Linewidth Measurement Simulation

I’d like to create a self-homodyne linewidth measurement setup by using a delay line to prevent any coherence effects between the two signals. I created the below circuit, where a CW input with 10 MHz linewidth is split into two. One arm is passed through a 10 km long lossless fiber. The beat note is then read with a photodetector and a spectrum analyzer. However, it seems like the system isn’t responding the way I’d expect.

  1. First of all, are there any coherence effects in INTERCONNECT to begin with? Can I expect to see anything with a setup like above?
  2. The electrical spectrum analyzer’s resolution is way too coarse to see anything. The first data point after 0 Hz (DC) seems to be 0.195 GHz, and I wasn’t able to find a way to change this. Refresh length and display memory length parameters didn’t make a difference. So there’s no way for me to zoom into the data enough to fit a Lorentzian lineshape to the beat note.

Thank you for your help!

Hi @esm Emir,

Thank you for your post. Regarding the questions:

  1. Yes, there are coherent effects in INTERCONNECT. With the self-homodyne linewidth measurement, you could expect to see the SSB 3 dB bandwidth for the detected signal to be roughly the linewidth of the laser source. However we do have to carefully tune the system a little bit to get it to work.

Please check out the attached file homodyne_linewidth_measurement.icp (209.9 KB) where I duplicated your circuit. I changed the length of the fiber to be 200 m (the coherent length for the fiber is roughly 20 m, which 10 km is way too long for this measurement) and set the bit rate to be 1 GHz and 16 samples per bit, both just to tune the resolution of the RF spectrum analyzer. Now if you enlarge the power spectrum, you could at least see several points in the MHz region. The script file plot_spectrum.lsf (334 Bytes) plots the received signal power spectrum only to 100 MHz, you could find the 3 dB bandwidth is around 6 MHz to 7 MHz in this figure. This is not fully accurate yet and we need to tune the system further to get a better result.

  1. The resolution of the RF Spectrum Analyzer depends on the received signal. You could enlarge the sample period to have a better resolution in the frequency domain. In the attached file, I enlarged both the sample period and the bit period. The “refresh length” only matters when you monitor the data while you run the simulation (the figure will update for every refresh length), so it won’t affect the resolution of the figure. You could keep playing with the system setting parameters for a better resolution.

I hope this could help :slightly_smiling:

there seems to be something wrong with this simulation.

I would suggest shifting the frequency of one of the arms so that the beat signal is created not around DC

Dear @gwang
Thanks for your homodyne linewidth setup model.
Regarding your model:
the resulted spectrum strongly depends on the “bandwidth” property of the RF spectrum analyzer.
for example in your file, the properties of the RFSA are “resolution= Gaussian Function” & “bandwidth = 10 MHz”. However, if we change the bandwidth param (for example to 50 MHz), the spectrum will completely change and the measured linewidth of the laser would be completely wrong!
In this example, we set the laser linewidth to 10MHz and we tuned the RFSA to show the linewidth correctly!
So my question is that how is it possible to use this homodyne setup to measure the linewidth of a complex structure acting as a laser that we don’t know the exact number of its linewidth?
foe example: in Complex_Laser_Linewidth_Example.icp (1.5 MB), there is a structure for a laser connected to the homodyne setup, fig1:

. The laser spectrum is shown in fig2:OSA_3
However, the resulted linewidth diagram is strange! fig3:RFSA_1

Thanks in advance for your help and support.
Sincerely,
Emad