Thermoelectric generator TEG powered Sky Ray King

I would like to make a thermoelectric generator to power my 3 xm-l Sky Ray King. The reason I want to make this is because the place where I live suffers from occasional power black-outs so this would be for back-up indoors lighting.
From what I understand the flashlight requires 3.6V and 9A power.
I have been looking at some peltier modules made by EuropeanThermodynamics that are optimized for power generation. Would four of these modules arranged series-parallel in a 2*2 square, and provided sufficient heating and cooling, produce enough power to run the Sky Ray King?
I’m not sure if I am reading the spec sheet correctly as I have little experience with electronics. Provided I could maintain the hot side at ~200’C and the cold side ~30’C, one module would produce 1.7V and 4.4A? Correct?
So 4 arranged 2*2 and wired series-parallel would produce 3.4V and 8.8A. Would that be enough to power the torch?

The thing I have in mind would look something like this but maybe with more candles or an alcohol burner to provide sufficient heating.

Thanks for any advice.

Your trying to drive the SRK directly?

You only have kerosene for a fuel choice? I think you might want to investigate solar or wind generator to run a 12 volt cell charger, like the i4, to charge
1 to 4, 18650cells. The SRK will run off one cell and a long time on low setting.

The Peltier junctions we use at work are expensive and not efficient. Are smaller, less powerful LED lighting more appropriate for emergency lighting or am I missing something?

TEG's are cranky beasts.

First of all, the loaded voltage of those modules isn't 1.7V. It's 0.85V. That's the real voltage you'll have to work with, and that's on a good day under ideal conditions. I would plan for even less.

Secondly, Keeping the Hot Side (Th) at 200C is easy. A little flame and 200C comes quickly. Keeping the Cool Side (Tc) at 30C, however, will be one heck of a challenge. Without active cooling (which you won't have in a power out situation), the only way I can think that you'd have ANY hope of maintaining 30C is with constantly flowing water through a heat sink. Even then, I suspect maintaining 30C is going to be very, very difficult.

The final difficulty is in determining whether the SRK driver will play nicely with the TEG or not. TEG's do not react well to a circuit that attempts to pull more current than they are capable of producing under a given Th-Tc ratio. If the driver attempts to pull more current than the TEG can actually provide, the output voltage will fall to nearly zero and the light will shut off. The cycle will start again, and you'll end up with a really strange oscillation that is hard to explain.

I wish you luck with the project, but I would be prepared for a huge learning curve and lots of trial and error.


@PilotPTK thanks for the reply.
So I should be looking at the ‘matched load output voltage’ value in the specifications of the peltier module?
I had a feeling I wasn’t reading the TEG specifications correctly.

I read somewhere that a typical candle outputs about 90 Watts of heat about the same as an Intel i7 CPU. I have 2 spare tower CPU coolers that I was planning to use on this thing so I thought that might be sufficient to deal with the heat.

@Illuminaria hi thanks for the reply.

Maybe a small diesel generator would be more appropriate. Usually I light up a few candles during a power-cut. Since I already have a SRK and a few CPU coolers laying around I was wondering if i could use the candles to power the SRK.

I don’t know very much about peltier modules so I don’t know how feasible the idea would be. It seems they are much less efficient at generating power than i first thought.


You are correct about the matched output voltage being the one of interest. TEG/Pelts are HORRIBLY inefficient. Even if we take this particular datasheet's numbers to be correct (I suspect they're highly optomistic), then you have to create 75 Watts of Heat on the Hot Side and Remove 75 Watts of Heat on the Cold Side in order to Generate 3.9 watts of output.

3.7/75 = 4.9% Efficiency. And this assumes everything is perfect. Also, if we calculate the REAL energy required (Add 75 on Hot, Remove 75 on Cold, Total 150 Watts of Heating/Cooling required), then efficiency is really 2.45%

It's really not the way to go unless you're just looking for something unique and interesting.


after reading your first post i thought of putting the cold side in a running stream, its about the only thing i can think of off the top of my head that could cool quickly and maintain a somewhat linear level of heat transfer, but under 3% output is not even worth it, better to get some solar panels and charge some li ion batteries in the daytime if away from civilization

looks like a cpu heatsink on a bunsen burner. Have you tried looking into Powerpot? Maybe offload that TeG pad or heck go water cooling PC style.

If lighting is your sole desire and kerosene is what you have at your disposal, you would be better off using a kerosene lantern than using thermoelectrics to run LEDs. And if you want electrical output to charge batteries for other purposes, photovoltaics would be a FAR better choice.

that seems wasting energy. try to use solar panel to charge battery or use generator

you could just buy this, the engineering has been done for you

I have one of these, and it is nothing short of REMARKABLE. Frankly, it’s better engineering than anything you’ll be able to home-brew, though it’s also quite simple in design. No unnecessary complexity. I unscrewed the orange base unit to clean ash from the heatsink, and it’s a very minimalist internal circuit / component layout.

At any rate, I think you get around 200 mA at 5V, from the USB output of this unit. Keep in mind that it is diverting some of its own power generation to a fan that runs the active cooling (which is a necessity in this type of design). When it gets burning, it’s loud, and it’s hot - and that’s still only for 200 mA of output (they say 400 mA, but I question that figure based on my observed charge rate).

So - 200 mA is a far cry from 9 A, and frankly, I don’t think it gets much better than this stove. Though I guess if you paralleled the output of 45 stoves, then you’d have your 9A :wink:

My suggestion? If this is really for emergency situations, why would you waste 9A on a SkyRay King? That’s more light than is necessary. Just drive one efficient LED at 200mA.

I agree with rhd. You likely don’t need thousands of lumens and in a blackout situation where available power is at a premium, you certainly don’t want to be using inefficient over-driven LEDs as your light source.

You can routinely obtain LEDs that offer 130-150 lumens/watt when driven at 350ma. So grab a few of these, pick an efficient 350ma driver (Meanwell LDD series claim 95-97% efficiency), power them from a lead-acid battery and call it done. You’ll have a 400-500 lumens that’ll last for days.

From what I remember, current Peltier chips that we are “allowed’ to buy are rather inefficeint.
Your link shows 375w heat @ 200F = 3.7w, that is under ideal conditions.

If you will, consider building a solar panel,.
If you’re comfortable with wiring DC, it should be pretty easy.

Individual solar cells are pretty cheap, even the broken ones work, just at reduced output relative to the sq surface area remaining.

Grab an inverter from harborfreight,
[ 200w for $20]
i.e. a 200w @ 110v = 1.8amps

Using an old car battery[$0-5], you can have fuel free power to charge batteries or run compact fluorescent lamps. [a 60w CF lamp pulls 13w]

FYI, I’ve used various inverters for decades and really find them to be useful and relliabe.


After seeing one at a friend’s house I kind of shamed him into getting me one (after all I’ve made him two or three lights already).

I haven’t had much time to play with it yet but did find it provides roughly 2.75V to the LEDs about two minutes after lighting the candle. When I have the time I’ll measure volts and amps when it’s fully warmed up. I’m also curious to see if adding a small fan to the back increases the output.

Seriously. OP is building rube goldberg contraption. Hydrocarbon~~EM>TECLED~~>EM

I have made experiments with powering LEDs with Peltier elements and it was disappointing…
So the answer is noway this is gonna power something bigger than some 5mm LEDs…

i wish those 5mm leds were sucked up by some black hole, i despise them