[ntp:questions] Re: gps; why pps?

Tim Keck timkeck at gmpexpress.net
Wed Mar 29 02:18:40 UTC 2006

Richard B. Gilbert wrote:
> Brian Garrett wrote:
>> "Folkert van Heusden" <folkert at vanheusden.com> wrote in message
>> news:20060328155232.GP26463 at vanheusden.com...
>>> Hi,
>>> Could someone please explain me why you want to only use the PPS signal
>>> of a GPS receiver and not the time indication?
>>> Folkert van Heusden
>>> -- 
>> Presumably because of the GPS receiver's tendency to delay the time 
>> display
>> anywhere from a half-second to two seconds behind the actual time.  The
>> processor's man task is calculating location and its firmware is 
>> dedicated
>> to that; by the time it gets around to spitting out time-of-day info, the
>> moment has passed.  Even using the raw NMEA data can get you into 
>> trouble if
>> you aren't clear about whether each sentence pertains to the second just
>> passed, or the one coming up.
>> Brian Garrett
> Brian,
> I don't think that's quite correct.   The processor solves a system of 
> four simultaneous equations in four unknowns: latitude, longitude, 
> height and time.  All GPS receivers can do that.  It does take a 
> substantial fraction of a second to solve those equations.
> Receivers designed for timing have maybe three special features: PPS 
> output, the ability to go into "position hold mode", and priority given 
> to reporting the time rather than the position.   Position hold mode 
> allows a timing receiver to solve one equation in one unknown since the 
> antenna is in a fixed, known, position.  Not all receivers are designed 
> for timing and not all receivers have all the features I mentioned.
> It is essential to understand the relationship of the reported time to 
> the PPS output if you are writing your own driver.  The reported time 
> can mark either the last pulse or the next pulse depending on the design 
> of the receiver.

As I recall, if SA is on, the accurate position is not known for some 30 
days; position hold involves a very slow filter to try to remove the 
affect of sthe clock variation imposed by Selective Availability.  The 
early units I played with solved the 4 equations with the postion and 
time to be valid at the next 1 second tick, but those were rack mount 
Collins units about 20 years ago.  We also had some PCI based units (I 
think Ashtech) that ran at 10 Hz solution rate, something like 15 years 
ago.  Have the manufacturers abandoned the "at the mark, the time and 
postion will be" approach for "at some time in the past, we were at" 
kind of solutions?  Sounds dumb to me.
Tim Keck

More information about the questions mailing list