[time-nuts] DC Voltage Ramp?
cfmd at bredband.net
Sat Sep 3 19:23:10 EDT 2005
From: Brooke Clarke <brooke at pacific.net>
Subject: Re: [time-nuts] DC Voltage Ramp?
Date: Sat, 03 Sep 2005 15:56:32 -0700
Message-ID: <431A2A20.3060901 at pacific.net>
> Hi Magnus:
> I was thinking of using a Datum 1000B Ultra Stable Oscillator (the one
> used in the FTS 4060 Cesium standard). The data sheet says that after a
> few years the aging rate gets down to parts in E-12/day. I would think
> that the Allan plot would be cleaner if the control voltage was well
> behaved than if it was derived from a noisy process like GPS or maybe
You have noise wherever you turn, and if you have a DAC, you need to create a
stable voltage reference for it, or else the noise, temperature and supply-
dependence of that source will ripple straight through the DAC (it will anyway)
and spoil your finetuning. The noise in the signal updating the DAC will also
go straigh thru, so you would like to spend some time to figure out what
biases you have there.
Also, you must learn and maintain the specifics of your particular oscillator.
As time goes, those parameters will change, so whatever source of reference you
have you must make comparisions and other complementary measures in order to
train and maintain the specifics of the particular oscilator in order to
free-wheel into the unknown when the reference is lost.
I would probably attempt to use a Kalman filter setup with some good models of
the oscillators as the basis.
> I think the dewar oven holds the temperature pretty well.
Yes, a dewar flask will reduce the temperature leakage, but you still want to
measure it and control it to a stable and known point. Recall that if your
temperature stabilization drifts in stable point, then the crystal frequency
shifts with it.
More information about the time-nuts