[Laser] differential detector

[Art]

At 11:22 PM 4/20/04, you wrote:
>I was browsing the web and found some articles on optical detection schemes.
>Most seemed to create a fuzzy spot above and behind my eyebrows ( OK, mostly
>above and beyond all of me. )
>
>I did get the idea that you can get some benefit from two detectors.  Trying
>to imagine how to build a device for experimenting, I thought of two
>photodiodes, cathode grounded.  The anode of each to the respective 
>non-inverting
>inputs of a low noise FET input dual op amp.  The rest of the circuit is 
>modeled
>after a high impedance differential input instrumentation amplifier, so 
>that the
>differential output of the two op amps is a high gain of the difference
>between the two photodiodes, without loading either.
>
>If you mount the two photodiodes in the focal plane of a telescope, they
>should get similar noise source input from the surroundings and the 
>"ordinary"
>light coming in from the telescope.  If you then point the telescope at a 
>signal
>source, such that its image falls on only one of the two photodiodes, then 
>the
>differential circuit should respond to the signal, which is a differential.
>The common input should be rejected.
>
>This scheme seems to have a significant benefit to small and point sources
>that are surrounded by noisy broad light sources.  An example would be a 
>laser
>signal from across a brightly lit city.  Maybe it would work if a laser was
>pointed at a cloud.   I would think that it would not work well with a 
>scattered
>signal as it would likely strike both photodiodes equally.  It does nothing
>about random noise that originates in the diodes and amplifiers.  In fact, it
>doubles the number of such noise sources.  ( I seem to remember if you add 
>the
>noise from two random sources you increase the noise power by 1.414 not by 
>2.0
>because sometimes the two random sources cancel. )  Also, I think that you
>would need a least fairly sharp optics, not necessarily telescope image 
>quality,
>but perhaps better than a plastic fresnel lens.
>
>Anybody tried this?  Or suggestions?



Hi James,

I just found your message, I had meant to reply long ago. Sorry it got lost 
temporarily.

2 photodiodes, 2 amps, an active mixer IC and extremely high grade light 
gathering optics is a high price to pay for rejection of ambient and 
interfering light. The optics is not a big deal, except that you want a 
large light gathering surface....so, you are going to need a 12 inch glass 
lens/mirror instead of a 12 inch fresnel. When you get into larger surface 
area optics, it is a very expensive proposition to have high quality light 
gathering optics.

I have 2 alternative suggestions, that is I am suggesting 2 countermeasures 
that should be considered before going to the added expense/complexity of 
the hardware you have suggested.

1)      Use small active area photodiodes. This is a common method for 
narrowing the field of view for optical receivers. A small photodiode 
placed at the focal length of the lens will narrow the receivers field of 
vision to the milliradian range (same at the collimated laser transmitter). 
If the focal length of the optics is long, this method allows one to take 
advantage of the fact that different wavelengths of light focus at slightly 
different distances...in effect, creating a nearly loss less optical 
passband filter!

2)      Use a 'leaky integrator' and a single photodiode instead. Make sure 
the time constant of the integrator is 10x slower than the frequency of the 
signal you are trying to receive (the desired signal). Feed the output of 
the leaky integrator back into the inverting input of the photodiode amp 
(at the junction of the photodiode and the op amps inverting input), in 
effect cancelling dc (constant background light level). Feedback must be 
through a resistor to avoid having the feedback take total control. This 
method allows full daylight operation of photodiode based receivers, an 
area that most have not considered to be worthy of experimentation because 
previous attempts at daylight laser communications have been implemented by 
dumbing down the receiver sensitivity to avoid overloading by ambient light.

Burr-Brown has an ap note describing this method, along with an actual 
design example, but I can't locate it in my archives here, it's not on the 
Burr-Brown website anymore and TI's website doesn't list it either (TI 
bought BurrBrown in 1999). This method allows a fully sensitive receiver to 
be used although the feedback electronics does generate a slight bit of 
additional noise. The only drawback is that the integrator takes several 
seconds (or tens of seconds) to cancel the background light level after the 
receiver is powered up.

I'm wondering whether anyone has tried method #2, or has the original 
Burr-Brown ap note describing this method???

Regards,

Art