[Fwd: receivers stability]

Andre' Kesteloot akestelo@bellatlantic.net
Sat, 26 Dec 1998 23:25:55 -0500

Andy Talbot wrote:

> >Andre'Kesteloot wrote:

> One question:  when you conducted your LF experiments at sub-Hz
> >bandwidth,  what kind of receiver did you use?  Most receivers I know
> >have a resolution of either 10 Hz, or 1 Hz.
> >
> >How did you  _know_ your receiver was tuned to the correct frequency?
> >Did you use a separate calibration injection (your caesium standard?) or
> >did you peer at the display until you saw the expected trace?
> Hello Andre -
> I use a Racal RA1792 Receiver which has all its oscillators derived from a
> common reference and tunes in 10 Hz steps, although later models do have 1
> Hz steps I believe.  The reference input is taken from my Caesium standard,
> which also drives a 10 Hz step conventional PLL synthesizer - actualyl 1kHz
> steps at 7 Mhz then divide by 100.   Since all drifts cancel when LOs are
> derived from the same source, using Cs accuaracy  is way Over-The-Top and a
> normal TCXO of 1ppm stability would have been good enough.   The DSP
> Downconvert / FFT software has an adjustable centre frequency, generated DDS
> style,  and the FFT span of 3 Hz is enough to capture even a badly adjusted
> transmitter.
> At the moment the CODEC clock is the only part of the system not locked to
> the master - but here the frequency error is around 2ppm at the sampling
> frequency of 8 kHz so insignificant frequency error (for now).  I did once
> try using a PLL on the CODEC xtal oscillator but it took 30 minutes to lock
> after the PLL had been booted so was too messey. It will be more satisfying
> to generate a Cs locked 1kHz tone, feed to the other input of the DSP and
> phase lock to that in software instead.
> The Cs standard  is only turned on a few times a year to calibrate its
> internal reference, this maintains around 10^-9 stability provided it is
> never turned off  which is not far off Rubidium performance.  I also have a
> frequency reference locked to the 60kHz MSF clock signal which can usually
> give better than 10^-9 short term, most of the time and 'almost infinitely
> good' in the long term as it is a National Standard.
> And just to complete the story,  I also have a backup system using TV sync
> pulses.   The BBC network is Caesium controlled, but phase changes due to
> network switching cause uncertainties in this one - 10^-8 is usually easy
> though.   TV sync is an 'official' frequency / timing source in Germany and
> they make more use of this medium - especially for the Microwave bands.
> Saw with interest the GPS derived reference in July QST but I mustn't go
> mad!!  The digital PLL in that is interesting,  my MSF reference uses an
> analogue design at just about the limit possible of this technology,  using
> a 100 Meg resistor and 40uF in its PLL filter.
> Andy  G4JNT