Peltier question.

Thanks for that. It's a bit beyond my abilities to make up something like that (also I've got shitloads of higher-ptiority 'projects' waiting for my attention. LOL, the only part of that circuit that I have on hand is a part that I wouldn't need - the 15A bridge rectifier.

I thought that it might be, going by a quoted switching frequency of 65kHz. I mentioned I'm not very knowledgable about this stuff yeah?

Nor am I. I don't have an o'scope - and wouldn't know how to use it if I did (but could probably pick it up, I'm quite good at the old RFM).

Yep, I can see how that might occur. I was intending on supplying the peltier with a grunty heatsinked cc / cv adjustable LED driver so that hopefully I could find a power setting for it that mesant it wouldn't switvch in and out too often - or not often enough. However those LED drivers are switch-mode doodads right? Maybe if I add a bigish capacitor after it might help.

Heh! The damn axolotl's more than 15 years old - I was told they only lived for 12 years maximum.

Cheers,

Actually the LED driver (which will probably be switch-mode) is likely to have a noisy output because it's only meant for driving LEDs, which don't care about such issues. It might be tricky to work out how well your additional filter caps are helping this without an oscilloscope. I guess it will be a case of just hoping it works well enough to do the job.

Yeah, so maybe not my brightest idea.

It might be better to see if I have a ~12v transformer of sufficient size in my junk collection and rectifying and 'filtering' the output (making an unregulated supply). But maybe there will be too much 50Hz ripple on that? I don't know enough (or likely have enough bits on hand) to make a regulated supply.

It's starting to look like running the little fridge at ~10 C is the best option.

Cheers,

It might work, but without a 'scope you'll just need to try it to find out.

The regulator in a linear (transformer) supply is there in one function to avoid the ripple that would be on the output without it. So an unregulated supply would have too much ripple. A 12VAC transformer with output through a rectifier (dropping 1.4V) is actually going to have an output voltage of around 15V, by the way. Vout-dc = (12Vac * sqrt2) - 1.4Vrectifier

If the current boosting transistors on the standard voltage regulator is too hard for you to do, you could go old-fashioned and use Zener diodes in parallel with the peltier. From a 12VAC transformer they would use about 21W power. Five 5W 3V3 Zeners might do the job. A current limiting resistor would also be required from the output of the rectifier/caps.

Sorry I'm out of time to draw you pretty diagrams etc. Maybe tomorrow.

From an old (1976) electronics catalogue I happened to look at yesterday, in a description of power supplies for Peltier coolers:

"The only limitation on the supply is that ripple be maintained at a point lower than 10-15%."

Make that five 5V 12V Zeners.

Really you guys, Peltiers are not that bothered by a little bit of ripple or noise. If you had 5% ripple, maybe you could expect up to 5% worse performance (but probably better than that). Any SMPS that could run a computer etc. will be more than good enough for a peltier.

The idea about not applying a lot of ripple to the Peltiers comes from warning against the practice of some people to use a standard PID temperature controller that applies pulse width modulation to adjust the amount of cooling. That is, a square wave of e.g. 0V low, 12V high, applied to a Peltier. That is HUGE ripple, 12V peak-peak in this case, and it still works almost tolerably well (though it is not recommended and probably causes excessive stress due to thermal expansion and contraction in the peltier). I would not use on/off PWM to control the peltier, but any SMPS that has a filtered DC output with a few tens or hundreds of millivolts of ripple would be perfectly fine to run the peltier continuously. The difficult part would be how to reduce the cooling when full power is not required, as most SMPS do not have a wide adjustment range. Some LED drivers are adjustable, but some of those use on/off PWM, you need to look for ones with adjustable DC output voltage or current. The safe ones won't be cheap and vice versa.

I still like my idea of putting a tiny heating element inside a wide-mouth thermos flask, and putting that in any existing ordinary fridge. You could use a standard off the shelf PID temperature controller and thermocouple to regulate the temperature inside the thermos to any temperature above that of the fridge in which the thermos is located. The heater element won't be bothered by PWM, ripple etc. and will be cheap. Also it avoids the need for a huge heatsink that would be needed on the hot side of any peltier cooler.

most adjustable buck converters will adjust down to 0V if you inject a little current into the voltage feedback divider. on a non-isolating buck converter this can be got from the input side.

The LED driver may well have a far more noisy output than a PC power supply with outputs designed to be suitable for powering logic circuitry (and therefore would be OK for this application).

He wants to use a thermostat type controller to maintain the desired temperature. Not the ideal method of control but perhaps not as bad for the Peltier as PWM.

Yes that would work nicely.

I thought that a thermostat controller does basically do very slow PWM, if it turns the current on and off with no ability to produce intermediate current levels. I think this would not be as good as an adjustable buck converter, and choosing this type of thermostat would certainly make more difference to the peak-peak current variation than whether you use a SMPS or 50Hz transformer with linear regulator before you switch it hard on and off with the thermostat PWM.

It might also be possible to find a LED lighting SMPS that is of the type where the current is always continuous DC with an adjustable value, (not the type that does PWM for dimming that puts out a square wave because the PWM is too low frequency to be filtered out by the output inductor and capacitors). Something like this but less powerful and less expensive might be nice:

The difference is that PWM control introduces a constant switching at a particular frequency, whereas thermostat control switches at a frequency that's orders of magnitude lower. Instead of the length of the "on" pulses being longer for more effect, as with PWM, thermostats vary the length of time between the pulses themselves.

With things like your fridge, these "on" pulses from the thermostat can be minutes apart. I note now that there may be an issue with using a Peltier to cool this way though: Once the power is turned off by the thermostat, the Peltier will begin to conduct the heat from its heatsink back to the surface to be cooled. This will probably result in much shorter "off" times to "on" times.

An automatically adjusting supply using either a suitably filtered switch mode buck converter, or a linear regulator, would indeed be much better suited. However the OP has indicated that the construction of the circuitry required would be beyond his abilities.

The OP's LED power supply does apparantly have some variable current control. The problem with LED drivers is that they do not need to design them for normally acceptable levels of noise on their output, because LEDs don't care about it (hence why some controllers can use PWM). Another type of switch mode or linear supply would be preferable, though with the LED driver the only thing available, it would be worth giving it a try.

Manually adjusting the current to the Peltier may allow the box to be kept in the correct temperature range without even using the thermostat, as the cooling may be set to balance exactly the heat ingress from outside.

Actually two 5W 12V Zeners would be OK if you accept that if you disconnect the Peltier while the power supply's on, they will melt. ______ __________ ___________ Resistor | |----| Bridge |+____ ___________| 25W 0.47R |____ ____ |12VAC | |Rectifier | | Filter |___________| ___|___ | |______|----|__________|_ ----- Cap/s 12V | / \ | | ^^^^^ Zeners /___\ | |___|______________________ _____| | - |__________| + | | | Peltier | |____________|

Assumes Peltier consumes 6A @ 12V (plus or minus about 0.5A).

The 25W resistor may be made up from 10W resistors in parallel. eg. 3x 1.5R 10W

Thanks Kev I've saved all of this for future reference.

Much appreciated.

Thanks for the input Chris.

I'm intending to presevere with the peltier idea because I've already got the polybox, large internal passive heatsink (cold side) and really good all copper ex-CPU heatpipe based cooler for the hot side. I don't have a wide mouthed thermos and getting wires inside the fridge is likely going to compromise the seal.

Thanks to you guys for running through all of this, It's not for nothing, I'm learning heaps.

My original intent was to use a >12v laptop brick for the DC power then put it reduce both current and voltage to the peltier something like this;

Then I was planning to swith the output to the peltier with one of these;

I was planning on setting an upper temp of perhaps 10 deg and lower of 8. I was thinking that I'd set the voltage and current so that, for an average ambient the cycles would be roughly equal on and off.

The problem with that is my home isn't climate controlled. In the hottest part of summer the ambient can cycle between 28 / 18 day / night and in winter more like 18 / 8. With the climate the way it is the upper and lower temperatures can vary by more than 5 degrees from one day to the next.

Do you think that it's worth trying this as planned (I'm still waiting on the higher power peltier) or should I just drag the small fridge in from the shed that I'm storing for a friend and set that to it's highest temp and see how that goes? (Bearing in mind I don't have much space and moving fairly heavy s**te around isn't easy for me. )

Thanks again guys.

Once upon a time on usenet ~misfit~ wrote: [snipped]

(Please excuse typos and nonesense words, I changed the format of how I was saying this and made errors.)

I hoped that, because both of the 'LED drivers' are also able to be used as Li-Ion chargers that there'd be acceptably low ripple (or the cells being charged would suffer surely?). I know that sometimes I expect too much from these cheap Chinese modules.

Also I can't justify the cost of even a cheap scope (which then I'd only end up second-guessing - is it the scope throwing errors?). As it is last year I bit the bullet and bought a Fluke 17B+ as I find that measuring tools need to be trusted - or what's the point of 'measuring'?.

Cheers,

By the way, the thermostat temperature switch would go between the filter cap/s and the 0.47R resistor on the positive line.

Ah right, I was on the wrong track. Try to find a 12VDC power supply and use that without the LED driver, with just the thermostat device for temperature control.

I think it would be easier just to leave it out of the equation - I'm not sure how it would help unless it were rewired to control the temperature itself. I don't think I'd charge Li-Ions with it myself.

I can do that, I have one 12v / 4A brick (But half a dozen 16 to 21v laptop PSUs.)

Ok thanks again for the input. :)