Hey Glenn,
Also FWIW, in our ASP pulsar backends (serendip5-based) we've used both 8
and 24 taps in a 32-channel PFB.
For the "simple spectrometer" application where the PFB output goes
straight into detection, I think Dan is right that a huge number of taps
isn't too useful. However, if the PFB output will be further spectrally
divided (or coherent dedisp, etc) then it's worth thinking about the exact
FIR filter shape and how much aliasing/leakage you can tolerate between
PFB channels, depending on your application.
-Paul
On Thu, 22 May 2008, David MacMahon wrote:
Hi, Glenn,
FWIW, the ATA correlator's 1024 channel PFB uses 6 taps.
Dave
On May 22, 2008, at 13:42 , G Jones wrote:
Hello,
I understand the theoretical benefits of increasing the number of PFB
taps in making a spectrometer, but I was wondering how many taps
people are actually using in their spectrometers. Do most people just
stick to two taps?
Thanks,
Glenn
