Observing teh ammonia resonance


Andy G4JNT
 
Edited

On the Timenuts Group there's currently a discussion on, amongst other things, the ammonia resonance and its possible use as a simpler frequency reference.
There is a narrow resonannce of the ammonia molecule at 23,870,129,007 Hz +/- 10 Hz  (ie just below our 24GHz band)
 
The resonance will be a Hz or so wide, and it ought to be possible to see this with relatively simple equipment.    First of all, get a length , several metres, of waveguide for 24GHz.   This needn't be 'proper' WG, just squashed water pipe will do to allow just the TE01 mode to propagate.    Fill this with ammonia (obvs. make the transitions gas tight)
 
Use an ADF5355 synth, and drive its reference from a 48 bit DDS like the AD9852.   Anything to be able to tune the 24GHz signal in sub-Hz steps, and to know what it is.   That chip will leak enough 4th harmonic from its VCO; launch this into the waveguide.   At the other end stick a 24GHz receiver tuned suitably.
Vary the input frequency a few Hz either side of the resonance and look for a dip.
 
One snag - if you use GPS locking, the residual variation of the GPSDO will cause it to drift across the resonance over a few tens of seconds.
 
Work out a way of feeding this dip back and controlling the synth reference in a locking loop and you have your own atomic frequency standard.   It's potentially  the same performance as cesium 
 
It seems in years gone by that caesium was possibly adopted instead of ammonia, simply because its resonance at 9.19GHz was easier to generate using ex-WW2 equipment that was 23GHz back in the 1940s when the first Cs standard was developed.
 
Andy
 


Robin Szemeti - G1YFG
 

A method of tracking the "dip" used in some other absorbtion resonance sensing systems (some HP rubidium vapour references I worked on at one point for example ) is to wobble the excitation frequency a little, say at 50Hz FM, with a deviation of a few Hz ... demodulate the output af the absorbtion cell in an AM demodulator ... if you are up the low side of the curve the demodulated 50Hz tone will be 180 degrees out of  phase with the driving signal  (as the frequency goes higher, more is absorbed, less output) .. if you are on the high side of the curve, the tone will be in phase with the driving signal ... this allows you to steer the centre frequency to the centre of the dip with a simple mixer circuit.

A further refinement to make sure you are centred on the dip can be to look at the 3rd harmonic of the demodulated signal. The FM modulated RF signal will be on the "cusp" of the absorption dip. This will produce a rise in the 3rd harmonic content of the AM demodulated signal, HP used a circuit to monitor the level of the 3rd harmonic and accurately keep the signal centred on the peak of the absorbtion band. I forget the precise details, but it was quite clever, and centred the RF within a few fractions of a Hertz on the rubidium absorbtion band.


On Mon, 20 Sept 2021 at 19:05, Andy G4JNT <andy.g4jnt@...> wrote:
On the Timenuts Group there's currently a discussion on, amongst other things, the ammonia resonance and it's possible use as a simpler frequency reference.
There is a narrow resonanance of the ammonia molecule at 23,870,129,007 Hz +/- 10 Hz  (ie just below our 24GHz band)

The resonance will be a Hz or so wide, and it ought to be possible to see this with relatively simple equipment.    First of all, get a length , several metres, of waveguide for 24GHz.   This needn't be 'proper' WG, just squashed water pipe will do to allow just the TE01 mode to propagate.    Fill this with ammonia (obvs. make the transitions gas tight)

Use an ADF5355 synth, and drive it's reference from a 48 bit DDS loike the AD9852.   Anything to be able to tune the 24GHz signal in sub-Hz steps, and to know what it is.   That chip will leak enough 4th harmonic from its VCO; launch this into the waveguide.   At the other end stick a 24GHz receiver tuned suitably.
Vary the input frequency a few Hz either side of the resonance and look for a dip.

One snag - if you use GPS locking, the residual variation of the GPSDO will cause it to drift across the resonance over a few tens of seconds.

Work out a way of feeding this dip back and controlling the synth reference in a locking loop and you have your own atomic frequency standard.   It's potentially  the same performance as cesium 

It seems in years gone by that caesium was possibly adopted instead of ammonia, simply because its resonance at 9.19GHz was easier to generate using ex-WW2 equipment that was 23GHz back in the 1940s when the first Cs standard was developed.


--
Robin Szemeti - G1YFG