Re: [gpsd-dev] PPS over USB

[Gary E. Miller]

Yo Ed!

On Mon, 07 May 2012 15:52:07 +0100
Ed W <address@hidden> wrote:

> > The S1 is my personal, at-home, FreeBSD+Garmin18 ntpd server.
> 
> Aha! Then in that case I definitely agree you are sorted!  I would 
> presume you can achieve low few 100s micro second sync (or better?) 
> quite easily?

My experience is that a host adjacent to S1 on a GigE link can sync
to better than 100 microSec jitter.  Variable depending a lot on 
the load on the hosts and the net.

> Problem seems to be that unless you all have the same
> reference (eg gps) then you could all be microsecond "accurate", just
> all at an offset to each other?

Exactly, thus the preference for a single time source for all units
in the test.

> I might guess that this is the goal of thumbgps - get the offset to 
> zero?

Plus or minus a handfull of  milliSec.

> However, it appears in principle that you might achieve almost 
> the same simply syncing to a carefully chosen pool of stratum 1 
> servers?

Chicken and egg problem.  How can you get good time over a network
link of unknown (and asyn) delays?  Thus the observed data that it
can be pretty hard to sync to an S1 over a cable modem to even
100 milliSec.  100 times worse than the GR-601W locally.

>  So I'm still a touch unsure what the ultimate reason is for 
> this?

To measure and understand the effects of bufferbloat on network
latency.  Since this is off topic for gpsd I suggesst you check out
the bufferbloat project html pages and mailing lists.

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

signature.asc
Description: PGP signature