LED Voltages

I built a simple LED analyzer a while back and ran some tests on various LEDs at varying currents:

Wikipedia also has a list:

Here's what I'm seeing for some Osram parts

My voltages are a bit lower than yours, maybe because the Osrams are "painfully bright" at currents like 5-10 mA. We run these at 1 mA typically.

One minor problem I see is that the slightest leakage current and in the pitch dark you can see the die glowing gently at ~2uA (sometimes less) with some fixed but surprisingly small voltage across it. Lowest I can recall was something like 1.6v for a modern red LED.

Curiously the voltage drop is below the eV energy of the red photon being emitted so I guess it must be some two step excitation process in the lattice.

Not a problem in ordinary lighting but it is when fully dark adapted.

One interesting question is if LED will emit light when operating in reverse current (like zener). I tried that; indeed it does although not as good as in forward operation.

VLV

Overunity LEDs; MIT had a press release on it some time ago.

The tail of the thermal bell curve has just enough energy to add to the bias, occasionally kicking electrons through. It's a heat engine, I guess where the energy at (band gap) balances with thermal energy.

I forget if the radiance is even sufficient to light a room, assuming the walls were covered in GaAsP. For sure, a bright room will generate photovoltaic current, stopping emission, so it's not going to be, say, sunlight bright. One could put an isolator in front, blocking ambient light from reaching the LED, but the isolator works by absorbing, not reflecting, so it doesn't help much.

Tim

Air conditioners are over-unity too, if you count heat rejection divided by mains power input.

Cheers

Phil Hobbs

I took a really good Avago green LED and some gear up to Truckee, where it's really dark at night, and got up at 2AM when I was dark adapted. With my eye right against the LED, I was able to distinguish on/off at about 800 pA.

As you say, at some point there won't be enough volts to make photons. It would be a cool project to see what the actual photon production threshold is.

A green photon is about 2.3 ev, so one might think that no green light would happen below 2.3 volts, maybe. LEDs run about 100 mv per decade current. Science project!

I've made a red LED flash white.

Cheers

We once had a tube of light emitting EPROMs. They didn't work... Turned out that as the parts got faster transients on the Vprog line became larger and eventually enough to damage the chip. The chip programming guy got through a half a tube before he bothered to tell anyone.

The fault was insufficient decoupling on the Vprog supply easily fixed.

Once, right? Then it became a NED (noise-emitting diode).

It's fun to put enough current through them that they change color due to heating. Or dunk them in LN2 and get a color change in the other direction.

The wavelength stability of phosphide LEDs is much better than nitride ones with bias current.

Cheers

Phil Hobbs

I remember that from uni - red leds turned green IIRC.

df

I came across this many years ago - somebody used a reverse-biased LED to e mit very brief flashes of light to test a constant fraction discriminator l ooking at the output of a photo-detector.

I think the authors were Maier M R and Landis D A in 1974 Nuclear Instrumen ts and Methods 117 pages 245-7, but it might have been Maier M R and Sperr P 1970 ibid 89 pages 13-18. The references comes from a paper I published in 1978, so the stuff isn't exactly fresh in my memory.

Interesting--I wonder if it works with modern devices.

I could imagine that the rising edge would be faster with an avalanche, because you don't have to wait for the carriers to diffuse in from the contacts, and the falling edge would be faster too since you've got a huge electron drift velocity helping separate the carriers. (Unless the effect were really spectacular, you'd usually be better off with a VCSEL or something nowadays, but they cost a lot more than LEDs.)

Cheers

Phil Hobbs

Yes; I always snicker when I see a resistive heater claiming "100% efficiency". I say to myself, "Okay yeah, so why not 300% and save us some bucks?"

It's too bad Carnot efficiency is awful over wide ranges, needing many stages to achieve anything; it'd be neat to have an overunity oven in the kitchen, or an overunity furnace for the foundry -- either one would save a lot of power!

Maybe that's one thing 3D printing will make available: many-staged heat engines. The main barrier seems to be desigining and building all the different parts; if they're all more-or-less the same, scaled accordingly, who cares when it's print time?

By the way, induction heating takes electromagnetic radiation and turns it into heat; surely there must be some way to make a piece of hot metal resonate to give off electromagnetic radiation. (They do that already, it's called black body radiation; so, what's needed is one serious wideband-to-narrowband downconverter to move it from IR to, say, 100kHz. :) )

Tim

Ah yes, the old trick of sawing open a 2N3055 and reverse-biasing the BE junction -- it does indeed work; it's very dim, and gets hot quickly, at

Tim

You can make it resonate, no problems--I used to make IR antennas for a living. The problem is that classical equipartition means that the resonance has an equilibrium excitation of kT/2. :(

Cheers

Phil Hobbs

Finger to head, ...'Scratch Scratch'... How could that work? I've been avalanching LED's.. but I don't see how light comes out. The carriers get thermally excited, accelerated by the reverse bias voltage, and then crash into some other atom and ionize it... but where is there any electron hole recombination in that picture? (Sure, maybe it's just my picture that is fuzzy.)

George H.

There will be some recombination, but it won't be very efficient, so the light won't be very bright. Once in awhile an electron knocked loose by impact ionization will recombine radiatively before being accelerated enough to cause further ionization.

At a rough guess, the efficiency will go down by about the same factor as the fall time, and for the same reasons.

Cheers

Phil Hobbs