Hi You could time stamp the replies against the PC's clock. That would at least let you spot drop outs.
Bob On Mar 17, 2010, at 9:32 PM, John Miles wrote: > >> OK, so I tried this myself. A sloppy setup but never the less... The >> CNT-90 time-base output via a T-connection on B-channel and then over to >> the A input of the HP 5328, selected START A and then the T.B. OUT over >> to CNT-90 channel A. At 100 kHz setting on the HP 5328, the TI A->B >> measurement on CNT-90 shows some jitter, but RMS jitter is below 400 ps >> at least and ADEV readings somewhat below that. > > There's one other potential error source in Bert's case that might make his > results look a bit worse than they really are. TI addresses the 5370 in > talk/listen mode, and I've noticed that it can sometimes get into a state > where the counter appears to skip every other reading. I've only seen this > happen with 1-pps sources driving both channels but it could conceivably > happen on a 10 MHz test as well. The result would be 2 seconds' worth of > noise/drift between readings instead of one, which would artificially make > the graph look a bit worse. > > So if the counter's display seems to change only after every other START > cycle, that could be happening. The next test version will allow the > counter to run in talk-only mode with arbitrary measurement periods, which > should avoid this behavior completely. > > -- john, KE5FX > > > _______________________________________________ > time-nuts mailing list -- [email protected] > To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts > and follow the instructions there. > _______________________________________________ time-nuts mailing list -- [email protected] To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
