16.7 volts contrary to my previous statement of 15.8 volts. Sorry

I would not run that pelt from a 12V supply if I were you. It won't damage the PSU because it will only pull around 3.5A at 12V but as it's running so far below it's rated voltage it will only pump around 30W-35W of heat, not enough to effectivly cool even a PIII.
I'd go for a pelt rated for a lower Vmax, as close to 12V as you can find (the Luefken 72W would be a good choice).

 

[It_The_Cow]

What if I were to switch around the wires to get 17 volts on one line

 

[LimeyGreg]

If you mean the -5v and +12v trick used for overclocking the fans then I'm afraid not, your negative rails are around 2 amps, your psu will shut down for sure if you try that.

 

[It_The_Cow]

Hmm. Thanks for all your help LimeyGreg