[ntp:questions] calculating peak to peak error

Drk Ryan ryandrk at hotmail.com
Wed Aug 27 19:40:06 UTC 2003


>>Without the kernel PLL, the time obtained will have an error
component that >>is a sawtooth shape and has a max peak to peak
amplitude of about 500us, at a >>CPU clock frequency error of 250ppm. 
A well managed system will have a >>frequency error of close to 50ppm,
in which case the peak to peak error >>should be short of 200 us.


How is the 200us calculated?



More information about the questions mailing list