Learning Forum

[Arka Dipta Das]

Hello!

I have the s-parameters of a 1x8 MMI in SiN that has a group delay of 2.4e-12, also made available in the Lumerical s-parameter export file with the "calcualte group delay" option.

When I use a ONA to analyse the element in Interconnect after importing the data, I can see that the group delay is is correctly displayed in the results. However, when send a SECH pulse in the input and compare it with the output with OOSCs, the group delay varies a lot. I have take into consideration aliasing and selected an appropriate sampling rate. I have even carried out a parameter sweep of the sapmling rates and looked at the respective group delays obtained. The result does not show a group delay that saturates towards the calculated group delay at high sampling rates. Rather, it has a saw-tooth like behavior and drops by an order of magnitude below the expected value.

I have attached a few slides to demosnstrate the problem. I would appreciate some insights into what might be causing this issue.

 

[GWANG]

Hi Arka Dipta Das,

 

Thank you for the post. I wonder waht option you use for the s-parameter element digital filter? I would recommend changing the option to "group delay" to try this again. Please let me know if this doesn't help. 

[Arka Dipta Das]

Thank you for your speedy response. I think this might have resolved the issue. I do remember trying the simulation with the group delay option once but at that point of time I was monitoring multiple stages. I will run a second sampling rate sweep vs group delays and share both the previous and new results for the benefit of anyone else who might come across the same issue. Many thanks again, in advance!

From my understanding based on the help articles, the group delay at the center frequency is used to factor-in the delay contribution due to the highest frequencies that cannot be detected by lower sampling rates. My question is, how is the ONA automatically aware that this can be utilized without the digital filter setting being set to "group delay"? 

[Arka Dipta Das]

Here is the comparison of the group delays seen when using the two different settings. In the second case, the GD is very close to the expected value of 2.4e-12.

 

[GWANG]

Hi Arka Dipta Das,

 

Thank you for the feedback. There is another parameter called "factional delay" which is in respect to the delay and sample time, and enables it allows using a fraction of the sample time to be the delay in the element (e.g., waveguide).