Pulse width modulation on Peltiers

[Cool Frank]

Hi,
I was just wondering :rolleyes: what kind of pulse width modulation has people been experimenting with Peltier TECs.

Does anyone have any info on the effects of switching a peltier at a high frequency? I found some makers suggesting to switch the peltier at 1kHz. Can I go at a frequency above 20kHz to be above the hearing range and not cause interference to audio equipment?

On the flipside, what about low frequencies? I saw a Peltier power driver product that was switching it on and off over a period in the range of 6 seconds. What are the advantages/ drawbacks of the various speeds?

Also, how popular is it to just plug a Peltier to the DC voltage and just leave it there?

Frank

[Paxmax]

PCM is probably a pretty ok way to power a TEC.
I would however recommend you to higher frequencies, preferably about or above 50kHz.

Having your switch frequency around 20kHz might produce audiable "noise" or EMC since there are usually some harmonics, they might be in the audiable range.

I have only read about one manufacturer actually endorsing and certifying it's product for PCM powering. Other manufacturers do not recommend it for their units. TECs have a finite numbers of ON and OFF cycles they can withstand. I don't remember how much but I think it was about 50-100k times.

Having a too low frequency will cause electromechanical stress to the TEC, much like the hum and vibrations of a 50/60Hz transformer. That will lead to premature death of the TEC.

If you move frequencies higher up, the short bursts can't induce vibrations and stress on the unit since the inertia of the pellets will stop it.

Cycling power on/off in 6 secs intervals is definetly the worst case scenario in my mind. The TECs are reliable devices, but the thermo-mechanical stress on the internal pellets will be huge with those slow cycles. Keeping the TEC at a constant cool or constantly off is the way to extend the TECs life.

So, do PCM but do it at high frequencies! also, Good Luck! :)

[lclark2074]

What wood be the preppies of switching power on and off? Temp control or affiance.

[Paxmax]

Preppies?? affiance?!?! what?!
Dude... only thing I understood was "Temp control"

wood??? = would?

Eh.. please elaborate.. in easier english for us challenged Swedes.... :D

[doc]

You are much better off sticking with DC to operate TECs. PCM and PWM have down sides only for shorting the life of the TEC. They are also less efficient when operated in this mode.

[Cool Frank]

>What wood be the preppies of switching power on and off? Temp control or affiance.

Temp control. Here is my setup: I am using a microcontroller with two temperature sensors. I have the Peltier "sanwiched" between a big aluminium heat sink and a copper plate. On each, I put a National LM35DT temperature sensor which feeds the microcontroller built-in ADCs. Thw switching on time is computed with the following error control signal: K1(Tcold_meas-Tcold_desired), where I choose Tcold_desired to be say 20C and K1 is large.

I also have a second condition, which is to check if the hot side exceeds the manufacturers specification. In that case I am in "limit mode"an alarm goes off and I disregard the cold side and reduce the on time until the problem is gone and go back to normal mode.

For my power supply, I am using the power supply of a discarded case. I have a few cases so I am using my 486 case. The problem is that if I unplug the motherboard from the power, the power doesn't come on. Anyone solved this problem? I think it has something to do with the minimum current reqirement of the power supply.

[Paxmax]

Oh, Hi doc!
Your input is always welcome!

While I agree with you, doc, about the efficiency thingy, I still wonder about the longelivity.

Why would a TEC die prematurely of PWM?
Even if it is PWM:d really fast?


Cool Frank:

Having the "limit mode" working is smart, but don't depend on it for normal operation. Meaning, don't let your equipment reach that trip point under normal circumstances!

If your MC shuts the power off often it will ruin your TEC.
Turning to power off for a few seconds will make the TEC experience thermo-mechanical stress, which might rupture the solderings inside it.

What doc said is true. The TEC will operate a bit less efficient.
When the voltage are applied the TEC moves heat. during the short "off" periods the heat will try to creep back into the TEC.
When the power comes back on, it has to first chill the TEC down again before pumping heat at maximum level.

All in all, it is better to lower the DC level to regulate the cooling power.

You can at least add a few big beefy capacitors to smooth things out. That will atleast provide the TEC a few more micosecs of DC. Adding caps will ofcoz also lower the total efficiency of the electric circuitry since the caps are not perfect, but it will save your TEC some strain.

You better doublecheck that the PSU will cope with the TEC.

....and I still don't understand what preppies or affiance means...

[L337 M33P]

Switching at high frequencies above the audio region = bad. Not only is it likely to wear out the TEC more, but capacitance comes into play. The capacitance is likely to just short out your TEC at high frequencies.

Using a hysterisis control mechanism is a rather good method of controlling the temperature but it will lead to more thermal stress on the module itself, along with mechanical shear.

I would reccomend using something like voltage feedback rather than duty cycle mode regulation. If you can build a switching power supply that is :)

I don't see why it wouldn't work but it might not last long...

Also to turn your AT PSU on, put a dummy load resistor on the 5v rail and another for the 12v rail. About 1.5Amps on each should do it.

[doc]

The life span of a TEC will be ok using PWM or PCM at higher frequencies, say 20K and up. I have tried this below 20K and TECs sometimes make noise, depending on the geometry of device. The noise they make can be self destructive. If it is making noise something is vibrating or moving that should not be. Kinda like a flyback transformer on a TV set when you get that high pitch whine.

[lclark2074]

As to PS to run without a motherboard you need to put a resistor between 5v-rail to ground.
I used to a pot and I cant find it to find the size.

[Cool Frank]

I still have to try this one out. Is it critical to which 5V to tie to ground? Does it have to be the one on the connector to the motherboard, or is using the 5V from the connector to the drives good enough?
Frank

[Cool Frank]

Exactly what happens when there is no load on the 5V rail? Is it not working because of a safety feature :cool: , or is something bad happening on the power supply? :eek: Does it depend on the power supply maker?

Frank

[L337 M33P]

All the 5v lines come from the same power rail, so it doesn't matter which one you put the dummy load on. The power supply usually doesn't turn on because there is not enough load to give it a proper duty cycle for the switching MOSFET, so it either sits doing nothing or latches in an off state. It doesn't damage the PSU though.

[Paxmax]

L337 M33P: I think that the capacitance in a TEC can be disregarded in this high current situations. The resistance in the device will provide enough barrier against destructive inrush currents.

[L337 M33P]

OK then :) I did some calculations and I agree that it would be a negligible effect for those sorts of frequencies.

[Cool Frank]

I found something interesting :D in another section of overlockers forum.

According to this link:

http://www.geekswithpcs.com/article.php?id=88

to power up a power supply with no motherboard all that is required is to connect the green wire with a black one.

It would also be a good thing to respect the minimum loads given in the power supply.

[lclark2074]

that article is for a ATX power SP! 486 had a AT power SP.

[Borisw37]

how accurate do you need your temps to be?
I built a very simple temp contoller for my TEC using a comparator. It basically has 4 connectors. Voltage A, Voltage B, controlled output and ground. When B>A, current can flow trhough controlled output to ground, when B<A, no current can flow. Then i ammplified the signal with a transistor and hook it up to a relay 12v coil, 120v contacts. For temperature measurments i used a 10Kohm @ 25c thermistor from Radio Shack. Connected to A side of the comparator was a voltage devided (i assume you would know what thta is) with a thermistor in it. Connected to side B is a voltage devider with a variable resistor in it. Turn the knob, set the temperature, and it turns the TEC 0n-off-on-off.. etc..

Schematic is available if anybody wants to see it.

[Cool Frank]

>that article is for a ATX power SP! 486 had a AT power SP.
Yes, I noticed the absence of a green wire on the AT.

[ookabooka]

hi, your in luck. I recently experimented with various power supplies. Get your trusty multi meter and put ground to ground (any black wire usually) then test every other wire on the atx/at connector. Once you find one that has about 5v in it, short it using a wire (to any ground/black wire). you shouldn't damage the PSU since when its off the only voltage comming through should be for the switch, especially in the older psu's. If you get the wrong one, it should just not turn on, try the others. It worked for my old P166 dell power supply. Good luck.

edit: wire color was gray in my dell, and purple in another old psu.

[wymjym]

Hi,
I have been using these peltier controllers for several years now. No dead peltiers, no condensation, no noise (that I know of).

click on the vpc link (http://www.techcool.com)

http://www.techcool.com/vpc/images/photos/200r.jpg

wj

[Cool Frank]

It looks nice. :cool: Do you know at what frequency it is the ouput switching at?
Frank

[wymjym]

I was told that it operates @ 2000 +/-, but that it actually is a two stage controller where the peltiers are always getting a steady (low) voltage and the pwm is actually added to this steady dc. I don't understand it all but for the past few years I have had no issues at all. (knock on wood)
wj

[Cool Frank]

According to the maker from the link you provided, the voltage is converted from the 12V DC supply to another DC voltage (say 15V). This seconday voltage is next applied to the peltier in the form of pulses according to the temperature. If the Peltier got a variable DC voltage according to the temperature feedback, there would be no need for two stages.

What kind of heat sink did you put on the hot side of the peltier?
Frank

[wymjym]

yeh, whatever...it works.


I made a couple of copper water blocks and actually created a peltier sandwich.

you can see the whole sheeeeeeebang here (http://westech.home.mindspring.com/chiller/mychiller.htm)

wj