Re: [gpsd-users] Acquiring PPS on mini PCI-e

[Jean-Michel Pouré - GOOZE]

Le jeudi 04 septembre 2014 à 00:53 -0700, Gary E. Miller a écrit :
> Yo Jean-Michel!
Dear Friend, Hi!

> On Thu, 04 Sep 2014 09:22:38 +0200
> Jean-Michel Pouré - GOOZE <address@hidden> wrote:
> 
> > GOOZE would like to release a cheap and interesting mini PCI-e card
> > with GPS and 1PPS output. We are discussing with the developers and
> > they propose to wire the PPS to a pin on mini PCI-e, when we propose
> > to use the UART to USB bridge. 
> 
> If you connect to a pin on the mini-PCIe you will need to write your
> own Linux/Windows drivers.  If you connect to the USB/UART you can likely use
> existing drivers.

Yes, this is the point. We wonder if it makes sense to develop a new
driver and/or extend kpps, when it took months/years to have it included
in Linux kernel. Let us not say other OSes.

> > What is the preferred way in your opinion?
> 
> A dedicated pin would be more accurate, but the USB/UART would be plug
> and play.  The USB/UART, using USB 1.1, would be around 1 milliSec
> accuracy and a dedicated pin could be 1000x more accurate.
> 
> A good compromize would be if you could use a USB 2.0 UART which would give
> you almost as much accuracy as the dedicated pin.

The issue is also : does it make sense with interrupts of a PC to drive
down to such a precision? Using Embedded hardware, probably.

Just a simple question to drive our decision: how precise can be PPS
pulse in theory according to GPS? From the reference documentation of
the chip, it is written : ±11ns = 11 x 10-9 s = 1,1 x 10exp-8 s. 

So it probably makes sense to use mini PCI-E, right ?

Kind regards,
Kellogs