Re: [gpsd-users] Garmin 18X-5Hz

[Gary E. Miller]

Yo Robin!

On Sun, 14 Aug 2016 21:18:04 +0200
Robin Schwab <address@hidden> wrote:

> Am 14.08.2016 um 19:42 schrieb Gary E. Miller:
> > You are using the 22 driver, and I, and the howto's, use the 28
> > driver.  Once again the 22 driver usually works, but it lacks the
> > fail safes in the 28 driver chain.  
> 
> Not being a software engineer I am confused by the many drivers that
> are around. 90% are probably obsolete but not marked as such.

Yup.  The ntpd driver thing is confusing.  NTPsec has been ripping away
at old drivers with a vengence.


> Or am I wrong and there are still thousands of Trak 8820 users around?

Not relevant.  Driver 22 is for generic PPS.  One of the reasons gpsd
added a PPS driver years ago was due to the deficiencies of the ntpd PPS
driver.  Current plans are to eventually rip it out and only support PPS
by way of SHM, driver 28.


> Which driver shall one use? 20,22, 28, and 46 are candidates for a 
> modern setup with NEMA and PPS.

I use 28.  gpsd does all the work and hands the time corrections to
ntpd to deal with.  That is how the gpsd time server howto does it, as
well as the in progress NTPsec RasPi howto.

    http://www.catb.org/esr/faqs/stratum-1-microserver-howto/

> But I do not find information which
> one to prefer and which NEMA sentences they need to work best.

See the above two references.  Don't worry about which NMEA sentences, as
long as your port speed is fast enough to keep up.  You only need to
NMEA to tell the current second and if the fix is valid.

> I.e. a
> driver could use DOP data to trust good fixes more than average ones
> and to even discard very bad fixes.

Most GPS manufacturers list the conditions under which you can trust their
PPS.  gpsd trusts the PPS if the last 4 fixes have been valid, which it
tighter than most any GPS insists on.  Never heard of any issues with that.

> That is probably what you mention
> as fail safes Gary.

No, I mean hard fail.  Like loss of signal, no PPS output, so ntpd goes
free running.  Or worse yet, just a wild hare output., like 14 Aug about
16:00 UTC here: https://pi2.rellim.com/day/

Tht one happens every few weeks, driving me crazy to find and fix.

chronyd was designed to handle loss of refclocks.  The authors
considered intermittent uptime and intermittent fixes.  It has
techniques to adjust for them and converges fast.

ntpd, by design, is for constant on operation with multiple inputs.  It
converges very badly and slowly on startup, it uses voting to select
averages of the best times.

If you are only going to have one input you should be using chronyd.

> As far as I know, diver 46 is the most recent and
> the only one to consider DOP.

 Are the howtos out of date?

ntpd goes back to the 1980's!  Take your pick, a lot are up to date, a lot
are not.

Right now I would say that the NTPsec one is the most up to date and
actively maintained one, even though it is not even released yet.o

Dom't trust any howto that does not have proof of how good their
technique works for them.  Like this:
        https://pi3.rellim.com/day/

If you are trying to nail it down right, you need to be able to tell the
good advice from the best advice.

RGDS
GARY
---------------------------------------------------------------------------
Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97703
        address@hidden  Tel:+1 541 382 8588

pgpnzh25qOYU6.pgp
Description: OpenPGP digital signature