1. Home
  2. Electronics Design
  3. Excess low frequency noise in LEDs

Excess low frequency noise in LEDs

Hi all, This is a bit of a shot in the dark (or a shot in the noise.) So we sell an experiment to measure noise. One of the bits is a light bulb that shines on a photodiode, and you measure the shot noise, which gives you the charge of the electron. (fun stuff)

In the kit we also include other light sources, IR and visible LED. (I've not looked very closely at the noise from these.) I got an email from a masters student in Austria, who is seeing excess (maybe 1/f) noise at low frequency from the visible LED. (And also some from the IR led.. though less.)

There are issues with vibrations getting into system and shaking the alignment a bit. But I warned him about that and he claims to have it under control. (Hmm thinking out loud, the beam spread on the vis. led may be tighter... and so it will be more subject to pointing error.. shaking.) At the moment I'm not sure if it's a technical error, or if it's a real effect. There is 1/f noise in almost all things electronic and finding some in an LED would not surprise me. I see some reports (on line) about excess low frequency noise in LED's (mostly Oleds and GaN.)

Anyone run into this? I'll have to repeat his measurements later this week. (At the moment I'm leaning towards technical source.)

George H.

Reply to
George Herold
Loading thread data ...

Am 01.02.2016 um 22:37 schrieb George Herold:

I'm not really sure if I have shown that here:

<
formatting link
>

These are forward biased LEDs abused as voltage references, and the electrical noise can be pretty small. I did not measure optical noise.

The LT3042 and the Zeners are also interesting.

regards, Gerhard

Reply to
Gerhard Hoffmann

Yup. LEDs, especially if driven too hard, exhibit 1/f conductivity fluctuations, which show up in the optical power output. Driving them a bit more gently, using a good-quality current source, makes a big difference.

Lots of 1/f noise is often a sign of a defective LED, either from some manufacturing defect or from ESD, and is strongly correlated with short service life.

Good LEDs driven too hard can have a lot of 1/f noise too, which seems to arise from electromigration--i.e. it's the noise of the device getting progressively disassembled by the stress.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net
Reply to
Phil Hobbs

Thanks Phil, I wanted to say that I get to see all the data. I've also seen the same effect at conferences/work shops, when I get excited, I talk loudly... and that gets right into the "shot noise"... :^)

I've got to show the student my poor man's spectrum analyzer. A DSO, normal trigger, set up near the top of the signal/noise. Average to max. and pick the right time base... hmmm for room vibrations, you may want less than the max number of signal averages. (?)

George H.

Reply to
George Herold

Gerhard (and Phil.) Thanks... Driving home, I was thinking this is most likely a lot of light and technical.. vibration noise. I've a plastic tube,with two plastic screws threaded near each end. Into each end you plug a photo diode and light source. Three sources, a little incandescent bulb, an IR led, and a visible one. (Sources are driven from a clean voltage source, 1-2 nV/rtHz. and a resistor for the leds.) I saw a plot of excess noise vs device and visible had most, then IR led, then light bulb. But the visible led has a sharp focus... the IR less... I was going to suggest he plot noise (density) vs drive level.

Reply to
George Herold

Could there be some contribution from stimulated emission? Unstable lasing would have a similar signature, correlated photons in bursts...

Reply to
whit3rd

On a sunny day (Mon, 1 Feb 2016 13:37:28 -0800 (PST)) it happened George Herold wrote in :

Now I am curious, how do you calculate that? Expensive kit? Circuit diagram ;)?

Reply to
Jan Panteltje

What is low frequency in this case ? Could it be thermal fluctuations of the LED ?

--
mikko
Reply to
Mikko OH2HVJ

If it is room vibrations adding a thin layer of diffuser to the light path should largely eliminate it. If it is source flicker then it won't.

For educational use and teaching purposes provided you don't want to go above audio Daqarta (sp?) is very good for that and you get a 30 days free trial fully functioning and a workable signal generator afterwards if you don't mind connecting your PC sound card to kit on a bench.

I have only ever used it with microphones. But it will do realtime audio spectral analysis very nicely indeed. Perfect for teaching and live lectures in combination with a projector.

--
Regards, 
Martin Brown
Reply to
Martin Brown

Yeah we sell it for too much money... But you can do it yourself for a few bucks. You'll need a tin can to keep all the local E-fields out. A photodiode and an opamp TIA or even just a follower with a resistor to ground.. With the PD biased (from a clean voltage source... cap multiplier.) and then a bunch more opamps... I gain things up to the volt level, but you don't have to go that high.. (And less gain mean's less chance that you make an oscillator... :^) Total gain depends on the BW, but ~10^4 to 10^6.

The theory is simple. The shot noise i_n^2 = 2*e*I_dc*BW. Where e is the electron charge, I_dc is the dc current.. you have to measure that, and BW is the bandwidth over which you measure the noise... getting the BW right is the hard part.. but for (say) 10% you can wing it and get close enough.

You could get i_n^2 from digitizing and a computer... or just a DSO 'scope type measurement.. or spectrum analyzer... or maybe just a precision rectifier. But what I do is run it into an analog multiplier and use that output.

(Do you need more?)

George H.

Reply to
George Herold

Hmm I could try that too... I always want more light! (it's a little PD to keep capacitance down and speed up.) And I butt the light source right up next to it... I could try a diffuser over the PD.. (scotch tape.)

Reply to
George Herold

Hi Mikko, Well the data the student sent me is plotted in a weird way, but it looks to be a pretty high corner frequency if it's 1/f noise... a few kHz. Which is yet another reason for me to think "technical noise".

George H.

Reply to
George Herold

Nope, it's conductivity fluctuations due to dopant migration. It mostly goes away when you use a current source. Stimulated emission is actually very quiet--it only has shot noise.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net
Reply to
Phil Hobbs

If you use cascaded RC lowpasses, you can do a lot better than that--1% or so at frequencies below a megahertz or two. I do it all the time for noise measurements.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net
Reply to
Phil Hobbs

1/f noise in LEDs is a well known effect, and it's frequently used as a diagnostic to weed out early failures.

See for example

and its references 5-9 (which I don't have in soft copy).

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net
Reply to
Phil Hobbs

I should add that part of the noise is also propagation of dislocations ("dark lines") into the emitting region, which is another sort of migration.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net
Reply to
Phil Hobbs

Arc sources (high pressure Xe for example) are pretty unstable as well. There the problem is a moving hot spot. Total light output doesn't seem to vary much but an image of the source does. Improving stability is easy - put a magnetic stirrer near the lamp.

Field emission sources for electron microscopes have a similar problem - gas molecules contaminate the tip and must be periodically removed through a 'flash' procedure or some form of electronic compensation must be used. Ultrahigh vacuum helps too.

--
Grizzly H.
Reply to
mixed nuts

OK thanks for all the papers Phil. I'm just going to have to do the measurement.

George H.

Reply to
George Herold

OK, I didn't want to get into all the ENBW stuff... How many do you cascade? (I assume with opamp buffers in between) There's an error associated with how much BW you really have above the RC corner frequency. With the two pole filter I use you need almost a decade of BW above that to get 1%.. or you've got to roll that into the calculation.

George H.

Reply to
George Herold

Arc lamps are orders of magnitude noisier than LEDs, due to magnetic effects (e.g. the rapidly moving hot spot in some lamps), plasma instability, and convection. They're really not in the same class at all.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net
Reply to
Phil Hobbs

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.