I wanna be a sock-puppet: led question!

[poor mystic]

Well, it goes like this...
I want a very constant light for use in an interferometer. I had been planning to use ultraviolet led's (high resolution is very desirable) and photodiodes.
While I have heard nothing averse about the (detecting) photodiodes, my sometime partner-in-discovery mentioned that led's are subject to flicker. Nano-second scale flicker.
Is this true?
The reason it matters is that I fear that if the led flickers, the phase of the light the led produces might be disturbed.

 

[electronboy]

It is a fact that all LED lights naturally flicker at a rate of approximately 60 Hz per second. Even though that is an extremely fast rate of speed, there are some individuals who can actually detect the flickering at that rate. This is known as a half wave, also called non-rectified light. This type of light is common to various Christmas LED light sets and may be more easily detected if the lights or the individual are moving. Unfortunately, this flickering in non-rectified lights is a common complaint of consumers in regards to some inexpensive Christmas lights.- http://www.ledlights.org/FAQ/Why-Do-LED-Lights-Flicker.html

 

[electronboy]

Fortunately, there are flicker-free LED lights on the market, which are also called rectified lights. They turn on and off (flicker) even more rapidly than non-rectified lights, at a rate of about 120 Hz per second. The flicker of rectified LED lights is virtually undetected by the human eye, and even though they flicker, they are called non-flickering or flicker-free. It would be more accurate to say they are imperceptible flickering lights! Unfortunately, rectified LED lights are less efficient than non-rectified lights, which is still a small price to pay since they are so much more efficient than any other type of light on the market anyway. Another ‘bonus’ of rectified lights is that they are much brighter than the non-rectified lights, by as much as 30% – 40%.- http://www.ledlights.org/FAQ/Why-Do-LED-Lights-Flicker.html

 

[(*steve*)]

Electronboy, the reason those LEDs flicker is because they're driven from rectified (and largely unfiltered) mains.

Run form DC, such a 60Hz (or 120Hz) flicker would be absent.

I have not heard of any flicker from LEDs. You're not talking about the shimmer you see from a laser diode are you?

 

[poor mystic]

I have no idea what the noise my friend mentioned could be; I've looked everywhere I would expect such a thing to be described and i just cain't find nothin'.
Hmmm... laser shimmer? is that due to lensing distortions in the atmosphere?

 

[electronboy]

you could use an incandescant (i think thats how you spell it) bulb but that would suck if you want it to be bright and efficient

 

[poor mystic]

you could use an incandescant (i think thats how you spell it) bulb but that would suck if you want it to be bright and efficient

The technique I am chasing (interferometry) needs all the light to be exactly the same wavelength (colour). Colour is determined by the electrical behaviour of atoms. For any given element in given conditions the colour emitted by the electron structure of its atoms is identical.
I have had some hope that led's might be able to produce single colour (monochromatic) light because they are made of very pure elemental materials.
My question remains moot, although a scan of available devices is not encouraging. I hope to have a 'chat' with someone more knowledgable than myself sometime in the next day or so, which may help me make my decision.
One easy way to get monochromatic light, by the way, is to use a neon lamp. I'd rather find a light source that produces a shorter wavelength than the (red) neon though, I want blue or even ultraviolet light to play with.
At the same time I don't want to use lasers if it can be avoided - lasers are dangerous and I'd rather leave them as a last resort.

 

[daddles]

LEDs don't flicker, at least like electronboy described, if driven with a decent DC source.

However, there may be some types of noise that manifest themselves in some types of use (I'm thinking e.g. of flicker (1/f) noise or shot noise). My intuition would be worried about shot noise. While it's a Poisson process because it's caused by discrete events, I believe the power spectrum is similar to that of Johnson noise. I have an old friend who's an optical engineer and I can ping him on your behalf if needed. You're also going to have to detect the photons and that will lead to shot noise too. If you can determine you have significant Johnson noise in your detector, consider using a thermoelectric device to cool the detector (detector cooling is often done in particle and photon detectors). I can't help much further, as I took my nuclear physics class before those guys landed on the moon...

Since you want to use the LED source in an interferometer, you're going to be integrating the signal over some spatial distance and I'd imagine something like shot noise would somewhat average out since it won't be spatially coherent. This shouldn't be too hard to check with e.g. a cheap red LED.

From the few UV LEDs I've measured, the FWHM is typically around 5-10 nm; the same is true for LEDs in the visible region. I've attached a graph of one spectrum I made about 5 years ago. I had some UV LEDs that had center frequencies of 350 and 365 nm (the mercury line), but I don't know where they are, so I can't measure them. I remember they were in a mailing envelope, so lemme look for them tomorrow. Alas, I think I tossed them out when I moved from the other room...

Edit: on second thought, that statement about the noise averaging out may be kinda dumb. All those noise pulses add up and you won't be able to ignore them. But the spatial part of the comment is probably OK -- it just means that because of the finite size of your detector, you'll probably see fewer spikes; they'll average out to a continuous noise signal. Or, all of this could be highly diluted eyewash...

 

[poor mystic]

Thank you for your kind offer daddles, and for your sensible suggestions.
I'm worried that the led spectrum shows a distribution of wavelengths rather than a single peak. Maybe I need a semiconductor laser. I had hoped to avoid lasers though. Is there some other way of producing truly monochromatic light in a little box at low voltage?

PS Documents found on the web suggest that led's driven by high impedance sources produce less shot noise.

 

[(*steve*)]

"LEDs driven by high impedance sources" is another way of saying "constant current".

 

[daddles]

Is there some other way of producing truly monochromatic light in a little box at low voltage?

Other than the laser, the only thing I can think of is to use a narrow bandpass filter with some light source. But these are usually specially-designed thin film filters and can run a pretty penny. Or, use a slit with a diffraction grating, reflection grating, or prism -- but these are old-school methods.

The last monochromator I used was a B&L device back in school. It had a hefty incandescent source in it (the body of the monochromator would get hot, meaning on the order of 50-100 W was being used for the light source. I believe this monochromator used a slit and a reflection grating to generate monochromatic light. It was nice because a big knob on top adjusted the light color. I don't know what the FWHM was though.

I'm not sure why you're averse to using a laser. You could use it in conjunction with some kind of diffuser and/or absorber to lower the irradiance if desired. The solid state diode lasers have the advantage of being cheap (red) to moderately expensive (green, blue).

I measured the spectrum of a Bosch distance measuring tool's laser with my StellarNet EPP2000 spectroradiometer (see attached plot). The measured FWHM is 3.8 nm; for reference, the sodium doublet is separated by about 0.5 nm (I remember as a student I could see the color difference between the two lines in a Fabry-Perot interferometer). The nominal resolution of the spectrometer is around 0.5 nm, so take that as a data point. I don't have any fancier equipment. A quick look on the web indicates that the FWHM of a diode laser should be in the MHz region, so I'm measuring substantially more.