Remote Living > Heating

A look at Peltier cooling

(1/40) > >>

Madscientist267:
I've been playing with these little Peltier modules for a little while now, and have come to realize their potential for use in the land of RE (and real life as well) for use as a source for chilling.

Translation? Refrigeration.

While they have been in use for many years now for just that, there's been a fatal flaw with them that for whatever reason was never addressed.

They are horrendously inefficient in their 'native' design specification, but I have found a cute little combination of hacks that make them potentially viable for use in this arena, even in RE.

I held out forever with this until I could prove to myself that it was viable. First hold up was that the idea of running them on reduced input didn't really cross my mind until recently. The other was time to set aside to play with them. Not that it was difficult to do the experiment, just hadn't gotten around to it.

Last night, I found out that they can be run in a 'maintenance' mode, using between 1/4 and 1/6 the power they are rated for (and even less with the right circumstances), and still keep things cool.

I have two of these modules, one is buried inside it's OEM implement, a new Coleman "44 can" cooler that I got from wally world recently, that can stand upright just like a micro-fridge. The other was removed from a much smaller cooler that I purchased years ago. The barebones assembly is the only thing intact, with the module sandwiched between the two heatsinks.

The Coleman module appears to use quite a bit more power than the other one, and so this was one of my motives for trying this. The power supply that came with the unit gets outrageously hot in operation, and I couldn't see the need for it.

Both modules are designed to run on 12V nominal.

Turns out, reducing the input to approximately 5V results in moderate chilling capability with much less power consumption than it is designed for. It was able to 'hold the cold' overnight, only gaining a couple degrees C over the entire span. On full power, it will hold the temp lower, but the return is nowhere near what it is given to do so.

I'm in the process now of designing a thermostat that will switch between 5, 12, and 15V (the output from the power supply that came with it) in an effort to economize this thing as much as possible.

I'm also looking into using very large heatsinks with multiple modules to work my way closer to the abilities that a carnot cycle fridge can provide, even though I'm aware that the overall efficiency will still be much lower. Efficiency, surprisingly enough, isn't the primary concern here. The lack of moving parts is what makes it so appealing. With a large enough hot-side heatsink, I may be able to even eliminate the fans altogether from the design. More on that at a later time.

The relationship between power consumption vs cooling ability is non-linear:

@ 5V : ~1.8A (~9W)
@ 12V: ~4A (~48W)
@ 15V: ~4.6A (~70W)

The couple of degrees of 'cool' that are lost by running it at 5V is trivial compared to the power savings.

I'm not sure if the module can handle much more than 15V, and at the moment, I don't really care to find out. Power consumption is apparently exponential as the input increases, with only marginal additional heat pumping abilities.

In 'maintenance' mode, the module cannot quickly recover much in terms of heat that has entered the chamber due to the door being opened, etc. So, I am working out a multi-stage thermostat (maybe even just a PWM design) to provide power based on how close the temperature is to the set (desired) temperature.

At the moment, this means 5V when holding, 12V for normal cycling (heat creep recovery), and a switch on the door to engage 15V to prep the heatsinks for maximum recovery ability when the door is open. Considering a timer for this as well, to hold the 15V for a short time after the door has been closed, at which point control would be returned to the thermostat.

Here are some pics:



The Coleman cooler, holding at 7 deg C (~45F). It flirts with 6 degrees C (~43F) regularly, even after sitting all night. This is cold enough to keep many foods from spoiling, at least in any immediate term. Not bad for 9 watts.




The inside, showing the bags of dihydrogen monoxide provided for thermal mass. This makes a big difference.  ;D




The immediate rise in temperature as seen by the probe in the short amount of time the door was opened to take the picture above. The probe is dangling below the tray, not really visible. The meter actually peaked out at 16C (~61F), and had already recovered to 14C (~57F) by the time I could take this pic.




My other Peltier module, with heatsinks. This one uses a little less power by design, but doesn't have quite the cooling capacity either. Good for getting an idea of what to expect from the bigger module in the Coleman, however. They operate on the exact same principle, and so far, hasn't led me astray. The small sensor wedged in the center chunk of aluminum is some sort of temperature limiting device (I'm assuming for automatic defrost - this one had some minor issues with that when it was in it's original implement.


The probe's temperature recovery is good (within a minute or so) with the bags of water and soda bottle as shown. Recovery is very poor without these present.

Figured someone might find this interesting... More to come as the experiments continue...

Steve

EDIT -

Maybe even just as a supplement to a carnot system, a hybrid.

Haven't run the math to get an idea of how this pins up against a comparable sized conventional mini-fridge's average usage. No samples.  :(

May be worth it - a deep sleep that uses almost no juice.

FWIW

dnix71:
There was one brand of Peltier junction cool boxes that works like you tried. It was almost as efficient as a real fridge. The cheap cool boxes have no real controls. This one did.

I can't find it on the web anymore. It was about $200 which is still much cheaper than an Engel.

Madscientist267:
Yeah I'm starting to think there really is something to this, but they get a bad rap because too much is expected from a single module.

I think spreading the power out among multiple modules is the answer here. A few designs have hit my mind's eye since I posted this; Seriously contemplating giving it a shot.

There's a place in china that has 40x40mm modules for like $4 US a pop, bought in lots of 10 modules ($40/lot), but I'm a little leary, something about the site doesn't really quite feel right, even though there's nothing outward that screams 'scam' either...

http://www.aliexpress.com/fm-store/901312/210188600-313416809/10-pcs-of-TEC1-12705-Thermoelectric-Cooler-Peltier-Plate-50W-free-shipping.html

What you guys think? $40 for 10 modules, but there are single modules for that much elsewhere on the webernets.

Traces of the PC repair geek in me always scrutinizes a little more than the average Joe, I suppose. But if it's real, I can think of a design I wouldn't mind trying using all 10 of a lot on one unit. Hope I can find as good a deal on extruded heatsinks too.  ;)

I mean either way, flop or scam - I'd be out only $40, so it's very tempting even if just for giggles to see what the possibilities are.

Steve

joestue:
I think i paid $1.76 plus shipping for a 40mm square unit with the same part number as that lot you linked to just recently.
it works as good as the other ones.

The key here is turning the junction off when its cold enough.
The amount of heat transfered is pretty much linear. zero at 55-65C difference, maximum at zero C.
so its pretty easy to calculate the max thermal efficiency if you know how much heat you need to remove on average.

I'm not sure, but it might be more efficient to run one at 40 watts input with heat pipes rather than two with just heat sinks and fans.
the kind of fridges i've seen have two junctions, and heatsinks inside the fridge with two 0.3 watt fans inside the fridge.

if you were to use passive freon filled heatsinks rather than fans it would probably work out to be on par with normal refridgerators.

another issue is that when you turn it off, the heat in the heat sink is back fed into the fridge, at half the maximum rate you can remove it
(assuming a 40F fridge and the peltier is 35F at the cold plate, 80f ambient and heat sink temperature of 95F during operation.)
it isn't that much more efficient to pwm the peltier either.
one option is to use a water based cooling system.
that way when you turn the junction off, there is no heat back fed into the fridge.

Madscientist267:
I'm kinda there, only not. The multiple module design may draw more at some points than the single module would, but I think it would be much more efficient about it in the process.

Thinking something along the lines of an 'active insulation' kind of thing, with 1 module per ~sq ft of heatsink 'wall'.

With all 6 sides covered, the heatsinks can be thermostatically kept at 40F (or whatever) and the space inside will intrinsically meet up with this value at some point.

This is in contrast to the current designs; a single module, being force-fed full power, and struggling to keep up with just what is leaking through the walls.

A multiple module design would also be able to provide serious cooling power when needed, such as for initial cool-down and extended door-open induced temp spikes.

I also need a traditional carnot unit to compare performance to. Cost is already roughly comparable for identical size base units, about $80 US new for either kind, but the performance of the Peltiers lags severely, and I think it's simply due to design.

Edit:

9 watts flowing maintaining 7degC overnight? In terms of wH, I'd be willing to bet that the carnot system uses more than that when the on/off cycle is averaged over the same period of time.

Obviously, though, there is no contest when they are compared with the Peltier running at full throttle. Carnot wins, hands down.

This is where that whole PWM/Active Insulation idea comes in... if the duty cycle is continuously variable, it will only use as much energy as it needs to based on how much it needs to change the temperature at any given moment.

With 10 of the same Peltier module working in tandem, I'd be willing to bet that the power needed by each module drops even more significantly yet. They start to show a noticeable differential when being fed as little as 2.5V!

Steve

Navigation

[0] Message Index

[#] Next page