Measuring Photodetector's Dark Current Noise

[strahd_von_zarovich]

Hi,
I am trying to measure dark current noise of a photodetector using transimpedance amplifier. I have read lots of whitepapers, but I couldn't find any information about how photodetectors dark current noise affect the measerement system. Does it contributes directly to opamp current noise ?(For example if In of an opamp is 0.1pA/Sq(Hz) and photodetector's In is 0.1pA/Sq(Hz), do I have to add them together to get 0.2pA/Sq(Hz) or do I have to calculate it by using the noise formula : Sq(In^2 + In^2)

If my feedback resistor is 10^7 ohm, then what is the total noise contribution from the opamp and photodetector current noise ?



Thanks in advance.

 

[FvM]

Uncorrelated noise sources always add as square root of squared values. Technically, current noise is the shot noise of dark current. In so far, the numbers are quite unreallistic. 0.1 pA/√Hz refers to a several 10 µA dark current which would simply saturate a TIA with 1e7 conversion factor.

 

[strahd_von_zarovich]

Uncorrelated noise sources always add as square root of squared values. Technically, current noise is the shot noise of dark current. In so far, the numbers are quite unreallistic. 0.1 pA/√Hz refers to a several 10 µA dark current which would simply saturate a TIA with 1e7 conversion factor.

Thank you very much for the answer.
I didn't understand how 0.1pA/√Hz refers to 10uA dark current. For 10kHz bandwidth, It must be 10pA RMS. Am I wrong?

So if the output noise of the TIA is 400nV/√Hz (with photodetector mounted) and 50nV/sq(Hz) without photodetector. Then;
Sq( (400nV/√Hz)^2 - (50nV/√Hz)^2 ) gives me the voltage noise density of the photodetector, so I can divide it by the feedback resistor to find current noise of the photodetector maybe?

 

[dick_freebird]

Detectors that are trying to "count photons" can also
suffer RTS / RTN and this is a very low frequency
(sub-Hz) that may elude classical noise measurement
(needs a real long integration time, or time domain
inspection on a real slow sweep).

You'd also like a baseline measurement of your TIA
with input impedance emulating the detector's Zout,
to get at (hopefully de-embed) the noise floor.