[Thumbgps-devel] Effect of baud rate USB GPS / NMEA wandering effect / GPGGA

[Ron Frazier (NTP)]

Hi guys,

I thought I'd pass along some random info from my testing of my 
GlobalSat BU-353 USB only GPS.  Some of this has been discussed on the 
NTP questions list, and I thought it might be useful to you.

A) I have a nice example of (apparently) NMEA wandering here:

http://dl.dropbox.com/u/9879631/drifting01-peerstats.20120312.jpg

This graph shows my pc locked to the GPS, which is the jaggy dark line 
centered around zero vertically.  The band of colored lines is a group 
of internet servers I'm watching.  You can clearly see them wandering 
away from zero.  Then, in the middle, my GPS has a heart attack, and the 
offset of the colored lines dramatically shifts.  Then it starts 
wandering again.

So, one of three things is happening:

    1) all the internet servers are wandering
    2) my gps is wandering
    3) neither is happening, and this is a reporting error in the 
peerstats file

#2 seems the most likely.

B) The graph above was taken with the GPS set to 57600 baud.  I don't 
really even know what that means in the context of a USB connection.  I 
did that to reduce latency and perhaps jitter.  However, it may also 
reduce the stability of the GPS.  I've had two instances where my system 
clock suddenly jumped to about 52 seconds off.  Cause unknown.  I 
reduced the baud rate to 9600 as a test.  Results are preliminary, but 
it looks like the unit may be more stable and may wander less.  It 
didn't seem to affect the jitter much.  Regardless of how the data gets 
through the USB, once the baud rate was reduced, I had to adjust my 
offset fudge factor by about 90 ms to get the reported GPS time to agree 
with the internet servers I'm monitoring.

C) I believe I have recommended using the GPZDA sentence here before, 
and indeed, the manual for the Trimble Resolution T also recommends it 
for timekeeping.  However, David Taylor and I got into a discussion on 
this on the NTP Questions list.  He pointed out that the GPZDA doesn't 
have any validity check field where as GPGGA does.  We looked in the C 
code for NTPD's refclock implementation, and found out that it does 
check said validity field.  I don't know what it does with the data.  In 
any case, I decided to revert my system back to using GPGGA.  
Theoretically, this should allow a more graceful recovery in case the 
GPS fails.  In my experience, GPGGA does produce slightly more jitter 
than GPZDA.

Sincerely,

Ron


-- 

(PS - If you email me and don't get a quick response, don't be concerned.
I get about 300 emails per day from alternate energy mailing lists and
such.  I don't always see new messages very quickly.  If you need a
reply and have not heard from me in 1 - 2 weeks, send your message again.)

Ron Frazier
timekeepingdude AT c3energy.com