LEDs as references in current sources

[nowhere]

Hopefully it's the right sub-forum to discuss the issue.

"Here, there and everywhere" (c) I see LEDs in current sources and the like. I.e. the LEDs are meant to "dictate" current according to I = (Vled - Vbe) / Re, where Vled is voltage on open LED (IIRC 1.8V on red LED, 3.3V on while LED), Vbe is about 0.65V for silicon transistors, and Re is emitter resistance.

So far so good.

I am no semiconductor physicist, but I vaguely remember that LEDs are noisy. I.e. I think they are noisier than conventional PN junctions. If so, current sources based on LEDs as references will be noisy too.

Since in this place we are talking about truly HiFi/high end audio, couldn't using LEDs be an issue ?

I mean places like, for example, all kinds of I -> V converters for DACs.

[martin clark]

LEDs are about the quietest reference going! The 'leds are noisy' meme is archetypal audio myth.

LEDs constitute a forward-biased PN junction and the best types can exhibit a dynamic impedance of 10ohms or less. That makes for exceptionally low noise generation.

Some years ago DIYAudio member Christer published an extensive series of tests - try a search for the results, it is good reading. More recently SY has quoted the figure of 0.4nVSQRT(Hz); from my own experiments this does not surprise me.

[nowhere]

Originally Posted by martin clark

LEDs are about the quietest reference going! The 'leds are noisy' meme is archetypal audio myth.

LEDs constitute a forward-biased PN junction and the best types can exhibit a dynamic impedance of 10ohms or less. That makes for exceptionally low noise generation.

Some years ago DIYAudio member Christer published an extensive series of tests - try a search for the results, it is good reading. More recently SY has quoted the figure of 0.4nVSQRT(Hz); from my own experiments this does not surprise me at all.

OK, I'll ask the question differently. If I compare LED to, say, 3 conventional PN junctions in series, which implementation is less noisy ?

Regarding LEDs - I read that reverse biased LEDs are noisier than reverse biased silicon diodes. So I thought there may be something in LEDs which makes them noisier when they are forward biased.

My vague and maybe wrong points are:

1) in LED hole <-> electron recombination can be more extensive - because this is what produces light, isn't it ? Aagain, I'm no semiconductor physicist);
2) diodes/transistors are made with signals in mind while LEDs are made with emission of light in mind, so maybe conventional diodes and low noise transistors use materials and technology with less noise producing imperfections.

[Chris Daly]

Hi
Best to read Christers post.... here:
Some noise measurements for LEDs and zener diodes



Cheers / Chris

[nowhere]

Originally Posted by Chris Daly

Hi
Best to read Christers post.... here:
Some noise measurements for LEDs and zener diodes



Cheers / Chris

Thanks for the pointer. The post contains an attachment: http://www.diyaudio.com/forums/attac...ements_1_4.txt .

In that attachment I see among other things:

1N4148 (unknown brand):
#1 @ 1mA: 0.28 0.28 0.28 0.27 0.27 uV
...
EL202HD (red):
#1 @ 1mA: 0.31 0.32 0.31 0.31 0.32 uV
...
BC549 BE diode forward biased:
#1 @ 1mA: 0.24 0.24 0.24 0.24 0.24 uV

So, LED noise is slightly higher, but since voltage drop on it is about 3 times higher than on Si PN junction it looks like indeed LEDs are less noisy per 1V of reference voltage than Si PN junctions.

...

You know what, I'm wrong. Since noise in each device is random, it adds up as sqrt(Vn1^2 +Vn2^2 + ... +VnN^2). So noise of 3xSiPNis about 4db noisier than red LED, but still noisier.

...

And maybe this whole thing is wrong. This is because self-noise:

Idle noise:
---------------------------------------------------
Measured idle noise of amplifier with grounded input:
0.19 0.19 0.19 0.19 0.18 uV

is significant. I.e. one has to calculated not sums/differences of voltages, but noise factor as it is defined for transistors.

But, anyway, it looks like per 1V of reference voltage LEDs are better.

And yes, LEDs should be insulated from ambient light - because they are also photo-diodes.

...

Reading the thread further - Christer in Some noise measurements for LEDs and zener diodes talks about square roots too.

[AndrewT]

Have close look at how Roender implemented his CCSs in that now famous amplifier.

He even uses close coupling of transistors with LEDs to compensate for temperature.

It really is a fantastic implementation of many design ideas.

[SY]

HLMP6000, a very common red LED, shows noise at (average) 0.4nV/rt Hz. That's extremely quiet. In my His Master's Noise phono stage, LEDs are effectively in series with the signal and their noise is well buried in the thermal noise of the cartridge and tubes.

[ahaja]

Very usefull thing: http://www.google.pl/url?sa=t&rct=j&...DdnKM4HPxIYHiA