Closed krytie75 closed 7 months ago
You're right that it's probably not needed due to noselect
. It's (if I remember correctly) an approximate estimation of the error of the serially transmitted time, which would be around 100-200ms, and by declaring that, it's another way to say: this is a 'bad' source on it's own. So, if you would try to synchronize a clock solely based on the serial connection to the GPS device, that's the order of magnitude of error you would end up with.
In terms of the accuracy of your system: that delay
value does not matter! The only thing that needs to be correct is the offset
value, which needs to be tuned so that the overall offset between PPS pulse and serial time from GPS is within the required boundary of 200ms (this has nothing to do with 0.2 delay value, it's coincidence that the same value appears).
The quality of the PPS signal (which is directly dependent on the number satellites the GPS receiver 'sees') is the determining factor of the overall precision and accuracy of your system.
That's really informative and more or less what I was hoping to hear.
Many thanks.
Excellent guide, very well written and extremely helpful, thank you.
I'm trying to tweak everything to get the most accurate time on my system and I'm a little confused as to why we include the delay of 0.2 in the following line from chrony.conf:
refclock SHM 0 refid GPS precision 1e-1 offset 0.01 delay 0.2 noselect
Chrony docs describes the 'delay' option as:
Presumably we're not doing this to influence the source selection as we also have 'noselect' defined for this reflock so I was wondering why we include the option and how you arrived at the value 0.2.