Further questions about ring modulator simulation in interconnect


#1

Dear Lumerical staff,

I come up with some additional questions about ring modulator in interconnect and I have attached several files.

freq_domain_R=10um.icp (826.6 KB)
freq_domain_R=50um.icp (949.5 KB)
neff_vs_voltage_thesis.txt (66 Bytes)
time_domain_ref.icp (2.4 MB)

  1. In frequency domain file I have to racetrack ring with 10um and 50um bending radius. I specify the round trip loss by an optical attenuator (2dB loss) and set coupler to have roughly the same power coupling (near critical coupling). Then As I try frequency domain simulation, I saw Q factor is different for two structure. My understanding is Q factor only depends on the round trip loss while optical pah length (round trip time) only determine the FSR. However it seems Q factor is also affected by optical path length here and do you think that is correct?

  2. I also run time domain simulation and use electrical eye diagram to check the signal integrity. However eye diagram seem to give me some wrong BER measurement (BER=7.5e-8 for 1.75dB ring loss and BER=3.5e-11 for 2dB ring loss).
    As there i hardly any difference in eye diagram, I don’t think BER can have such difference. As I only use the default setting for eye diagram, do I need to change some setting to make output correct?

  3. I have input and output electrical waveform and I wish to calculate energy efficiency (fJ/bit). How can I proceed if I only know the voltage and bit rate?

  4. Currently I assume OM follows the lookup table for arbitrarily high modulation speed. In reality there can be some limitation such as carrier drift speed that perturb using such perfect relation. Is there any Lumerical example covering this aspect?

Best
Arthur Teng


#2

Hi Arthur,

Regarding your questions:

  1. I modified the frequency domain files a bit (to match the Optical Modulator length and the ring path length) and the Q factor of the two models sort of matched now. Here are the modified files:

  2. The BER is calculated from the Q-factor and if you do a frequency domain analysis, with attenuation = 2 dB, the Q-factor is actually larger. I will investigate this a bit more and follow up with you in a separate post.

  3. The energy/power information can get from DEVICE. Then the energy efficiency can be calculated accordingly.

  4. I assume the carrier drift speed can get from DEVICE simulation? There is no available example covering this aspect; I am not an expert on DEVICE, let me discuss about this with my colleague and follow up with you.

I hope this helps.