Dear Community,
I performed a very simple PSD calculation for a periodic pulse train that is generated using Verilog-A modelling. The pulse train has fixed high voltage 1V and low voltage 0V, fixed duty cycle of 50%, and fixed frequency of 1s. I ran the Transient Noise simulation using ADE L for 1000s (usingt Transient Noise analysis in order to directly acquire ADE L -> result -> main form -> PSD once the simulation finishes). Below are the signal waveform and transient noise simulation setting:
The PSD calculation expression is therefore: db10(psd(VT("/M") 1 993 198400 ?windowName "Rectangular" ?smooth 1 ?windowSize 12400 ?detrending "None" ?cohGain 1))
However, after calculation, I got the spectrum as shown below, which has the frequency-domain spikes with height 10.99dB at fundamental (1Hz) and 1.45dB at 3rd harmonic (3Hz) and so on, which is counter-intuitive. Because theoretically for a standard 50%-duty-cycle pulse train the PSD at fundamental should be 10log(Cn^2)=10log(0.3183^2) = -9.94dB and at 3rd harmonic 10log(0.1061^2) = -19.49dB, and so on.
Is there anything incorrect with my method of acquiring the power spectrum?
Also, I tried to perform the PSD calculation using the same method (but updated simulation time and sampling frequency, etc.) setting the 50%-duty-cycle pulse train frequency to 100Hz and 10kHz. But the frequency tones at all the harmonics are -20dB and -40dB lower throughout the spectrum compared with the 1Hz case. Why is that (should the heights at the corresponding harmonics not be the same for those three cases)?
Many thanks!