Home Forums SYSTEM Power Penalty Calculation

Viewing 1 reply thread
  • Author
    • #18421
      Parikhit Dutta


      I am having a problem regarding power penalty calculation. I have two systems which vary slightly, one with WDM Add/Drop Components to realize an Interleaver and one with FBG and Power Combiners. Both the projects are attached here.

      A Lithium Niobate MZM is used to suppress even order components and obtain the first and third components symmetric about the optical frequency. If f1 is the optical carrier frequency and f2 the frequency of RF signal, the components at the end of MZM are (f1 – 3*f2), (f1 – f2), (f1 + f2) and (f1 + 3*f2). The third order components are modulated by a PRBS-NRZ data signal following which (f1 – 3*f2) and (f1 + f2) are coupled for transmission over a SMF while (f1 – f2) and (f1 + 3*f2) are coupled for back-to-back transmission. An excel file is attached with the results for calculation of BER in order to compare a back-to-back connection as against an optical fibre.

      I have used an optical attenuator in order to get a BER in the range of 10^-15 to 10^-5 and one with 0 db attenuation. However I noticed that as attenuation monotonically increases the power doesn’t decrease monotonically and even the BER doesn’t increase. I have highlighted that in the excel file.
      Is the power level dependent upon the data signal or are there any other parameters. In the BER graphs that I found in several literature the log(BER) always decreases as power increases. Do I need to take the best fit values to obtain such a graph?

      Thank you so much.

    • #18447
      Parikhit Dutta

      I had mistakenly uploaded a different result. Sorry about the inconvenience. Attached herewith is the correct file.

      • #18453
        Damian Marek

        The randomness of the PRBS Generator is giving the different results here. If one sequence has many 1’s then the power will be higher, although the overall trend is a lower power. To resolve these differences try increasing the sequence length from 128 to a much larger number. This will give more accurate results for BER and smooth out the randomness of the PRBS.


        • #18469
          Parikhit Dutta

          Increasing the sequence length from 128 to 1024 has solved the problem. The BER increases with increasing attenuation.
          Thank you so much Damien.

Viewing 1 reply thread
  • You must be logged in to reply to this topic.