[time-nuts] M12+ phase jump?

James Maynard james.h.maynard at usa.net
Sun Apr 23 20:58:13 EDT 2006


I Might be seeing the same thing you (Ulrich, Ernie, Brooks) are seeing, 
although I would not attribute it to WAAS or EGNOS, as I am not using a 
receiver that is enabled for differential corrections.

I am using the 1 pulse-per-second output from a Motorola M12+ Timing 
receiver (M12+T) to disipline my 10 MHz frequency reference, which is a 
Stanford Research Systems (SRS) model PRS-10 rubidium standard. (The 
M12+T is mounted together with its antenna on a pole above the roof of 
the apartment building in which I live.)

I have been observing the phase diference between the M12+T pulses and 
the PRS-10 10 MHz output (counted down to 1 Hz) using a SRS model SR620 
time interval counter.  The SR620 is set to display the means and 
standard deviations of N-sample "measurements" (teminology from the 
SR620 instruction manual) on an oscillosope. The oscilloscope shows a 
graph of most recent 250 of these 600-sample measurements. I have set N 
= 500, so 250 measurements of these measurements amounts to 250*500 = 
125 000 s, or somewhat longer than a day.

The jitter (stamndard deviation) of the 500-sample measurements is 
consistently about 10 ns.

The time interval between the two 1 PPS signals (that is, the phase 
difference between the two 1 PPS outputs -- one from my PRS10 house 
oscillator and the other from the M12+T GPS receiver) varies over the 
course of the day.  For example, over the last day or so there were 
maximum and minimum phase differences as follows:

measurement # 238:           +60 ns (+59.797 ns)
measurement # 302 (250+ 52): -48 ns (-47.799 ns)
measurement # 409 (250+159): +70 ns (+70.250 ns)

The interval between the two maxima of the phase differences is about 
(409-250)*500 s = 86500 s, or about a day.

This suggests to me that the M12+T output might be exhibiting two phase 
jumps each day, one positive and one negative.  These phase jumps would 
perturb the PRS10's internal phase lock loop (PLL) that disciplines the 
PRS10 to agree with GPS time. (I have set the PRS10's PLL "PT" parameter 
to 9, which gives the PLL a time constant of about 3 hours. I am 
reluctant to set the PLL time constant to a greater value until I have 
an M12+T with Rich Hambly's hardware sawtooth correction to the 1 PPS 
output.)

To view this phenonemon in more detail, I really should write some 
software to interface with the PRS10 and record the time stamps of the 1 
PPS pulses from the M12+T.

I'm also trying to scrape the money together to buy one of Richard 
Hambly's "CNS Clock II" receivers with his "High Performance PPS" 
options that provides a hardware sawtooth correction to its M12+T's 1 
PPS output.  That will probably reduce the jitter I see over a 
100-second interval from about 10 ns to something less, perhaps 2 or 3 
nanoseconds. It would also let me set a longer time constant on the 
PRS10's PLL by which it synchs its 10 MHz oscillator to agree with GPS time.

However, upgrading my M12+T to one with the hardware sawtooth correction 
on the 1 PPS output would not, I think, have any effect on the 
twice-per-day phase jumps that I think we are observing.

-- 
James Maynard, K7KK
Salem, Oregon, USA






More information about the time-nuts mailing list