[time-nuts] Home built cesium clocks???
bruce.griffiths at xtra.co.nz
Sun Jun 29 09:35:27 EDT 2008
Magnus Danielson wrote:
> From: Predrag Dukic <stijena at tapko.de>
> Subject: Re: [time-nuts] Home built cesium clocks???
> Date: Sun, 29 Jun 2008 12:10:46 +0200
> Message-ID: <126.96.36.199.1.20080629120421.01ebecc8 at tapko.de>
>> I did my homework and collected everything from the internet I could
>> get. I have this one too.
>> If I go optical way, I still need 9+GHz electrical source, so I will
>> first recreate electronics. Using off the shelf DDS chips, or using FPGA.
> A YIG oscillator or similar should be considered. The FPGA would be great
> for the state handling, but be sure to externally reclock the signal
> before use to remove the FPGA jitter.
>> Also I need 9 GHz AOM, so that I can split the same optical beam and
>> have two wavelengths 9ghz apart. That is a problem because I have
>> only 350 MHz AOM.
> In the article he referenced, the AOM only needs to handle 250,1 MHz.
> Only if you intend to achieve 100% pumping and detection ratios you need
> a full set of frequencies. Notice the important note on relation between
> laser linewidth and S/N relationship. Luckilly those are limitations
> outside of the cavity.
> The interesting aspect with an optically pumped cesium is that one of the
> common failuremodes, the contamination of the masspectrometer is removed.
> The detection is off-axis from the beam. Wonder if an open oven could not
> be installed there. That would allow for a ping-pong mode of operation,
> which the optical pumping itself fits very nicely too. It would cancel
> some of the systematic shifts due to assymetries in the microwave
> assembly which to the best of my knowledge is hard to compensate normally.
> Maybe state of art designs have found a way to handle it properly.
>> Multiple pass is difficult, it would take cca 27 passes to get 9
>> GHz. On the other side, I wouldn't need the last stages of SRD
>> multiplication to get 9 GHz microwave.
>> No doubt, I will have a lot of fun with the project.
> Surely. It could be hairpulling too.
Even PTB's CSF2 caesium fountain doesn't require a 9 GHz AOM:
If one starts with a low noise 9GHz oscillator and divides down to a
suitable PLL loop frequency using regenerative multipliers then SRD
multipliers arent required.
Alternatively if you can still find a suitable NTL multiplier they are
somewhat quieter than SRDs.
More information about the time-nuts