[ntp:questions] New to NTP on Linux

Dusty Bin junkmail at bitbucket.mmscomms.demon.co.uk
Tue Nov 2 09:56:50 UTC 2004


On 30 Oct 2004 13:46:59 -0700, mayer at gis.net (Danny Mayer) wrote:

[...]
>> 
>> I did attempt ref clock type 20 on W2K but it didn't work - I didn't
>> try very hard because the Windows info on ref clocks suggests only a
>> couple have been implemented.  The Garmin was on serial port 1 and I
>> think I tried both 127.127.20.0 and 127.127.20.1 without success.
>>  
>The NMEA clock driver appears to be supported on Windows. I have no
>idea why it didn't work for you. Please note that on Windows there
>is no COM0. COM starts at 1. You need to configure your devices for
>the appropriate COM port number.

Thanks Danny - you are quite correct!  After I got NTP running on
Linux (without nanokernel), I went back to see what W2K would do - it
was working (now NTP v4.2.0 in both cases)...
As I said, I didn't try very hard last time, so probably I hadn't
programmed the Garmin unit to sent the compatible NMEA message
format.
It is currently working with 127.127.20.1 and COM1 as you suggested.

The issue with both Linux and Windows is that the timekeeping is
rubbish (about +/- 80mS over periods of a few days, and time1 set to 
around 0.6 seconds).  Before attempting to complicate matters with Linux 
nanokernel, I want to understand what is going on and how 1PPS is used 
(or not, as the case may be).
I have become very confused about PPS commands and whether my current
set up is actually using the 1PPS signal at all.
- 'enable PPS' is mentioned as a configuration option, but appears to be
related to an obsolete PPS i/f (tty_clk).
- 'PPS' command is mentioned but I didn't find a description of this
command.
- 'fudge ...  flag3' is mentioned, but also appears to be obsolete.
- 1PPS support is stated to be incoporated into the standard drivers,
so I presume I don't need to use the clock type 22 in addition to 20.

Certainly Windows didn't want to know about clock type 22 at all and
under Linux it just remained with reach 0.

Looking at the other recent thread that concentrates on the Oncore, 
solution it is mentioned that there are other links to the serial port
required and other configuration files to be used in order to get PPS
working - so I am wondering if there is a similar requirement with the
NMEA clock driver?

Can anyone help me put this PPS jigsaw together, or is it a matter of
ploughing through all the source code to work out what is going on and
what is needed?

I realise that the Garmin GPS25LVS isn't ideal, but it would seem
suprising that the 1PPS signal is actually as inaccurate as indicated
by my current results.  From a hardware point-of-view, the 1PPS signal
is connected to DCD and enabled, but has no level conversion to RS232
- so, although this is unlikely to cause a problem, it may be
something I need to check just to be sure.




More information about the questions mailing list