PWM and Peltier Junctions?

andrewpro

New Member
Not entirely picaxe related, but the PWM would come from one so...

Has anyone, or does anyone just plain know, used a PWM signal to control the temperature of a peltier junction?

My radio telescope DC processor (If I ever finish it! It's in redesign number 7) needs temperature control. The cooler the better, but I can't afford liquid nitrogen or anything like that. What I'm going for here is more of stability than any increase in performance. I dont need fast dramatic temp changes, just the ability ot keep the temps of a few components at a reasonably constant level, and ovenizing is out of the question.

--Andy P
 

hippy

Technical Support
Staff member
Google would seem to be your friend here. There are plenty of articles saying PWM is ideal to control the Peltier, and if a PICmicro can do it then I'd imagine that so can a PICAXE with PWMOUT.
 

nbw

Senior Member
WOuld be very interested to know how you get on. I had an idea about using a Peltier to cool a fish tank during summer, controlled by a PICAXE and temp sensor.
 

ylp88

Senior Member
I've never used a Peltier device before but this is a quote from Oatleye Electronics website (http://www.oatleyelectronics.com/peltier.html):

<i>Switching on and off a Peltier below 2KHz frequency causes increased thermal stress on the device </i>

Not sure &quot;exactly&quot; what this means, but it appears that you shouldn't switch the device on and off very fast?

ylp88
 

hax

New Member
No, you should not switch a peltier on and off very *slowly*

The thermal shock of cooling down and heating up causes the peltier to break prematurely. If you use high frequency (relatively) then it should be OK.

Also it is important to turn them on and off with PWM gradually, say over 40 seconds or so, for maximum life.

A frilly fact about peltiers is that they actually produce electricity if you heat one side, and keep the other side cool.
 

bobrayner

Member
Don't forget a Peltier device transfers heat from one side of the device to the other depending on current flow direction.
nbw. with your fish tank application the device would have to be attached to the glass with good thermal contact and not simply immersed in the water.
cheers BobR
 

andrewpro

New Member
Yeah, I should have tried more google searchng. Most of what I got the first time were CPU cooling &quot;hacks&quot; for overclockers...really not what I was looking for.

When I put it thermoelectric PWM control instead of peltier PWM control, I got alot more of the scientific data I was looking for.

I read some conflicting reports on the frequency of the PWM, but appearantly faster is better (not allowing it to heat cycle) so I'll take the 2khz+ suggestion.

I'm interested in your comments haxby. Do you mean to slowly increase the duty cycle over a longer period of time?

And thanks for the links micheal. They explained alot on the control aspect.

I havee used PJ's in both cooling ant heating before, but never for any kind of precision temperatue control. Usually just on or off, and once using current control. Seems PWM is a much easier way to go, especially as it's a feature of the picaxe I plan on using for the control.


Thanks everyone! I'll let you know how things progress

--Andy P
 

BeanieBots

Moderator
I have first hand experience of peltier heat pumps and can confirm that PWM can be used but must be fast. How fast depends on the pump size and how hard you are driving it. Several kHz should be fine for just about any size. One thing to remember is that as the temperature gradient rises, a voltage is also developed which must be overcome. This makes them difficult to control by simply applying a PWM voltage. It is much better (safer for the peltier) to use a PWM current.
Consider using a switching regulator to maintain a constant current and use PWMout from the PICAXE as a voltage source to control that current.
Do you know what sort of power levels you need to pump? If they are not too high, it might be possible to use a linear current control rather than a switcher.

nbw, I too looked at using peltiers to cool an aqaurium. Several problems arose. Firstly, they are not cheap, and the size/quantity to get any real pumping power (~100W) makes alternatives far more attractive. The second issue is how to interface between tank and pump. Through the glass is way too inefficient especially when you consider how expensive your pump will be per watt. The alternative is build a heat exchanger and have the water make contact with the peltier surface. This is very tricky in practice and also requires circulation either from another pump (which adds yet more heat) or from your existing system which slows it down and may well then need beafing up and so the cycle goes on.

Edited by - beaniebots on 10/5/2005 8:02:00 PM
 

nbw

Senior Member
Hi BobR, yes I was thinking about sandwiching the Peltier with heat grase between 2 6mm alumnium plates, including after that a heatsink and maybe CPU fan. The device would sit on the outside of the tank against the glass.

I'm trying to be able to cool a 150 litre body of water at 27degC down 0.1degC say every 8-10 minutes, with ambient temp 25-30deg C. I have a 40x40mm Peltier, it'
s about 40W I think. I know there's a meaty equation around to check all this... no idea where.
 

skyv

New Member
Hi NBW
There's an aricle in this months silicon chip for a &quot;bird brooder&quot; using a picaxe 18x and a peltier module.
Might be worth your while to have a look.

skyv
 
Hello Everyone, I submitted the article for the bird brooder to silicon chip. Its a bit different to what is required here. I used a h bridge because I wanted heating as well as cooling, I din't use PWM as I had concerns regarding the frequency the peltier could handle. Make sure you have a large heatsink with fins on the exterior side of the peltier, with a fan. I found that the effeciency of the device varyed greatly with the size and speed of the fan.
 

andrewpro

New Member
Thanks for the comments beaniebots. I'm not going to be cooling more than a few watts at most. Basically all I want to do is keep the components at as close to thermal equilibrium with each other as possible, and somewhat cooler as cooler=less noise. I wont get any dramatic improvement in sensitivity, just stability.

How does one go about making a PWMed current rather than a voltage? Am I correct to assume you mean using the PWM to control a current regulator at a constant voltage? I'd love some info on that =)

--Andy P
 

nbw

Senior Member
just a quick thought about the Peltier - you could start it up by pulsing it at 2Khz, and then over the next (say) 30 seconds, slow the pulsing until it's just on continuously. And, reverse the process when it's time to turn the Peltier off?

Is this what needs to be done to avoid stressing the modules?

cheers
 

BeanieBots

Moderator
By PWM current, I really mean a constant current that can be controlled by a PWM signal. Instead of taking control feedback from the supply voltage, use a current sense to supply feedback.
A peltier behaves a little bit like an incandescent light bulb. At the point of switch-on, the &quot;resistance&quot; is very much lower than it is after it has been on for a while. It is this initial high current that causes the thermal shock. They need to be brought up slowly.

Another thought. If your power level is not high, have a look at building a constant current source around a LM317. The app notes show how do it using just one resistor. Put this in series with a transistor to turn it on and off via PWM. A better analogy is to compare a peltier to an electric motor. Very high starting current if a voltage is simply &quot;applied&quot; but comes down quickly as it spins up (as a temeperature gradient is developed for the peltier).

Edited by - beaniebots on 10/13/2005 11:41:06 AM
 

Sputnik

New Member
&quot;150 litre body of water at 27degC down 0.1degC say every 8-10 minutes, with ambient temp 25-30deg C.&quot;


Whoah, your gonna need a lot more than 1 40mm pelt to cool that much water down that much. You will be very lucky to get just 1 litre down even a few degrees with that and it will not happen in 8-10 minutes thats for sure.

You would probably need 1 40mm pelt per litre and liquid cooling on the hot side to cool it that much.

You may have to scratch the pelt idea and go with a phase change system. You can get a small window air conditioner very cheap, you may even be able to find a scrap one for free if you look around. You could make a container to immerse the condenser in to cool the water and pump it to wherever you need it.
 

nbw

Senior Member
Yes, I thought I'd need a few more than one. I thought it might be useful to set up a minitank - about 20 litres - in the same conditions and see how much cooling I can get. I'd be running the peltier off a PC supply (200W) so feeding the peltier's hunger wouldn't be a problem.

I'd be happy with a small bank of up to 10 peltiers in the real system, even if it took a little longer to cool.

I'd be interested to know if anyone has any experience / ideas on whether cooling glass quite sharply in one area while the surrounding is warm would stress it (the glass) at all?

thanks for the feedback thus far.
 

andrewpro

New Member
It depends on the glass. Normal, non tempered glass, your looking doom in th face. Heck, even with tempered glass it can happen. coolinga small section of a larger section of heated glass only 10 degrees in a very rapid manner will cause all kinds of evil crystal shearing thigns to happen!

--Andy P

Edited by - andypro on 11/7/2005 11:14:37 AM
 

TEZARM

Senior Member
My radio telescope DC processor (If I ever finish it!

andypro. I for one would love to know what this is and how it works?
 

premelec

Senior Member
Doesn't sound like a good idea on the other hand I've occasionally resorted to a propane torch to defrost an iced car windscreen - and by mistake hit the flame on glass instead of ice... hasn't ever broken :)

What's a few hundred gallons of aquarium water anyhow? An incentive to build a PICAXE flood detector! [and catchment and sump pump control].
 

nbw

Senior Member
Good points!! I could get the sump of the new tank made in tempered glass. I'll make a small prototype first of course. thanks for the advice and pointers all.
 

andrewpro

New Member
TEZARM wrote:

&gt;My radio telescope DC processor (If I ever &gt;finish it!

&gt;andypro. I for one would love to know what &gt;this is and how it works?

I writes:

Well...a radio telescope is a telescope that, instead of looking at optical light, looks at radio waves comming from space. Different things in space emit different strengths and frequencies, and you can learn alot about a celestial object (or the lack of one) by those radio emissions.

The DC processor is what takes the radio waves from the receiver and antenna and turns them into a signal that a computer or chart recorder can read. It involves a special kind of rectification and alot of am-p0lification, prior to feeding it into an ADC to be sent to the computer for analysis.

You can check out more at

HTTP://radioastro.plistin.com
<A href='http://radioastro.plistin.com/ ' Target=_Blank>External Web Link</a>

It's not much now as I'v ebeen busy working on other things, but it'll give you an idea what radio astronomy is about (briefly) and the hardware involved (basically).

--Andy P
 

Cheeky

New Member
One of the problems with these devices in applications such as processor (or radio telescope) cooling is the complexity of either placing components requiring cooling inside evacuate sealed chambers OR controlling the device so that its surface temp never goes below the dewpoint of the air around it. These devices are usually rated with a delta T, that is the achievable temperature difference, this figure is typically 40 to 65 so the surface temp can be easily below the ambient dew point and probably below zero so dew would then freeze and accumulate on the device until it turned off and defrosted, probably not good in many applications.

Another problem with them is the very close proximity of the hot and cold surfaces, typically 4 to 5 mm, making the use of the cooling surface fairly tedious.

As for the use as an aquarium cooler, probably not really suitable. Based on your figures of 150 L, 0.1 C and 10 minutes you would require 103 W of cooling (not input) remembering that this is to cool the given quantity of water by 0.1 in 10 minutes without allowing for any additional heat from ambient gain, lighting or pumps so you would probably require several peltiers before you allowed for other gains and these would need to be removing heat more directly than thru glass, probably better to use the peltiers chill coolant or distilled water in a seperate insulated tank to create some stored capacity and then pump that fluid on thermostat demand through a heat exchanger coil in the tank (or filter etc), this could be stainless although titanium is best, or experiment with much longer lengths of polyethylene or nylon tubing.

I have built around a dozen refrigerated marine tank coolers ranging in size from 500 to 1400 W, if you want some more info let me know.

Regards,

Cheeky

Edited by - Cheeky on 11/15/2005 11:25:14 AM

Edited by - Cheeky on 11/15/2005 11:26:09 AM
 
Top