spectrumMeasurement function

That's not really enough information to go on. What does the input data look like? I assume the OUT signal is the output of an ideal DAC? The best thing is really to do this via customer support, but perhaps if you upload a dump of the OUT waveform (for two different stop times) exported from ViVA, with the appropriate spectrumMeasurement expressions you used, then I may be able to take a look.

Andrew

Thanks for the reply Andrew. DOUT waveform is below for reference and is the output of the transistor level ADC not verilog code. The input waveform is also shown 

tstop = 100us

tstop ~ 52us

expression for snr - 

spectrumMeasurement(sample(v("/DOUT" ?result "tran") 9e-08 VAR("tstop") "linear" 1e-07) t 9e-08 5.125e-05 512 19530 5000000.0 3 "Hamming" 0 0 1 "snr")

thanks 

Sounds like a rather odd transistor level ADC which has the "digital" output being a continuous single value looking like a sine wave with 500V peak. How is that an Analog-to-Digital-Convertor?

Anyway, pictures are not what I'd need - I was asking for the actual waveform data.

Andrew

Dear

sidm said:

I am using the spectrumMeasurement function to evaluate the SNR of an ADC in IC618 and am seeing that the output of the function changes with different values stop time of transient analysis. 

Is there any other approach that I can use to estimate SNR of ADC ?

I noticed from your expression you are using a Hamming window. I also tried to expand your two waveform plots to estimate the frequency of your input sinusoid. It appears to have 6 periods in 10 us for a period of about 0.60 us. Your analysis time is of 90 ns to 5.125 us suggests their are a non-integer number of periods in your analysis. I think if you want the most accurate estimate of the ADC SNR, I might suggest you choose an analysis period that is an integral number of input waveform periods. The motivation for this is the signal power will all be contained in a single frequency bin. In your case, it will be spread over multiple bins. The window you are using will assume all of the power is in three frequency bins - which may not be the case. It will also impact the amplitude of the signal as it has a frequency response that is not 0 dB. Using a integral number of periods of the input wave in your analysis period and a rectangular window will assure a much more accurate estimate of the signal power.

Why are your SNR results different for the two simulation stop times? I don't know how "different" your results are, but it may simply be that the simulation setting you chose for each of the two simulations do not produce exactly the same sampled waveforms. Have you created the sampled waveforms and examined if they they differ?

Just a couple of thoughts sidm...

Shawn

Dear sidm,

If it is of any interest, the Spectrum Assistant has a number of subtelties - one of which impacts its reported SNR. We had a discussion of this in the Forum post at URL:

https://community.cadence.com/cadence_technology_forums/f/custom-ic-design/51696/snr-calculation-wrong-by-3-db/1383427?focus=true

and I placed a note on a study of them at URL:

https://www.dropbox.com/s/lxjvo8ys6ci5m3x/test_sinusoid_rnoise_snr_study_061522v1p1.pdf?dl=0

This is included in the Forum post.

Shawn