[ntp:questions] Choice of local reference clock seems to affect synchronization on a leaf node
unruh at invalid.ca
Tue Nov 8 00:23:37 UTC 2011
On 2011-11-07, David Woolley <david at ex.djwhome.demon.invalid> wrote:
> unruh wrote:
>> Another approach might be to use the PID concepts ( in which one uses
>> the present offset, the derivative of the offset and the integral of the
>> offset to drive the correction) to control the clock to get faster
> ntpd uses at least two terms.
Which two? As I recall, ntpd determines the current offset, it then
alters the frequency so that in future that offset will be smaller. The
exact coefficients used (how does the current offset alter the
frequency) determines the convergence factor, and is adjusted to more or
less make the convergence be that of a critially damped system.
The PID controlers use the current offset, the difference between the
current offset and the previous one, and the integral of the offsets
with some exponential damping factor to determine the change in
frequency. Now in the cases of those heaters, it does tend to be an
on-off system ("frequency correction is either large-- heater on-- or
0-- heater off) which is certainly not what one wants to do in the clock
case. How much of the advantage of PID systems comes from that
limitation of the control system I am not sure.
More information about the questions