Future Directions

What we can do

Although the GPSClock 200 provides reasonably good time for many applications, it is possible to attain much greater accuracy, both in the GPS UTC second pulse and in the computer to which the GPS timing device is attached.

The GPSClock 200 itself is not quite as accurate as some other GPS devices now available that are more well-tuned for timing applications; it only guarantees PPS accuracy to a few hundred nanoseconds whereas the Motorola M12+ Timing unit, for example, advertises accuracy to a few nanoseconds. Even this can be improved upon, though, by using the GPS tick to discipline a local ovenized oscillator. A good ovenized oscillator typically has a rated stability of 1x10E-10 (or better) in 24 hours. That is, a stability of a few tens or hundreds (at worst) of picoseconds, and possibly better, depending on the oscillator.

Of course, we can't deliver picosecond accuracy to a network; the electronic components alone will introduce jitter on the order of nanoseconds, but it is certainly possible to do much better than the few hundred nanoseconds the GPSClock 200 gives.

Despite a time source of such high accuracy, though, we are limited by the jitter in the clock on the computer that talks to the GPS device. This jitter can be on the order of microseconds and is often at least on the order of a few hundred nanoseconds. We can improve this by controlling the environment of the system, but the only really effective way to significantly reduce that jitter is by replacing the low-stability oscillator of the computer with the (multiplied) output of the GPS-stabilized oscillator above. This will give us both a very stable pulse (in the long term from GPS and in the short term from the quality of the oscillator itself) as well as a very stable system clock with which to count the time between ticks.

The question then becomes, "How much will all this cost?"

Where we're going

There are several large companies that produce very highly stable time servers. These devices typically cost a few thousand dollars, at least. I think I can provide something for significantly less.

This is a work-in-progress. Please check back often for further updates.

Copyright © 2003 GPSClock.com
Comments? Questions? Criticisms? Contact Us