|Re: [chrony-users] How to plot statistics to show tracking accuracy?|
[ Thread Index |
| More chrony.tuxfamily.org/chrony-users Archives
On 07/05/2012 17:24, Bill Unruh wrote:
On Mon, 7 May 2012, Ed W wrote:
Hi, I'm getting very decent results with chrony and a usb gps.
According to the statistics the std dev is around 1ms, which I might
hope is slightly too high and that as low as 0.5ms is feasible? I
think that this might indicate that we can do better measuring the
arrival time of the $ in the NMEA sentence, but now I need to collect
some evidence for that
Wwhich gps card? The evidence I have seen is that the sentences have
jitter than that, and comments in the ntp newsgroup suggest it is more
10s of ms variation. Ie, the gps chipset makes no effort to make sure
sentence come out with any specific delay from the seconds turnover.
Venus 6 and other Venus chipsets have a highly accurate NMEA output. A
couple of people have now confirmed that the first bit of the $ZDA
sentence is aligned with the PPS to about the same accuracy as the PPS
(ie within the accuracy of the scope of anyone who has checked. And
anecdotally it is suggested that the rest of the bits come out with low
jitter - that in itself is interesting and could be used to effectively
have a higher freq PPS output...)
So with several venus chipsets it doesn't matter whether you measure the
PPS or the emit time of the initial NMEA sentence.
In the case of USB connected devices, everything is quantised to a 1ms
poll of the USB bus. However, it occurs to me that under these
circumstances the PPS and measuring the $ are about equal in accuracy
(1ms). However, it occurs that one might get somewhat higher accuracy
measuring the NMEA sentence because we might measure additional
characters (roughly 1-4 every 1ms poll) and how full the buffer is after
each 1ms poll and from that we might infer the character arrival time
(which is believe to be of similarly low jitter). Knowing the exact
arrival time of any character in the sentence we can work back to figure
out the arrival time of the $.
I observe quite high jitter time (1-4ms) using the output from gpsd, but
I have a suspicion that this may be due to the algorithms used in gpsd.
I need to measure externally to figure out the details
Interesting option anyway. Potentially you can get 0.5ms accuracy, and
perhaps below, using a simple $15 GPS dongle...
To unsubscribe email chrony-users-request@xxxxxxxxxxxxxxxxxxxx
with "unsubscribe" in the subject.
For help email chrony-users-request@xxxxxxxxxxxxxxxxxxxx
with "help" in the subject.
Trouble? Email listmaster@xxxxxxxxxxxxxxxxxxxx.