Sources of error in OPU measurements

Hello,

I did some experiments last year using the Aurora OPUs. We have y = |Ax|^2 where A is roughly iid standard complex Gaussian. I would like to understand the sources of errors in y.

I understand that one source of error in y is due to quanitzation because the camera takes 8-bit measurements. Is there some Poisson noise as well and what is the amount of this Poisson noise?

Furthermore, is there some correction done so that ‘floor’ noise is removed. By floor noise I mean when a zero signal, x, does not give a zero measurement, y?

What other sources of error are there in y?

Thanks!

1 Like

Hello Sidharth, good to hear from you :slight_smile:

You essentially said it yourself: there is shot noise (Poisson process) and floor noise (we do not remove it). The other sources are negligible with respect to these two. If you need a very precise characterization I can give you more details on these negligible contributions.

Best,
Iacopo

1 Like

Hello Iacopo, nice to hear from you too!

So is the amount of Poisson noise proportional to the intensity of the optical field hitting the sensor? So proportional to |Ax|^2 ?

And, my understanding is that the floor noise is because the camera is unable to detect lower intensities. Why is this? And does this mean that for a zero input signal, the shot noise is proportional to the floor value instead?

Kind Regards,
Sidharth

1 Like

The fluctuations due to Poisson noise scale with the square root of the mean intensity.

The noise floor is due to a complicated combination of noise sources due to the system. Thermal electrons generated per pixel that get amplified by the ADC are a source of noise, then there is a contribution due to the fluctuation of the reference voltage, and burst and pink noise. This article goes a bit more into detail. You can characterize the noise floor by taking images with a zero input - usually, suppliers give the median of the distribution (for example Figure 2 of this article).

Best,
Iacopo

1 Like

Thank you Iacopo! This is useful and helps. I will follow-up if I have further questions.

Kind Regards
Sidharth

1 Like