pichuang
Newbie level 6
Hi everyone,
I am working on a project right now that involves with a FM modulator and a frequency discriminator. The output of the FM modulator goes into the frequency discriminator consisted of a Dual Modulus Divider (DMD) and a Digital Flip Flop (DFF). To determine the dynamic range of this design, I need to perform a FFT analysis at the output of the DFF. However when I convert the time domain signal to frequency based, all I see is white noise so far.
I test the components individually and they are all working properly. Right now I think the cause is the incorrect FFT configuration. Does anyone has experience with this type of analysis?
Pierce
Added after 21 minutes:
Hi,
After a little search on the board, I found the following message and I think this is where my problem is
By naalald @ "A Question About Delta Sigma Modulators"
I got the answer! It's mainly related with the ".FFT" syntax parameters. There are two points:
1- the start point (tstop) should be selected exactly before the end of phi2 phase.
2- the stop point should be selected so: tstop=tstart+nfft*ts in which nfft is the number of fft points taken and ts is the sampling period.
Make sure nfft is a power of 2.
jiangxb was right! Thanks jiangxb!
Does anyone know what naalald mean by that the start point should be selected exactly before the end of phi2 phase?
For the second point, assume My sampling frequency is 225Mhz. and the tstart is lets say 20uS. Then for 4096 samples, tstop = 20uS + 4096*(1/2250000) = 1.84ms?
For the psb function in the cadence calculator, I also need to specify the window size. Can anyone give me some ideas on how to select the optimal window size?
I am working on a project right now that involves with a FM modulator and a frequency discriminator. The output of the FM modulator goes into the frequency discriminator consisted of a Dual Modulus Divider (DMD) and a Digital Flip Flop (DFF). To determine the dynamic range of this design, I need to perform a FFT analysis at the output of the DFF. However when I convert the time domain signal to frequency based, all I see is white noise so far.
I test the components individually and they are all working properly. Right now I think the cause is the incorrect FFT configuration. Does anyone has experience with this type of analysis?
Pierce
Added after 21 minutes:
Hi,
After a little search on the board, I found the following message and I think this is where my problem is
By naalald @ "A Question About Delta Sigma Modulators"
I got the answer! It's mainly related with the ".FFT" syntax parameters. There are two points:
1- the start point (tstop) should be selected exactly before the end of phi2 phase.
2- the stop point should be selected so: tstop=tstart+nfft*ts in which nfft is the number of fft points taken and ts is the sampling period.
Make sure nfft is a power of 2.
jiangxb was right! Thanks jiangxb!
Does anyone know what naalald mean by that the start point should be selected exactly before the end of phi2 phase?
For the second point, assume My sampling frequency is 225Mhz. and the tstart is lets say 20uS. Then for 4096 samples, tstop = 20uS + 4096*(1/2250000) = 1.84ms?
For the psb function in the cadence calculator, I also need to specify the window size. Can anyone give me some ideas on how to select the optimal window size?
