On 02/01/14 18:10, Gregory Muir wrote:
Reading all of this brings back memories of a project I was involved in back in
the early 70's in the Denver area. NBS-Boulder was experimenting with
injecting their time standard into the video of the analog signal before it hit
the transmitter. We installed a prototype unit that was slaved to Boulder and
performed the function of adding the time signal into the video. Part of it
also provided station sync slaved to the unit as well. NBS then measured the
system prop delay and adjusted the timing accordingly to compensate for any
latency before it went to RF. The experiment lasted for a couple of months and
then the equipment was removed.
Alas... satellites and digital... given buffering, processing and routing
delays, you are sort of now on your own in this one-bit world.
Just to help confusing matters even more, SMPTE-12M time code (which is
what you can suspect to be in the TV-signal) and it's Drop-frame
algorithm causes a drift of time over the day, as the drop-frame
mechanism isn't perfectly aligning up to 30000/1001 frames per second
over the 86400s day. This requires the production-time to be jammed into
alignment regularly, such as every day (off-hour). With the evolving
standards, the halting mechanics of drop-frame correction is not
changed, but just standardized. The jamming mechanism is also used for
leap-seconds and DST change-overs.
So, in the US and other 30000/1001 frames per second countries (formerly
NTSC), encoded time is not going to be useful for precision work. For us
in the 25 frames per second world, we only need to jam for leap-seconds
and DST change-overs, but that is enough of an upset, but can be more
easily predicted with only a few handful of bits extra information.
I prefer using MLS measurement for audio delay measurement. If you do it
right, you get 20,833 us step resolution, as a result of the 48 kHz
sampliung clock. MLS delay measurement is trivial using the Analog
Precision test-set.
Cheers,
Magnus
_______________________________________________
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.
