[time-nuts] 5370 processor boards available
kb8tq at n1k.org
Wed Oct 8 17:48:54 EDT 2014
The purpose of time stamping: You sample the phase of a signal and log the time at which you recorded the phase.
If you then want frequency, you can take the delta time between stamps and the delta phase between stamps and get frequency. If both time and phase are very accurate, the resulting frequency estimate will be accurate as well.
Taking a PPS output and time stamping it to picoseconds is one way to do this. It is always a zero phase (rising edge for instance), so it drops back to a delta time measure.
Looking at a 10 MHz sine wave with a counter is another way to do this. You likely also have a zero crossing, so again it’s a measure at zero phase, time stamped to picoseconds (or what ever).
It both of the cases above, your instrument actually reads out a “delta time” between the edge / zero crossing any your time stamp. If you look at the delta time as a phase, then you do have phase / time stamp data.
If you have a system that samples a beat notes out of a DMTD system (say with a good ADC), then you can indeed have a true phase number.
The whole “is it delta time or delta phase” is one of those things that can be caught up on in by reading several dozen papers that argue the terms back and forth. The result is that common usage is to call it phase.
On Oct 8, 2014, at 4:26 PM, John Seamons <jks at jks.com> wrote:
> On Oct 8, 2014, at 1:36 PM, "Tom Van Baak" <tvb at LeapSecond.com> wrote:
>> Time-stamping is wonderful. But note -- it does not imply using NTP! Best to time-stamp with some sort of fine nanosecond or picosecond XO or OCXO or Rb or Cs or GPSDO -- and not against the gross millisecond or microsecond PC / SBC / hardware / software cauldron of NTP.
> Okay, then I didn't quite understand that requirement of time-stamping.
> I made a real mistake by not running the 5370's 10 MHz oven clock, that was available right there on a processor board pin, to a GPIO on the Beagle so it could be accurately counted with the built-in event counter and software overflow (that clock used to drive the processor clock of the old MC6800). A terminating resistor and level-shift are required if I remember correctly, so it would be slightly more effort than a "blue wire" fix.
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
More information about the time-nuts