Can a Lock-in Amplifier Handle -30dB SNR?

[SvetaP]

Hi,

I wrote a simulation of lock in amplifier using Matlab and now i want to test it for different SNR.
I added to the signal white gaussian noise. In all the spec that i read, always written that the lock in amplifier can handle -100dB of noise.

When I add a noise of SNR=-30dB the lock in stops working, when i look at the spectrum of the signal+noise i see that the component that i am looking for is covered by noise so no wonder that the lock in doesn't work.

what i don't understand how is that in spec always written that lock in works at -100dB? i think that i am missing something.

attached spec of lock in amplifier : http://infrared.als.lbl.gov/content/PDF/equipment/SR810830_specs.pdf

look at the part of dynamic reserve.

I didn't add any ADC and quantization yet. So the system only consist of sinus signal that is the signal and reference and Gaussian noise.

Thanks for any help.

 

[f95toli]

If you have truly white noise you can recover any signal if you average for long enough (the SNR improves as the square root of the number of samples).. Hence, you do not even need a lock-in amplifier (although using one will make the measurement much faster).

What is the time-constant of your simulated lock-in?

You should also be able to see the peak corresponding to your signal by just averaging over many simulated spectra.

 

[SvetaP]

Instead of using LPF in the end of lock in , i used averaging as perfect LPF.
explanation of simulation and results attached.