[time-nuts] HP 5061 zeeman frequency experiance
magnus at rubidium.dyndns.org
Tue Oct 4 03:06:42 EDT 2011
On 04/10/11 04:56, paul swed wrote:
> Hello to the group.
> Still working on the 5061 with a 5060 tube. Have home brewed a new CS oven
> controller thats appropriate to the 5060 tube and its working quite well. 2
> op amps and a couple of transistors. It handles the 5060s higher current
> needs. Average current is 650 ma at 98 ohms for the thermistor.
> The zeeman frequency is still bothering me. Granted I may be dealing with a
> weak tube.
> But I can adjust the cfield and get the various peaks. But they are
> small. The normal HP meter is a 50 ua movement. Using a 25 ua larger scale
> meter to monitor the beam current results in,
> No modulation 8 ua
> Driving 42.82KC at +1dbm I actually see the meter drop to 7.5ua.
> Then adjusting the cfield to 20 ma I can see the middle peak at 8.5 ua maybe
> a little higher.
> Is this the peak? Or better question whats a normal old CBTs beam current
> and the peak expectation.
> I am wondering if I simply am way down at the edge of usability.
> I can adjust the ionizer and get a higher reading just unsure were to stop
> at. Burn something out perhaps like the ionizer is the risk.
> The other thing is that there is still a drift. You can see the system lock
> for about 3-5 minutes then slowly become unstable and slip 5ns. Depending on
> the zeeman peak used it goes right or left. The green op light stays on
> however so the slips awfully small.
With a very small trick you can sweep your tube using a network analyzer
and a mixer.
Take the Port 1 of your network analyzer and sweep around 12,6 MHz which
you insert in replacement of the synthesis frequency. Hook the 10 MHz of
the crystal to the network analyzer input for additional stability. Use
the detected signal (as amplified) and control a mixer which takes a bit
of the 12,6 MHz and feed that to Port 2 of the network analyzer. That
way you modulate the amplitude.
Should give you an overview of the 7 peaks, there spreading and their
Note that this also works for spectrum analyzers having a sweep output,
as no phase is recovered this way anyway.
More information about the time-nuts