I'm editing and checking our noise manual. (Making sure everything works they way we have said.) I was looking at the shot noise from a red LED and found that it has a noticeable amount of low frequency noise... starting to rise above the shot noise at frequencies of about
1 kHz and below. A light bulb and IR LED show no such =91excess=92 noise at low frequency. Anyone seen this before?
It may be that I=92m driving the red LED a bit too hard... I=92ll try reducing the current and see if it=92s still there.
Interesting. Is it leaky in reverse bias? If so, it might have had a bunch of junction damage.
How are you driving it? RF power transistors work a lot better if you servo-control the bias current with some reasonable bandwidth--that prevents the device noise from intermodulating with the signal, and cuts
1/f noise in both amplitude and phase fairly dramatically.
Now don't throw the mug at me for bringing this up, it's just meant as an idea: Did you cover the LED during the measurements? Those things also act as photodiodes. When the ambient illumination carries a lot of "gunk" on it and you are measuring miniscule signal levels this might cause issues. That would be a lot less severe in the IR range and light bulbs can't receive at all unless we are talking about an almost direct lightning hit.
I have no idea, 50mA was the number on the spec sheet for absolute maximum current. With warnings that if you ran it over 30mA for a long time bad things might start to happen... reduced light output. Is it internal heating that limits the maximum current through an LED?
Mostly, but big 1/f noise sounds like you were propagating dislocations or maybe interstitial atoms. LEDs always have a bit of 1/f current noise when run from a resistor--since modern ones are pretty efficient, it would be interesting to see how the optical 1/f noise depends on the drive impedance.
This is a high intensity red led from Avago. I don't recall the part number. It may not be a problem, I think we can turn lemon leds into lemonade light sources that have excess noise. White noise is boring after a while. Did the other color leds not have an aluminum problem? I've got other colors from the same series in a drawer at work. Monday,
I'm curious: What kind of red LED is this? Mfr/brand and part no.? Any indication of chemistry type, what sort of efficiency, whatever?
I am aware of a couple interesting red LEDs:
1) Part number ER300 or ER-300, I don't know who made it. This is one of the first GaAlAsP red LEDs, made with GaP substrate. GaAlAsP was a breakthrough in high brightness, high efficiency red LEDs that became available in 1985 or so. Of course, now we have red LEDs ~20 times as efficient as the ER300 was, though the ER300 back then was considered laserlike blinding bright (300 mcd at 20 mA).
One interesting thing about the ER300 is that with reverse bias around 9 volts, in my experience they tend to have faint specks of white glow. I have not seen this in other LEDs yet, not even ones of the same chemistry such as Rohm SLA591LT3 (IIRC).
2) "Low Current Red" LEDs, chemistry GaP doped with ZnO. This dates back to the mid 1970's. One common characteristic is peak wavelength being especially long for a visible red LED - usually stated as 690, 697 or 700 nm. Their spectral bandwidth is more broad than usual for visible red LEDs. Examples are TLR147 (an older part number) and everythying I have ever seen by Panasonic listed in the Digi-Key catalog with a wavelength of 700 nm.
These are not always referred to as "low current red". Some post-1985 LEDs mentioned as "low current red" do not have this chemistry.
These have a nonlinearity where efficiency peaks when current is around or under 1 mA, and decreases quite noticeably as current increases past 3-4 mA. They are only a few times as bright at 20 mA as at 2 mA. Sometimes they look not much over twice as bright visually when current is increased by a factor of 10.
One interesting thing about these: Their spectrum has a secondary band peaking in the green (yellowish green) around 555-560 nm. This secondary spectral feature does not lose efficiency with increasing current the way the primary spectral feature does. As a result, many of these that have untinted packages glow orange at currents that are within their ratings.
In full overcurrent, the diode would develop destructive hot spots. Below that current, there would be thermal profile fluctuations (and you should be able with a video camera to see light output area getting patchy and fluctuating).
Yeah, I did notice that when running 'over current' (50mA) that the DC light output was *slowly* decreasing. Maybe a part per thousand in 5 minutes?? I didn't write down any numbers. It may not be linear... It was running on my lab bench for several hours, while I did other sutff.