Some noise measurements for LEDs and zener diodes

There has been many discussions on the forum over the years
about how noisy various voltage references are. I can't really
remember anyone pointing to any measurements or satisfactory
theoretical arguments, but maybe it is my memory being bad.
It seems to be generally agreed (maybe) that LEDs are less
noisy then zener diodes, but I don't think anybody has tried
to put figures on the difference. There has also been discussions
about what LEDs are lesat noisy. The only claim I can remember
reading on the forum is that green LEDs are the least noisy,
but no argument was provided to back it up. Hence, I decided
to see it it was possible to get some answers to these questions
using only some simple DIY test equipment. I don't have the
time or money to do a proper scientific study, but maybe this
can be considered a somewhat systematic pilot study for
further investigations into this matter. For now I have done
some measurements on various types of LEDs and zener
diodes and some of the results were quite unexpected and
begging for further studies.

i didn't do this just because I had nothing better to do, because
I had, but since I like the forum and it members so much I
decided to put some time and money into doing this study
(OK, I admit I also did it to still my scientific curiosity :) ).

The study and the measurements so far are described in
the attached document. Please note that the study is too
small and uses too questionable equipment to be considered
properly scientific, so no hard claims or conclusions can be
drawn from it. There are hints at some interesting phenomea
though. Had I know beforehand what the results would be,
I would have used a slightly different selection of components
for study, but thats usually how it is.
 

Attachments

  • noise_measurements1.txt
    6.9 KB · Views: 9,250
  • Thank You
Reactions: 1 user
Excellent work Christer, very informative (even though you claim it is not scientific, I believe it to be very good work).

So if I'm not mistaken, the Infrared Led at 20ma is nearly as good as it gets! Wow I never tought I'd use a voltage reference in a CCS to heat my dinner :)

Thanks for your devotion to this forum!
Sébastien
 
I'll try to briefly summarize the conclusions one might perhaps
try to draw from the results:

The least noisy choice is an IR LED running at close to maximum
current.

Blue LEDS share the property with IR LEDs that they get less
noisy with larger current, but they are so noisy compared to
other LEDs that they are best avoided.

All other LEDs seem to have an optimal current for low noise.
however, don't draw the conclusion that this optimum is at
5mA just because I happened to use that as one of the test
currents. Further investigation is needed to find the optimum
current for various LEDs. The differences between colour in
the red to green area or for differenc currents are small though
and nothing to loose sleep over. Red LEDs at optimum current
seem the least noisy, though (next to IR LEDs at high current,
that is).

Zener diodes seem an altoghether different and more confusing
story. Below approximately 6V they have true zener breakdown
and above this they have avalanche breakdown. It was thus
obvious to test zeners close to and to each side of this
"transition region" and also zeneres far into each region.
Surprisingly, the important distinction seems not to be which
type of breakdown they exhibit, but how close to or far away
from the transition they are. Low volt and high volt zeners
turned out to have moderate noise and the high voltage
avalanche type could even rival most LEDs. Those diodes
close to the transition, on the other hand, turned out to
be noisy as h**l (maybe the electrons are confused and
don't know if they are doing a zener breakdown or an
avalanche one? :) ).

Zener diodes generally share with IR and blue LEDs that
the get less noisy with larger currents.

For high power zeneres vs. low power zeners, the high power
ones seem more noise in the zener region and less noisy in
the avalanche region. Since these diodes could take higher
currents, the noise could probably be brought down by
and rival the LEDs if going far above 20mA.


Please note that the study is much too small and has to many
error sources to really claim any of the above conclusions.
 
tool49 said:
Excellent work Christer, very informative (even though you claim it is not scientific, I believe it to be very good work).

So if I'm not mistaken, the Infrared Led at 20ma is nearly as good as it gets! Wow I never tought I'd use a voltage reference in a CCS to heat my dinner :)

Thanks for your devotion to this forum!
Sébastien

Thanks for the kind words.

An added bonus with the IR LED is that you will avoid stupid
questions why you have put LEDs inside the box where nobody
can see them. :)
 
Regarding the interesting results for zeners, i just made a
quick check with an 8.2V zener and it had reasonable noise
values of 1.1 uV, which should be at least as good as stringing
up a corresponding number of LEDs. Hence, it seems there
are mainly the zeners just around 6V that are terribly noisy.
i suspect it has something to do with these diodes
having a mixture of zener and avalanche breakdown or
something like that.
 
mandat said:
The same colour LEDs have got a different kind of semiconductor inside. Have you got such information for tested LEDs.

I am not sure I understand what you mean, I am afraid. If you
mean that different types of LEDs with the same colour use
different chips, then yes I am sure they do and a more
comprehensive study should include several different types of
LEDs with the same colour, among other things.
 
diyAudio Senior Member
Joined 2002
Hi,

Well done, Christer.

If ever you'd stumble upon some indepth study on the noise behaviour of zeners/LEDs or even a spectral analysis of their noise I'd be happy to see it.

Parallel to your "exercise" we did similar analysis in VR tubes (the granddaddy of zeners) and saw similar changes in noise levels with changes in current.

The spectrum of the noise was generally evenly spread, not peaky as withsome other components.
This is important for use in audio gear as this seems to less annoying compared to peaky noise bands.

Again congrats, much appreciated.

Cheers,;)
 
Thanks Frank,

yes I think an obvious next step would be to look at spectra
too. However, already the RMS values raise a number of
new interesting questions that should also be addressed
by me or someone else. How does the noise vary if testing
a larger number of the same device type, especially from
different batches? How does it vary for similar but not identical
types of LEDs? Where is the oprimum current for LEDs?
Are the observations on the strange noise values for zeners
universal, or do they to some extent have to do with statisticcal
variations? ...?
 
mandat said:
I mean that LEDs may be manufactured from a different type of semiconductors: AlGaAs, GaAs, GaP, InGaN ect. For example diode may be manufactured from GaAsP and the colour is:
red UF=1.65V,
yellow UF=2.7 - 3V.
UF - voltage drop for IF=10 or 20mA

I am no expert on this, but doesn't the specific doping
substances decide the spectrum, so all LEDs with the same
colour have the same doping substances? Hm, I just checked
som datasheets and there seems the same substances, for
instance GaP can cover a wide spectrum, so presumably it is
the relative ratios of substances that matter?

I am still not sure what your point was?
 
I think the point was that you cannot say that a given color led has a certian behaviour. You need to specify probably the voltage drop, color and perhaps the material in the diode, or the brand and model as you did.

:bigeyes: Thanks for sharing these results. It was one of the things i have been looking for.

I just have one question. I am trying to figure out if you stack LED's to acheive more voltage do you add the RMS noise figure or?
 
Christer said:
The study and the measurements so far are described in
the attached document. Please note that the study is too
small and uses too questionable equipment to be considered
properly scientific, so no hard claims or conclusions can be
drawn from it.
Impressive work :)

Did you rule out the noise from the power supply?

I have an another question to the experts.

In my youth I measured the I/V of diodes and LED's, different colours.

Is it true that the leakage current is extremely small for LED's, I mean in real life? If you convert the I/V to the diode formula, this leakage current will take the value of fA!!! Can this be true?
 
Re: Re: Some noise measurements for LEDs and zener diodes

hjelm said:
I think the point was that you cannot say that a given color led has a certian behaviour. You need to specify probably the voltage drop, color and perhaps the material in the diode, or the brand and model as you did.


I just have one question. I am trying to figure out if you stack LED's to acheive more voltage do you add the RMS noise figure or?

I am currently measuring some other LEDs from my junk box to
get some more variation to the mix.

Noise add geometrically, or how it is called. That is if you have
n noise voltages v1,..,vn you get the total noise as
sqrt(v1^2+..+vn^2)


peranders said:

Did you rule out the noise from the power supply?

I use batteries and as can be seen from the schematic
each op amp as well as the current sources for the DUT
have a series resistance of 10 Ohms on the supply line
and is decoupled with 100uF in parallel w. 10nF. Further,
I took care to have a separate ground lead for the op
amp decoupling to separate this from the low-level ground.
As you can see from the idle measurements, the measured
noise coincides well with the calculated theoretical noise.
The major error factors here are the unknown bandwidth
of the soundcard input filter and that I calibrate the program
using my scope which has never been calibrated in over
25 years (the point of the experiment wasn't to get any
exact absolute figures anyway, but try to understand the
relative performance of various diodes.) I should add that
I could not see any hint of hum on the scope for any of
the measurements, except if attaching the DVM to the DUT.
Hence, I could not measure voltage drop while measuring
noise.