QAM sequence generator issue

[Dr. Dhiman Kakati]

Hi all,
I am finding some issue with electrical 16QAM sequence generator when it is operated at 10Gbps(one constellation point is missing), but it is working fine for 100Gbps and 1Tbps input sequence, I have attached .osd file please go through.

With regards,
Dhiman Kakati

[Dr. Dhiman Kakati]

Same problem for 15Gbps also.

[Dr. Dhiman Kakati]

upto 17Gbps the problem exists, But from 18Gbps it is working fine.

Thanks

[alistu]

Hi Dhiman,

I ran your simulations and I believe they work fine. I actually saw all constellation points. The reason for one constellation point missing is probably the fact that your sequence length is too short for all constellation points to be included in it. Sometimes they happen and sometimes not. Please increase your sequence length, say, to 4096 and try again.

Regards

[Dr. Dhiman Kakati]

Thanks a lot Alistu, It works.

[alistu]

You’re welcome!

[Dr. Dhiman Kakati]

Dear alistu, I got the solution but one more thing I need to mention that the same sequence length(ie, 128) works fine for higher bit rate but not working for lower bitrate. i dont find the logic.

[alistu]

As I mentioned in my previous comment, it all depends on the random bit sequence that is generated by your PRBS generators. Sometimes all the patterns to make all constellation points are created and sometimes not. I don’t think it depends on the bit rate. However, if you use a user defined sequence and define it so as to contain all patterns, you always have all constellation points. I suggest you use larger sequence lengths instead, though.

[Abhishek Shrama]

Hi! Dhiman Kakati

Suggestion made above is good as it solved your problem.But it is not quite obvious that works every time.However the increase in sequence length increase the accuracy of calculation and show less fluctuations as in the case of lower seq. length 128.So using longer sequence length for any system makes it more accurate but due to increase in total no of samples it take more time .
Thank you

[Dr. Dhiman Kakati]

Thanks Abhishek for the reply, according to me the higher bitrate should require more sequence length for accurate detection of bits, in my case I am getting the opposite one.

With regards

[Dr. Dhiman Kakati]

Thanks Alistu for the clarification.

Regards

[alistu]

Glad to be of help. You can even check and see if what I said is actually the reason for the problem you mentioned by checking for the bit patterns (while choosing short bit patterns) by using a visualizer that shows the patterns generated by PRBS generator. However, if you agrre with me, checking is not necessary as I am rather sure about this.

Regards

[Dr. Dhiman Kakati]

Dear Alistu, Thanks for your support. The question was in the PRBS generator the bit rate. According to me the higher bit rate need higher sequence length to detect accuratelately, but in my case the lower bitrate requiring higher seq length to come all the constellation point. as the higher seq length simulation more simulation time. As my circuit which I am working is quite big so putting such a hi sequence length increases the simulation time. So i was mentioning the same question again.

Thanks for cooperation

[alistu]

I don’t actually think increasing bit rate has anything to do with increasing sequence length, and I would be glad if you explain it more. But in general, when the bit rate goes up, the systems are more susceptible to error, and this may occur either when the bit sequence is short or long. But generally, the longer the sequence, the more reliable the BER results.

[Dr. Dhiman Kakati]

Dear Alistu I agree with you that “bit rate goes up, the systems are more susceptible to error” so i mentioned that as all the constellation points are coming properly for higher bit rate with less sequence length of 128 go 100Gbps so there should have come more clear constellation point for 10Gbps keeping the other parameter constant. So the main issue was this. It may be due to constraints taken or mathematical equations or principle on which this 16QAM block works but as the problem is solved by increasing the sequence length for more accurate results and the problem is solved, I think we should close the discussion here.

Thank u all for the replies.

[alistu]

Thanks for the explanation. I only needed to know on what basis you mentioned your former statement. So we agree short bit sequence of the discussed issue here.

Regards

[Dr. Dhiman Kakati]

you ar welcome Alistu.

Regards

[Author]

Posts