[prev in list] [next in list] [prev in thread] [next in thread] 

List:       gpsd-users
Subject:    Re: [Gpsd-users] deluo gps ntp time offset and strangeness
From:       "Chris Kuethe" <chris.kuethe () gmail ! com>
Date:       2007-02-19 0:34:21
Message-ID: 91981b3e0702181634xf6363a9s907d16e18668f67a () mail ! gmail ! com
[Download RAW message or body]

On 2/16/07, Kerry Hoath <kerry@gotss.net> wrote:
> I admit to switching it into SiRf Binary; afterall the device is cheap, I
> have a screwdriver and don't mind resetting the unit.

That's fine. Gpsd can switch it back to NMEA with the N command,
gpsctl should be able to kick it back as well.

> This is one of the little square gps units on a cable; see
> http://www.deluoelectronics.com/ and seems to contain the Sirf3 chipset.

I'm told that the SiRFstar 3 is not the optimal choice for precison
timing applications. For +/- 1 second though, it should be fine.

> I have configured ntpd to take time from gpsd 2.34-1 and the peer billboard
> looks as follows:

It's been ages since I last used ntp.org as my time daemon.

> Now currently the time offset is 14 seconds, however I have seen the
> difference be as large as 80 seconds.

14 seconds is understandable - leap seconds. Not sure where 80 is coming from.

> At night I seem to get 6 satelites and a reasonable fix however in the
> daytime the fix can fail completely or drop to 1 or 2 satelites.
> the unit sits on a window cill in our block of units; so perhaps the view of
> the sky plays a part but I am wondering why I would see time errors
> as large as 80 seconds?

Yes, I also live in an apartment with a GPS in the window; around here
I'm lucky if I get 4 satellites; often I get 2 and a ton of multipath
error.

I'm wondering if maybe it's something in SHM. I seem to recall hearing
that running gpsd at debug level 5 will give some insight into the
workings of SHM.

> I'm happy with 5 seconds either way worst case sanario but 80 seems a bit
> wrong.
>
> Running gpsprof -f cycle yields:
> ...
> I don't know Python but it seems as if gpsprof gets values that it does not
> like which are out of sequence.

I don't know python either. ;)

> I read in the gpsd manpage that running in a binary protocol at reasonable
> baudrate would improve accuracy or could do so.

It can reduce certain transmission delays, and running in binary mode
means you can get a full navigation solution in a single packet,
rather than having to reassemble it from NMEA sentences.

> The unit seems to return reasonable possition information when I query it
> with the p command, and as previously stated q shows fixes between 3 and 6
> satelites.

With 3 satellites you don't get a real 3d solution - you can make some
good guesses, but any receiver intended for timing use I've ever seen
insists on having at least 4 satellites.

> Is there anything I could try to potentially improve accuracy to below 5
> seconds or to stop the unit from loosing fix? The gps time source appears to
> occationally become unavailable which causes the server to lock onto another
> time source and given the time differences stay locked onto that source.

> Am I expecting too much from a cheap SiRf3 unit? Should I go out and buy a
> Garman or similar?

No, you're not expecting too much to be able to get time within a few seconds.

> I also have a couple of other gps units; wondering if it is worth trying one
> of these.

Go for it... it can't hurt.

> I am happy to buy another gps unit for a time source but am trying not to
> use a bluetooth unit if possible since I don't really want the 2.4 ghz radio
> noise in the unit block and a cable seems more reliable potentially.

Heh. Bluetooth has its own issues, notably the tendency to not switch
port speed correctly. I prefer wired also.

> Any help or insight would be greatly apreciated. Happy to provide any debug
> dumps or similar information including switching gps to NMEA or similar.

telnet to 127.0.0.1 port 2947 and send "N=0"

CK

-- 
GDB has a 'break' feature; why doesn't it have 'fix' too?
_______________________________________________
Gpsd-users mailing list
Gpsd-users@lists.berlios.de
https://lists.berlios.de/mailman/listinfo/gpsd-users
[prev in list] [next in list] [prev in thread] [next in thread]