[time-nuts] GPS accuracy specs

Said Jackson saidjack at aol.com
Sun Feb 16 18:53:29 EST 2014

There are two numbers that need to be specified, one is relative (back and forth wander around some fixed offset) one is absolute (how close to the real UTC pulse is it?).

Most of the time relative stability is given which is the first number, and one has no idea how much unit to unit offset and offset to UTC there is. Unless you sit at the USNO or NIST office there really is no way to tell how much offset to the real UTC your receiver has. 

Please note that almost all GPS receivers I know claim to generate UTC time not GPS time. There can be up to 100ns offset between the two but lately the offset is held at around 10ns or less.

Having access to a lot of GPS receivers, I can say that the Motorola M12M is very close to the ublox receivers (better than 20ns on typical units), and a typical ublox to ublox offset would be about 5ns on the timing units with SBAS in position hold mode.

The days where GPS receivers only had 1us stability or accuracy are long gone and today even typical navigation receivers achieve better than 25ns accuracy and stability.

Can you get down to 2ns stability? Supposedly with the M2M using sawtooth correction. But who knows what the offset to UTC is on late model iLotus M2M..

Not an issue since most receivers have antenna delay correction features that lets a user fine-tune the UTC offset to some reference if that is desired so you can tune a batch of GPS to within 1ns of each other on the bench.


Sent From iPhone

On Feb 16, 2014, at 13:31, "Richard H McCorkle" <mccorkle at ptialaska.net> wrote:

> Jim,
> Generally navigation receivers don't include survey and position hold
> features so the time solution accuracy is typically about +/- 1us.
> Timing receivers survey their position over a large number of samples
> (typically 10,000) and go into position hold mode once the survey
> completes. The fixed position allows higher accuracy in determining
> the time solution, typically to +/- 1ns. However the 1PPS output is
> placed on the nearest GPS clock edge, typically derived from an XO,
> so the pulse placement resolution is limited by the GPS clock period.
>  The GPS XO clock drifts so the 1PPS placement also drifts over the
> clock period, creating a "sawtooth" like displacement in time over
> the GPS clock period. With a receiver like the M12+ the placement
> varies roughly +/- 12ns for a 25ns 1 sigma 1PPS accuracy. For better
> accuracy the M12+ also includes a message with the predicted 1PPS
> placement error of the next pulse to the +/- 1ns time calculation
> resolution. The combination of the 1PPS placement to the nearest
> clock edge and the sawtooth correction message giving the placement
> error allows resolution of the GPS time to +/- 1ns using either a
> software correction of the sample data or hardware correction of
> the 1PPS pulse using a variable delay.
> Richard
>> I've looked at several different manufacturer GPS datasheets now regarding the 1
>> PPS output in an attempt to compare apples to apples. Some of them rate their 1 PPS
>> output as something on the order of "PPS signals have an accuracy ranging 10ns"
>> which seems ambiguous. Does that mean the leading edge of their 1PPS is within 10ns
>> of the GPS clock? Or simply that the stability of their 1 PPS is within 10ns? Or
>> both?
>> Perhaps there's an industry standard for these specs of which I'm unaware?
>> The datasheet for my (presumably much older) Globalsat ER-102 seems, to me at
>> least, to be much more clear stating "time reference at the pulse leading edge
>> aligned to GPS sec., +/- 1 us". Which I interpret as the leading edge of my
>> receiver's 1PPS is aligned with the GPS's clock to within +/- 1 us.
>> Jim...
>> N5SPE
>> _______________________________________________
>> time-nuts mailing list -- time-nuts at febo.com
>> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>> and follow the instructions there.
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.

More information about the time-nuts mailing list