Re: [gpsd-dev] RFC 2783

[Gary E. Miller]

Yo Hal!

On Tue, 10 Mar 2015 14:17:00 -0700
Hal Murray <address@hidden> wrote:

> address@hidden said:
> > Is there always, and necessarily, only *one* pps device?
> 
> No.
> 
> If nothing else, I want to be able to connect two so I can compare
> them.

Yup, I do it often.  validate one against the other.

> The kernel time stamping also counts events.  That's a convenient
> hack for measuring things.  So gpsd and ntpd have to be sure not to
> gobble up a PPS device that isn't theirs.

If gpsd has a GPS device, it assumes it owns the matching /dev/ppsX.

If someone opens /dev/pps0 first then gpsd fails on open, but continues
without the /dev/ppsX.

So just do not configure ntpd to grab a device gpsd will be told to use.

> Here is a graph made by feeding the 60 Hz power line via a
> transformer into a PPS input and counting cycles:
>   http://www.megapathdsl.net/~hmurray/time-nuts/line/Dec-2013.png

I did that in 1975.  How all consummer digital clocks used to work.

> I think the Linux kernel supports PPS via the printer port.  I
> haven't tried it, but the printer port is an old standby for hardware
> hacking.

And also via GPIO, very common in RasPis world.

RGDS
GARY
---------------------------------------------------------------------------
Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97701
        address@hidden  Tel:+1(541)382-8588