Solid state cooling

I read with interest mention of Dendrobates solid state chiller. It added to some thinking that I'd been doing and prompted some further exploring.

(Edit just discovered I can't post links for a while as a new member so I'll take out the links for now).

Hoping that I can prompt some pooling of ideas as there seem to be a number of members with a technical bent. My original chiller (not Haelia) died a few years ago without me noticing for a couple of days and I lost most of my coral so relying on one chiller is a nagging concern, likewise concerns about the impacts of an extended blackout if that should happen.

I'd been considering the use of Peltiers as a backup chiller but keep wondering if I'm missing something basic.

Not sure I'd initially want to use an ebay sourced home built chiller as a primary chiller for a moderate sized reef tank but the idea is very tempting as a backup strategy.

It also has the advantage that in conjunction with a decent deep cycle battery (and maybe some charging off a car) the thing could be a coral saver during an extended blackout. If the circuity gets a bit more complex then the units should be able to double as a heater as well, needs a H-bridge driver circuit for the peltiers. I'm still getting my head around the issues but thought it might be worth seeing what others think.

Queries
- How well does the rating of a peltier compare to the rating of a chiller? Eg I have a Hailea 1/4hp chiller which will in theory handle up to 800L (ignoring local climate etc). That seems to be about 186.5 watt so what size peltier would it take to achieve the same amount of cooling (assuming that heat transfer between water and peltier was not an issue). My assumption is that the ratings should be roughly similar.
A Hailea HC-250A at 1/6hp (around 114W) is nominally rated for up to a 600L tank.

I've found a couple of sites that helped me do some basic calcs on heat transfer. Online physicsclassroom lists a formula Rate = k•A•(T1 - T2)/d
- Rate of heat transfer in Joules/Sec (Watt)
- k coefficient of heat transfer (for glass 1.05 from Roymech in the UK)
- A Surface area(m2) - 40mm * 40mm (a lot of peltiers) = 0.0016 m2
- T1 -T2 Inside temp minus outside temp/ temp diff for a 136w peltier at full power is 67 degrees C
- d - thickness in m - I've not measured but say 7mm (0.007)

So around 16 watts of heat transfer at full power across 7mm glass for a nominal 40mm by 40mm peltier. At first glance not so good.

What happens if I stick the peltier onto a thickish plate of aluminium with a larger surface area (say 120*120*5mm) which is then attached to the glass with heat transfer paste (or similar)? Aluminium is listed with a k of 237. I can potentially transfer 144.72 watts through the glass to the aluminium and as far as I can tell the transfer across the aluminium would allow something like 95 watts per side eg 381 watts total or better as most of the heat is not coming from the edges. The exposed aluminium would need to be well insulated.

Not sure how much actual temp difference I want across my glass so the maximum temp difference maybe somewhat less than 67 degrees C.

From there
- a pair of 136 watt peltier devices on ebay can be obtained for $15.15
- 15v 150w power supplies starting at $28.50 each.
- Still exploring suitable heatsink/fan combo's as the high power rated CPU ones I keep finding are pricier than I'd like (say $65 in a 150/160watt rating) and I've not explored how hard it would be to fit them to a peltier. I suspect that far better options will be available
- Add in a Arduino (or picaxe/Raspberry pi/etc), a temp sensor of some kind (eg a DS18B20 encased on epoxy maybe), some mosfets to control the peltiers, a case and some bits and pieces and a 136 watt chiller could be built for around $150 (add another $25 for a serial enabled LCD display if you like). I've not priced the aluminium but in 5mm thickness not much.

If far cheaper options were found for the heatsink/fan combo then that helps the overall pricing a lot.

It occurs to me that it may be cheaper and more reliable to use smaller peltier units and more of them as that reduces the required rating on the heatsink/fan combo.

Thoughts?

Bob

Hi Bob

The theory is sound, though the many variables cause me concern, from the heat transfer and dissipation to the true efficiency of such a cooler- and yer, the heat sink will keep you warm on those cold nights in the UK! with the said theoretical design, though that said, it would be cheap - and perhaps a good back up as you say.

As far as a rating comparison goes - sorry mate I have no idea myself, though other more technical here may have some experience or comment?

What size tank do you have?

Ignobolis, I'm in Thorneside, close to a couple of other posters I think.
Much warmer than the UK so no need for exta heat in the house at the moment. Enough coming out of the chiller as it is.

The tank is nominally 500 l.

Cheers
Bob

Hey Bob,the old chiller,was it the eclectics-controller that went on it.

If so I am half way through a thread on DIY repairing this your self.

Next week I should have the pics and explanation and suppliers for the parts finished so others can just copy and fix their own chiller.

If it is the electrics on yours, which is what 99 percent of the time is the problem.

there are a few common faults with those thermostats. relays, probes and switching transistors. all easily sourced and replaced.
as long as you know who to talk to.
its the mechanical side of the chillers that are the biggest problem.
a chiller made from semi conductors.... you sir have my attention.

liquidg, it was the electronics. It sat around here for a long time while I pondered options then went to the tip. Love that you are documenting repair processes. One of the wonders of the internet is how much easier it has become to learn from others.

Makaira, from what I've seen of it the efficiency of the solid state chillers does not generally come anywhere near that of compressor driven units. I don't think that they would be a good option for a larger tank (but I hope to be proven wrong). I do suspect they might be a coral saver as a backup to mechanical systems (and I'm wondering about viability as a quieter alternative at night or during the cooler months of the year to save having that compressor running in the house). Cheap(ish) and able to be run from 12v are both appealing attributes.

Probably perfect for small tanks (and I've started wondering about setting up a pico reef now after seeing Dendrobates write up then seeing some other stuff on the web).

Some more things I've learned since my earlier posts (I may repeat some stuff here)
- When they are off the heat from the hot side bleeds back into the cold side so short on off duty cycles can be very inefficient.
- Pulse Width Modulation without filtering has the same problem, during the off cycle heat from the heatsink is warming the cold side.
- Any frequent hearing/cooling cycle imposes mechanical stress on the units that can shorten their lives. An additional problem for both the above scenario's.
eg a lot of the stuff those of us who have used to manage resistive heating is not suitable for this.
- Some claims that they are much more efficient at way lower than the rated power. Eg 2 units running at half power might do more cooling than 1 at full power. I've not tested this yet but it makes some sense. That from a discussion group working on using them for improved camping fridges etc.
- The supply does not have to be extremely clean but should have enough ripple removed to not create a heating/cooling cycle during use.
- Best to run them continuously at a voltage that produces the required cooling. That can be achieved via PWM feeding into a filter circuit.

So the emergency cooling system for the aquarium might be based around 2 or 4 peltiers hooked up in series and sandwiched between an aluminium plate bonded to the tank and a high power heatsink/fan unit.

Bob

My neighbour set up the massive one for the reaserch center on straddie and he is showing me as i take pics then i will do mine and then put it all on here for others to hopefully learn from.

Saves throwing away a perfectly good chiller!

pwm through a filtered circuit?.

if what your saying is that the frequent switching (on\off) of the peltiers causes them to "reverse cycle"
and loose effeicency then would it not be better to run a constant variable dc voltage via a thermostat.

you could still use pwm to drive it but use a rectifier circuit to make it constant. (i think thats maybe what you ment by filtering)

led driver with variable pwm input could work just have to find one that uses dc source. ill talk to my suppliers.
whats the forward bias voltage drop on those peltiers?

Makaira, as far as I can tell there is no forward bias voltage as such. They are not a true diode. I don't think I can post links yet but the Peltier FAQ on the tellurex website has the following "Without a depletion region, a TE couple cannot act like a diode; the couple will conduct in both electrical polarities and there is no fixed voltage drop across the couple (unlike the nominal 0.6 to 0.7 VDC typically dropped across a forward-biased silicon diode)."

Some basic testing I've done with an old car cooler peltier hooked to a low power variable power supply shows current flow from very low voltages. At 0.26V I have what I think is around 30mA flowing (need to dig up some extra leads to do better monitoring of voltage and current at the same time). The cool side of the peltier is noticeably cooler than ambient at that setting by the way.

And I am wanting to smooth the output to produce a variable regulated voltage output, preferably controlled via a micro of some sort - in my case probably an arduino. I'm think a PWM drive into a filter/regulator to take out the ripple and some monitoring on the resultant output voltage.

I set up a dallas 1-wire temp sensor (18B20) with araldite and heatshrink to seal the connections to the unit and connected it to an arduino running a hack of some code from another project so hopefully I now have a temperature probe suitable for immersion in salt water. The lcd display I was trying to use did not work so I'll have to try another later. Managed to get output back on the PC via a serial monitor.

The next step may be to find a suitable small glass container with at least one flat side (perhaps a fighting fish tank), attach a peltier and insulate the thing well then do some experiments with cooling rates. With a known volume of water it should be possible to get some decent measures on how long it takes to lower the temperature by some a given amount at different input voltages.

I may eventually find there is some key stumbling block to all this but so far nothing is giving me that impression.

Bob

dude my company actually manufacturers salt water nmea (serial) temp probes. lol.
id love to get a data sheet of one of these peltiers. bit hard for me to comment further without one.
a variable regulated voltage driven via pwm. im sure such a beast exists ill keep my eyes open.

My peltiers arrived today. Not sure when I will get to working with them next. I have not chased down my power supply or fans yet and am having a think about approach. Thinking of mounting the peltiers onto a sheet of titanium with water pumped through into a cavity in perspex or similar which the sheet forms one side of. Should be far better heat transfer that way than trying to attach to the glass. Bob

Cool Bob

Hey if you can please photo document the build and findings - I'm sure a few others here will be very interested in how your build goes and what the findings are....I for one are keen

Thanks mate

Good suggestion.
I'm thinking I could experiment with my revised design with some aluminium sheet and if it behaves as expected order some titanium.

Bob