Using LEDs as photovoltaics

Have you considered the solar cells used to power calculators? Would probably take a couple in series to get 10 volts.

Dan

You are probably better off with a 3v silicon PV panel like the ones on pocket calculators or garden lights at the receiving end. There was a thread about this recently but with a higher current requirement.

Finding a way to drive it with a much smaller voltage might help.

They say short wavelength UV have highest forward voltage. I believe that's what will limit the output voltage. also they'll pull down the gate ain when unpowered, so maybe a si blocking diode might improve things(rev I in the uA range). Esp if you plan to pulse the drive led above limit.

Is there some reason you can't use a photodiode as current generator? (I guess you'd need more in series) You might google Forest Mimms III who has done some led on led stuff.

In general I think you'll want the drive LED to be of a shorter wavelength than the detector led. And beware of colored lens/ plastic.

George H.

Not quite sure what you mean by "need a bit more distance" here, is it because the clearance between the package pins isn't enough on the photoisolator, or because you can't place the isolator physically close enough to the MOSFET for it to drive it effectively?

I think I've seen this done by taking an phototransistor like the 4N28 that brings the base terminal out to a pin, connecting the transistor as a diode, and stacking them in series, at 10 volts probably good for couple hundred uA with the IR diode string running full tilt.

But it sounds like maybe you want to shoot power across a gap of a couple inches, or something.

Just break apart a garden light. those element are meant to charge a 1,2 volt battery, and can easily switch on any transistor(0.6-0.9volt). I used it to switch a tr. to drive a relay for a nightlight.

That seems right - my 'white' LED does nothing with either red or green laser pen at point blank, but a small UV torch seems to give more output than using a very bright white LED lamp as a source.

Cheers

Small power makes it sound easy, but 10V is rather hard; visible light photons aren't so high energy, it mayl take a dozen receiver cells in series to hit that target.

Receiver area is another concern; emitters are SMALL, most light aimed at 'em will miss. The lenses help (small LEDs with narrow output for emitters, large LEDs with wide output might be the best receivers). A silicon solar panel has the receive-area advantage, it'll catch a lot more photons.

Lasers, even UV types, exist; you could use a laser to hit a small target, but it still takes several receivers in series. Enough lasers to hit 10V worth of point receivers is going to be pricey. Maybe a cluster of receivers around a phosphor dot, and laser-illuminate the dot?

EL backlight material and a PV cell might work over an inch or two

Well, my 'white' LED gives 2.2V open circuit and 10 or 20uA short circuit (it's the last digit on my DVM flickering 1&2) when illuminated with a single UV LED from a torch. I'm guessing a blue LED receiver (ie without the phosphor) would do better.

Forest Mimms III as suggested by George H suggests that LEDs are narrow band receivers, so I'm not sure why the UV LED works so well as a source. It is partly visible with a sort of yellowish tinge.

Anyway, it seems it would work. Needs some experiments.

Cheers

e?

eries to hit

at 'em will miss.

with wide

eceive-area advantage,

et, but it still

oint receivers is going

d laser-illuminate the dot?

I would not expect LEDs to be "narrow band". For a photon to be absorbed i t would need to have sufficient energy. That equates to being short wavele ngth enough which is what George H said.

As others have pointed out the area of the receiving diode is rather small so it will be hard to gather enough energy. The IV curve should be fairly flat on both the current limited side and the voltage limited side with the max power point in the knee of intersection. Try reading the voltage under the load you will need.

Right, The wavelength of light that is absorbed or emitted is a measure of the bandgap in the semi conductor.

George H.

Cree, and several other people, make "high voltage" multi-chip LEDs, in white and blue and other colors, 12 to 48 volts. The 12v white ones that I have look like welding torches.

Blue-to-blue 12 volts might work.

In another thread about the same subject, I had done some test. I could easily generate 2V on a LED, 5 in series is 10V :-)

They are fairly narrow band with a typical FWHM of 50nm around their nominal wavelength. You might get optimum performance with the emitter wavelength slightly shorter than the detector. Possible for some "pure" blues and greens these days.

It is a general rule that emitters at a particular frequency are also good absorbers at that frequency when illuminated.

I like JL's idea of a single compound blue LED with 12v nominal drop.

A visible gap and visible light is comforting when switching high voltage, and makes for a simple 'mechanical' interlock.

A few mm would do.

Cheers

Thanks, I'll look into that. Not literally.

Cheers

Could you use one modulated IR LED (e.g. SFH 4725S), driving one BPW34 photodiode that feeds a small step-up transformer and rectifier? It will not be suitable for turning the output voltage on or off quickly as the smoothing cap after the rectifier will slow things down. You can also have two photodiodes in anti-parallel for push-pull drive to the step-up transformer but you'd have to illuminate them with two LEDs in anti-phase.

Are you suggesting that an LED is not responsive to shorter wavelengths than the wavelength they emit?

I like the idea of using some photocells, but the world is a strange place.