The example shows the typical behavior of TV-PLL
synchronized units of this design.(like designs by DK2DB and
DK6RX ).
Course of
the reference hardware at the 27th of August 2001. Starting 5
seconds after switch on, until almost 2 hours behind.
The shown precision is a calculated average and can not directly
be used ! For an user of such HW only the slope of the current
graph is important. Only that slope is an image of the current
precision.
The worst case here is better than 1*10exp-8.
This is a rather good value for a regular counter.
The average about 4 hours shows a precision
of about -4*10exp-11. The problem however is, that the usable precision
for a counter varies between -1*10exp-8 and +1*10exp-8.
Rubidium vs. GPS
The reference signal was taken by GPS and the oscillator
under test was a 10 MHz Rubidium Standard Frequency Generator
by Rohde&Schwarz.
The top graph shows the course of short-term
stability of the Rubidium Standard and the calculated Accuracy(green line) over 12 hours.
The mid graph shows the same
result, however the source is an IIR filter which causes a much
more smooth graph.
The bottom graph shows
the noise of the GPS reference and the Rb-Standard from second
to second. The resolution is around 10ns.
3-SAT Doppler
The screen shot below explains why a TV-Satellite
based reference can not be a reliable reference source.
The picture shows the precision of
the 3SAT TV-Satellite within 24 hours. Due to the motion of the
satellite, the usable precision changes dramatically per day.
2 times a day the precision is excellent (gradient is zero).
But 2 times a day the precision is rather lousy. Between that
both extreme values, any precision can be guessed. The Doppler
effect is responsible for that behavior.
GPS-Receiver
Differences between TIM-LP and MS1
Both GPS-Receiver can be recommended
for the CRO as a 1PPS reference source. However, the MS1 is better.
The picture on the left shows the differences.
The MS1 has a 1PPS jitter of +/- 30
ns, the TIM-LP has jitter of +/- 40 ns. Also it seems that the
1PPS signal from the TIM-LP is more restless.
Important information about TIM (not TIM_LP!)
The current TIM firmware (E010) has
a bug. The bug causes that the 1PPS pulse jumps any (around)
35 minutes for a certain amount of
time. It can be seen in the Online-Analysis SW as a recurrent
disturbance (blue marker) after the mentioned time interval.
u-Blox knows the bug and they promised
to fix it in the next SW release.
From the jitter point of view the TIM
seems to be a bit better than the TIM-LP .