Synchronous IR beam break detector for outdoor use with indirect sunlight

[e100]

I couldn't find any info on how commercial systems work so by trial and error I have arrived at the following system which uses a pulsed IR LED and photodiode.
Below is my current algorithm.

• With the LED transmitter off, take multiple measurements of the photodiode current for 100 milliseconds, calculate the average background current and peak to peak noise.
• Pulse the LED (1 millisecond) and measure the photodiode current.
• Subtract the previously measured background current.
• Divide the result by the previously measured peak noise current to give the signal to noise ratio.
• Compare this to some threshold value.

Is this how commercial systems work?

Developing indoors where there is little ambient IR and testing outdoors in constantly changing conditions is a real challenge.

Mike

[bobaruni]

The way some beam break detectors work is by transmitting a modulated signal (square wave) and receiving the signal and only amplifying the AC portion of the signal though AC coupling, this eliminates the effects of ambient light.
In essence, the method you have described will work almost the same as AC coupling the signal, take care you have enough headroom (low enough gain so as not to overload in direct sunlight) in your trans-impedance amplifier.

[e100]

In essence, the method you have described will work almost the same as AC coupling the signal, take care you have enough headroom (low enough gain so as not to overload in direct sunlight) in your trans-impedance amplifier.

As I was stepping into the unknown, I went with DC measurements so I could detect if the ambient light was overloading the receiver. By luck it seems that I have about 50% of the dynamic range left when outdoors.
Indoors the ambient is virtually zero and the signal to noise is about 3:1.
Weather permitting I'll do some more outdoor tests tomorrow. 

[Kleinstein]

In principle the suggested method is ok. However they tend to use a faster sequence and one should also average over more pulses. So more like a 10-100 µs pulse every 1-10 ms.
The difference might also determined in an analog way.