Re: ubxtool -- setting DYNMODEL

[John Ackermann N8UR]

On 5/8/20 5:05 PM, Gary E. Miller wrote:

> Sadly, the clock offset is less than the clock granularity if any known
> linux hardware.  There has been work on kernel drivers to echo the PPS
> pulse to an M8T or M9T, capture the interrupt latency, then apply a
> cirrection factor to the NTP idea of system time.  Only a few have dared
> venture there so far.  Eventually it is the way forward.

For my application, we don't need the PPS signal at all.  If you can
feed an external reference clock into the receiver, the PPP results give
you a clock offset in nanoseconds that you can process as phase data to
derive frequency offset, noise, etc.  You can do much better than
hardware PPS that way.  Unfortunately, though, the F9 series don't have
an external reference input.

I was disappointed in the EXTINT performance of the F9P and F9T.  The
resolution is < 10ns but there appear to be random missed inputs that
mean you can't count on getting a continuous phase record.

John