[time-nuts] You can build a fountain from the things you find at home...

Magnus Danielson magnus at rubidium.dyndns.org
Sat Oct 1 09:13:25 UTC 2011

On 01/10/11 10:49, Attila Kinali wrote:
> On Tue, 27 Sep 2011 19:46:17 +0000
> "Poul-Henning Kamp"<phk at phk.freebsd.dk>  wrote:
>> In message<CAP6i9MmntLTia=XXUafbyrp+e3=koc1TRSEvL=g05emepQ4v-w at mail.gmail.com>
>> , brent evers writes:
>>> So at its most basic, I'm wondering what type of clock would make the
>>> most sense to consider - cesium fountain, or hydrogen maser?
>> Based on what I can gather, a fountain is probably easier to make work
>> than any of the other, from a pure _mechanical_ point of view.
> What makes you think it is easier to build?
>  From what i gather, the size of a Cs fountain is smaller than of a H maser,
> but otherwise the difficulty of the components you need around it is comparable.

A Cs beam can be smaller than a H maser, but a Cs fountain usually 
isn't. A particular issue with Cs fountains is that there is a fair bit 
of time between the firing of the Cs ball until it comes down again for 
observation, so the time between observations becomes quite long. The 
fly-wheel oscillator needs to be very stable, so you have a choice 
between a very high stability crystal oscillator such as OSA 8607 with 
palladium electrodes (only a few ever sold... to the big labs) or a 

> Not to mention that handling H is a lot easier than handling Cs.
> And as i've written before, the electronics for a Cs would be a lot more
> difficult to get working than for a H.

Very little of the electronics is at the atomic reference. A typical Cs 
or Rb produces 60 or 90 MHz and a suitable mix frequency 12,6 or 5,3 MHz 
respectively. That is built very well in through-hole. The signals is 
combined onto a coax with a DC bias to hit a SRD sitting inside a RF 
cavity. The one Cs beam that deviates from this scheme is the 5071 which 
actually has an oscillator running at the Cs frequency, but no pain no 
gain as that architecture provides better performance. Still not very 
much electronics run at the full frequency.

My scope has a 50 GHz channel and a pair of 20 GHz channels. Badly need 
a spectrum analyzer that goes that far and my network analyzer only gets 
me to 500 MHz. Luckilly I have some stuff at work. :)


More information about the time-nuts mailing list