Re: PPS USB GPS recommendation

[Martin Boissonneault]

On 2020-03-30 01:42, David J Taylor wrote:
> List members,
> []
> 2. If one were to compare using the NTP servers provided by the NIST 
> vs using a PPS via USB GPS device, which would seem the better choice 
> and why?
> []
>
> Thanks in advance to everyone that reads or answers my post.
>
> Very Respectfully,
>
> Stuart Blake Tener, N3GWG (Extra), MROP
> Computer Scientist
> ===========================================================
>
> Stuart,
>
> Some primary servers (perhaps NIST) are very overloaded, and you would 
> get better results using the "pool" directive to allow NTP to use 
> multiple servers, many of which may be better performers than an 
> overloaded popular server.  You can still have an NIST server in the 
> list.

I recommend everyone keeps some of their countries official NTP servers 
in their ntp.conf (server) for sanity checking, but also use the (pool) 
directive with pools that are geographically close. Of course, that 
implies you are using a recent version of NTP or NTPsec that understands 
the (pool) directive.

> PPS over USB is likely to be /worse/ than using Internet-based 
> servers, as USB is not polled very frequently.  A direct PPS 
> connection is /far/ preferable.  Of course, if your only choice it 
> would be better than nothing, and would stand alone in the case of 
> Internet loss.
>
> PPS only takes one wire from the GPS direct to the GPIO pin or DCD pin 
> on an RS-232 port (possibly via a level converter).
>
> Cheers,
> David

Also, if you use a GNSS receiver with PPS, (tos minsane) default of 1 is 
OK (as if Internet disconnects, your GNSS is the only time source), but 
if you rely on internet servers only, consider using something like (tos 
minsane 4) or higher. As written in the documentation, you need 4 to 
kick out 1 falseticker.

Good success!
Martin VE2MRX
Pi 3B+, NTPsec, GPSd, u-Blox MAX-M8Q