Demodulation of an laser interferometer signal

[elch21]

Hi,

I am currently building a heterodyne laser interferometer in a Mach-Zehnder configuration. The output of my photodiodes give a phase (frequency) modulated signal with a carrier frequency of 80 MHz. The modulation signal has a maximal frequency of some KHz.
Because I am completely new to electronics and signal processing, I have some fundamental questions regarding the demodulation and processing of the signal. It would be great if someone who is experienced in those fields could give me some short answers:

- My first thought was to just digitize the whole signal and demodulate it afterwards. Unfortunately, I would need an ADC with a sampling frequency of at least some GHz, which seems to be pretty expensive. Is there any way to digitize a 80 MHz signal accurately without spending several thousand dollars?
Is it maybe possible to downconvert the signal to a certain extend in order to reduce the necessary sampling rate?

- Another option seems to be analog demodulation using a PLL and only digitizing the demodulated signal. As far as I can see, PLLs are quite cheap and should deliver a voltage depending on the phase of my signal. I know that analog processing would be more sensitive and less accurate, but I would accept this for now.
Are there any important specifications of PLLs I would have to consider in order to have the best possible results? I know that I have to pick the right input bandwith

Kind regards and thanks in advance
elch21

 

[KlausST]

Hi,

Why do you think you need GHz?
According Nyquist every sampling frequency above 160MHz is suitable for 80MHz.

And I assume in your application you may even use undersampling.

Klaus

 

[filip.amator]

Hi,

I am currently building a heterodyne laser interferometer in a Mach-Zehnder configuration. The output of my photodiodes give a phase (frequency) modulated signal with a carrier frequency of 80 MHz.

Depending on the application at the output of the interferometer you will get a signal which amplitude is proportional to the length difference of the arms (fringes). Describe your application more precisely.

 

[elch21]

Thanks a lot for your replies!

Depending on the application at the output of the interferometer you will get a signal which amplitude is proportional to the length difference of the arms (fringes). Describe your application more precisely.

I use exactly the setup shown here: https://en.wikipedia.org/wiki/Laser_Doppler_vibrometer

My application is measuring vibrations and displacements of different reflecting surfaces.

So in my opinion, the signal should not be amplitude modulated but only frequency modulated proportional to the velocity of the reflecting sample.

Hi,

Why do you think you need GHz?
According Nyquist every sampling frequency above 160MHz is suitable for 80MHz.

And I assume in your application you may even use undersampling.

Klaus

Are you sure this holds for frequency modulated signals as well? I am aware of Nyquist, but I would have thought that I need a higher sampling rate in order to keep track of small frequency variations... I have to admit that I am not shure at this point, would be great if my thoughts were wrong :grin:

 

[filip.amator]

You can put into RF double balanced mixer two signals: Fb which drives AOM/Bragg cell (LO port) and signal from photodiode (RF port) - as a result you will get at IF port signal proportional to displacement of your object (after filtering it).
Your photodiode will give you a beatnote signal between two laser beams: unshifted laser beam taken by the first beam splitter and laser beam with shifted frequency and with extra phase shift coming from vibrating object.

Do the math with frequencies at the diagram, you will get Fd at the and.

 

[elch21]

Great idea, thank you! This makes in general a lot of sense, I just wonder then why people in most cases bother to use PLLs...
Maybe there are problems regarding the phase information if you are just mixing it like this?

 

[filip.amator]

Ordinary double balanced mixer has got linear range of +/- Pi/2, it means that the output from the mixer/phase detector will wrap if your vibrations are bigger that 1/4 of the wavelength of the laser.
To be honest a mixer in that particular configuration is almost one third of the PLL. You don't have to use PLL, a simple frequency to voltage converter (AD650) will also do the job.