Re: Xtal oscillator injection locking / Pn effects

Neil Smith G4DBN

If the osc has low PN and is free-running close to the locking frequency and the injection level is very low and through a resistive divider, I've found that the phase noise of the osc output is *much* lower than that of the locking signal.  As the level is increased, the pull-in range increases and the output does indeed start to look like the PN of the reference.

If the level is too low, the signal squegs about between the free run and lock frequency until the level is just high enough to lock it.  There is a fair bit of hysteresis as well.  Once it is locked, you can drop the level again by a dB or two and it stays locked but then lets go and squegs again.

Pulling a 117MHz crystal by 6Hz or so with -30dBm or so of locking signal had almost no discernable effect on the output spectrum, but pulling it 40Hz with -5dBm made it look pretty much like the spectrum of the injected reference.

I don't think I have the spectra saved anywhere, must repeat the tests.  That was into the osc of a number of DB6NT transverters.  Gut feeling is that the osc in a G4DDK Anglian locks at a much lower level than the DB6NT and the output is less affected by the PN of the injected signal, possibly because the injection is on the colder side of the xtal and it somehow cleans up the PN a bit more than the DB6NT topology, but I don't have any evidence for that.  Must do more tests.

Neil G4DBN

On 06/08/2019 12:18, alwyn.seeds1 wrote:
Dear All,

Within the locking bandwidth the locked oscillator takes on the phase noise characteristics of the injected signal.



Alwyn Seeds, Director
114 Beaufort Street (Management) Company Ltd.,
114 Beaufort Street,

+44 020 7376 4110

114 Beaufort Street (Management) Company Ltd., Registered in England and Wales: No. 02797775 Registered Office Address: 114 Beaufort Street, London, SW3 6BU

Join to automatically receive all group messages.