A Rubidium frequency reference generates a 10MHz signal which is divided down by a decade counter chain and fed through a rotary switch to a BNC port. This is the frequency generator. Meanwhile, the 1Hz output from the counter is fed to a PIC microcontroller which uses it to count seconds and display the day and time. A couple of buttons and a switch allows the time to be set. It's possible to manually set the time accurately to within a fraction of a second, and after that, it won't gain or lose even a millisecond in my lifetime. Theoretically.
Thanks for your message. To be super-accurate it probably does need calibrating, but I've found it's accurate enough for me at the moment. I'm only setting it manually anyway, and I'm manually comparing it with a GPS clock and a Colorado radio clock and it has stayed right on the dot every second. For all I know it could be several orders of magnitude away from its designed accuracy, but that's fine for now.
I have a lot of experience with FEI ex-telecom Rubidium oscillators, and none with the Symmetricom ones. So I'd be interested in hearing about any differences. But one big thing I found straight away with the FEI ones is that they have excellent long term frequency stability, but their actual calibration is usually gone. So you can't just take one and use it as the timebase for a clock without calibrating it first.
Again, the Symmetricom one may be different, but I'd still want to check its frequency against (say) GPS first. If you can calibrate it, then at that point, you'd have a very stable clock.
Thanks for your message. To be super-accurate it probably does need calibrating, but I've found it's accurate enough for me at the moment. I'm only setting it manually anyway, and I'm manually comparing it with a GPS clock and a Colorado radio clock and it has stayed right on the dot every second. For all I know it could be several orders of magnitude away from its designed accuracy, but that's fine for now.