The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> DFT Assistance
https://designers-guide.org/forum/YaBB.pl?num=1510971759

Message started by MWJ1975 on Nov 17^th, 2017, 6:22pm

Title: DFT Assistance
Post by MWJ1975 on Nov 17^th, 2017, 6:22pm

Hi,

I am trying to simulate the DFT response in Cadence. I have constructed an LC Cross-Coupled oscillator operating at approximately 1GHz. I used the 'frequency' function in the calculator to obtain the value above and I am getting a free running frequency of 1.005GHz. I would like to analyze the harmonics of the signal (at least of to the 4th harmonic).

I attempted to use the DFT function but I am getting an error when I try to plot my response. The first thing I did was run a transient analysis for 300ns. I then clipped the waveform from 50ns to 100ns. Setting the 'From' and 'To' of the DFT function is where my main confusion arises.

From the initial values set, I am assuming this is the start and stop time. I would like to examine 1GHz and 4GHz, so my logic was to start at 0.2n to 2n. Which I was assuming would include frequencies from 500MHz to 5GHz. I chose a sample value of 2^15 and left all the other options to the default value. I am not able to get any sort of response.

I am having trouble locating the documentation for the DFT function, so I last resorted to this. Could anyone provide any tips or relevant information? Thank you.

Title: Re: DFT Assistance
Post by sheldon on Dec 7^th, 2017, 9:21pm

MWJ1975,

The simple answer is that this problem is why Shooting Newton and
Harmonic Balance with autonomous-mode are provided.

If you don't have access, then you might want to consider the folllowing

1) Use the freq function to make sure the oscillator has settled to steady
state. As I remember the frequency returns the frequency of the last
period, you need to reach sinusoidal steady-state before performing
a DFT. If you don't the settling will contaminate the spectrum
2) There is no reason to assume that the frequency is the same at
50ns and 300ns, you need to check the settling before choosing the
window to perform the analysis.
3) Your time window of 50ns and period of ~995.02488ps are not
conducive to useful results. You will need to either define the
correct sample window based on the oscillator frequency or
use a window function
4) The resolution of your FFT is 20MHz, yet you need 5MHz resolution.
1.005GHz --> means you need tones every 5MHz or the DFT will
will be corrupted by spectral leakage
5) Did you strobe your waveform to produce the samples you need
for 2^15 FFT frequencies, the tool will interpolate values if the
simulator does not provide the data required for the DFT. Interpolation
error degrades accuracy

What you want to do can be done but you need to be more thorough
in your simulation setup in order to achieve accurate results.

Sheldon