measurements on spectrum

Dear vamshiky,

vamshiky said:

1) calculator expression to find second/third/foruth maximum power tone from a spectral plot.

Also the frequency bin location of Nth max tone (analogous to xmax for time domain signals)

I think if you take a few minutes to think about your request, you will quickly realize that you need to write a custom SKILL or ocean based function. To my knowledge, an existing calculator function can not take the output of its dft() function and interactively scan the result to determine the ordered peak amplitudes and their respective frequencies. Further, as you have not indicated if the tones are integrally related to the sample frequency, their power may extend over several adjacent frequency bins. Hence, the entire power of a given tone is more accurately assessed by integrating over a number of frequency bins and not simply the amplitude of the contents on a single frequency bin. Hence, simply providing the power of a single frequency bin that meets your request may have limited accuracy.

I do have an ocean procedure that does basically what you are interested, but cannot release the code as it was done for my company. However, I assembled an example using a Virtuoso schematic that generates a sum of sinusoidal tones, performed a transient simulation using a strobe period that includes the dft size to assure the dft() function does not need to interpolate the simulation time domain data. I then used the dft() function to determine its Fourier components. Figure 1 details a plot of the resulting Fourier components. The ocean procedure writes the peak amplitude data to a comma-separated file as well as printing its result to the stdio. Its printed output follows and you will observe that the amplitudes and frequencies it determines correspond to those in Figure 1.

Simulation case number 1 has valid input.scs file.
process_case:
typicalmid
num_dft_points = 32768
tstop = 1000.000 ns.
sim_temp = 25C.
Number of dft points = 32768.
Clipping data between 0.000 ns and 1000.000 ns (1000.0 ns window).
Searching for tones whose amplitude exceeds minimum threshold of 10.0 mV...
Number of data points in sdd5e_TB_test_fft_tran_single_freq waveform vout (32769) allow specified dft length of 32768.
Tone 48: 46.67 mV at 48.0000 MHz
Tone 60: 58.33 mV at 60.0000 MHz
Tone 80: 333.33 mV at 80.0000 MHz
Tone 100: 83.33 mV at 100.0000 MHz
File name defined as: "sdd5e_TB_test_fft_tran_single_freq_vout_dft_spur_removed_dft_10.csv".

vamshiky said:

2) spectral measurement with time. Lets say  have a random signal and want to measure sfdr/hd3/enob etc with time.

I mean with multiple spectral measurements by charging the start time for DFT and finally plot sfdr with time.

This also cannot be performed with a single calculator function to my knowledge, but could be achieved with a custom SKILL or ocean procedure. Alternatively, one might create a number, N, of calculator dft() expressions to use in Assembler/Explorer with each expression containing one of the N start times you are interested. Please note that if you are changing the elapsed time over which the dft() is performed but maintain the same number of points when performing the dft(), without due care, some start times will likely require interpolating the data - which will degrade the accuracy of the dft() function. There is a Cadence Application note detailing accuracy considerations when using the dft() function at the On-line support URL:

Getting accurate results when using the calculator DFT and the Fourier component.

I hope this provides some insight into what you might need to get started vamshiky!

Shawn

Simulation case number 1 has valid input.scs file.
process_case:
typicalmid
num_dft_points = 32768
tstop = 1000.000 ns.
sim_temp = 25C.
Number of dft points = 32768.
Clipping data between 0.000 ns and 1000.000 ns (1000.0 ns window).
Searching for tones whose amplitude exceeds minimum threshold of 10.0 mV...
Number of data points in sdd5e_TB_test_fft_tran_single_freq waveform vout (32769) allow specified dft length of 32768.
Tone 48: 46.67 mV at 48.0000 MHz
Tone 60: 58.33 mV at 60.0000 MHz
Tone 80: 333.33 mV at 80.0000 MHz
Tone 100: 83.33 mV at 100.0000 MHz
File name defined as: "sdd5e_TB_test_fft_tran_single_freq_vout_dft_spur_removed_dft_10.csv".

Figure 1

Hi Shawn,

Thanks for the response, yes I understand  and need some skill/ocean script to get started here.

-Vamshiky