A question about 2-1-1 1bit MASH Sigma-Delta adc!

[gdhp]

Hi everyone

I met a question in the designing the mash sigmadelta adc.

In my design, it is a 2-1-1 mash, every stage has 1-bit flash. I use the scaling factor as in book(cascade sigma delta adc for sensor and telecom).

I have finished the schematic, and is doing the spectre simulation for top. I simulate for 250u, and get the output data, do fft in matlab.
A problemn is arised. If i use the output data from 50u~160u, its snr is about 79dB, as with my expection. If i use the output data from 50u~180u, its snr is only about less than 60db. And if i use the data from 180u~250u, its snr is also ok, is about 79dB.

So i think the data from 150u ~170u is not correct. But what is the reasons?
What is the root course?

The results of fft results is as the figure attachment.

 

[gdhp]

can some give some suggestions?
By now, i think maybe it is due to the opamp stability.
Can someone give some?

Thanks

 

[JoannesPaulus]

It cannot be a circuit issue since the data is good between 50us and 160us and then again between 180us and 250us... **unless** you have a reset condition due, for instance, to overloading.
What happens if you take the fft between 70us and 180us (i.e. the same amount of data as the first good SNR)?

 

[gdhp]

thanks JoannesPaulus for your reply.
As you said, if i use the data from 160u~170u, the fft result is also not correct.
By now, i haven't found the reasons. Whether due to the opamp? or just because simulation data number is not much enough?

Thanks

Added after 2 minutes:

another i only add the sigal with -5dbFs, so overloading should be not happened in theroy.

 

[gdhp]

any suggestions?