CXR Larus 80-100-400
Issue 1, July 2006
______________________________________________________________
____________________________________________________________
5−
26
measurement logic. The calculation is performed as follows:
a. The clocks start out by taking a phase measurement, which is the average
phase difference between the clock's output and the reference input. This
first reading acts as a signpost for the algorithm. The clock will attempt to
keep the output close to the input phase as represented by this first
reading.
b. Another reading is taken and compared with the first. This difference is
the error in phase between the input and the output. This first difference
may be seen as representing the position error of the phase.
c. A calculation of the difference of two successive readings is taken. This
calculation expresses the rate of change of the phase and may be seen as
representing the velocity error of the phase.
d. The first difference is multiplied by a scale factor and added to the second
difference multiplied by another scale factor. This is called the correction
factor. The correction factor represents the scaled error from the correct
phase as well as a value that expresses how fast the correct phase is
being departed from or approached.
e. The control value to be sent to the DDFS is then calculated by taking the
last long-term average of the control values, adding the latest successive
average of the control value and the correction, times the damping value,
less the last control value, and dividing by the long-term average sample
number.
f. The long-term average is formed by taking old control values and
averaging them into another damped response summation to yield the
long-term average control value.
g. The algorithm returns to take another reading and repeat the above. If,
before the control value is to be updated at the DDFS, there has been a
condition that requires the clock to go into holdover, the control value is not
updated from its last value and the unit goes into holdover. Once the condition
has cleared, the algorithm proceeds from the top by taking a new first reading.
5.7013 Thus the value used for control prior to holdover represents a damped
average of the short-term corrections and the long-term offset average. The
time taken for this process is about 2 minutes for the Stratum 3E/LNC clock
and about 15 minutes for the Stratum 2/TNC clock. Complete stability and
convergence may take as long as 4.5 hours for the Stratum 3E/LNC and 32
hours for the Stratum 2/TNC. These times are based on the worst case
long-term average offset, which is not normally present. Typical times are 30
minutes for the Stratum 3E/LNC clock and 3.5 hours for the Stratum 2/TNC
clock.
5.8 Model 54591 GPS Stratum 1/PRC Track and Stratum
3E/LNC Hold Clock Card
