Hi,

I am trying to characterize the effective bandwidth of my track-and-hold circuit during the hold mode by using the method specified in "www.designers-guide.org/Analysis/sc-filters.pdf".

My understanding from page 16 is that time-averaged PXF analysis (as opposed to time-averaged PAC) provides the bandwidth accurately since it shows the transfer function from specified I/P sidebands to the O/P baseband (and hence avoids the artificial sinc roll-off in higher O/P sidebands due to the zero-order-hold).

The problem with the above approach is that it gives the correct gain only at integer multiples of the PSS clock frequency (the other frequencies are still colored by the sinc roll-off). Also, since the transfer function is distributed across sidebands, it is hard to automate the effective 3dB bandwidth calculation.

However, if I use the sampled-PAC analysis (not time-averaged) and look at the direct gain (Maximum sideband = 0), I can still plot the transfer function accurately. My understanding of sampled-PAC analysis (please correct me if I'm wrong) is that it samples the O/P at a specified time instant and creates a discrete-time signal with samples separated by the PSS beat period.

Thus, if I sample the track-and-hold O/P at a certain time instant in the hold mode, it is equivalent to repeating the sinc coloring at integer multiples of the PSS beat frequency and adding all the images together. This removes the sinc roll-off and you can see the bandwidth accurately.

Also, since I'm only looking at the O/P at a single sideband, it is easier to directly calculate the 3dB bandwidth.

Please let me know if my understanding is correct.

Thanks,
Arvindh