[time-nuts] How Rubidiums make their frequency
Tom Clark, K3IO (ex W3IWI)
K3IO at verizon.net
Wed Apr 19 18:33:34 UTC 2006
Christopher Hoover asked:
one issue remains: i have to crank the magnetic field setting almost to
its high limit (9.91/10.00) to get 5 MHz out; lower settings give a
frequency that is too low. i presume this is unusual.
i have a rudimentary understanding of the rubidium oscillator physics, but i
do not understand what would cause this. can i buy a clue?
I don't know the Tracor, but I imagine it is like most of the other
Rubidiums in it's innards.
Inside the physics package of a Rb, a cell with some Rubidium is
heated to (that's why Rb's run not!) enough so that it is turned into
a gas. Both light and microwaves illuminate the cell. If no magnetic
field is present on the cell, the Rb gas has a hyperfine resonance
(the difference in frequency between two infrared transitions of the
Rb gas) at 6.8346826128 Mhz. When a magnetic field is imposed, the
energy difference between the two hyperfine states changes.
In the RF part of the signal path (here, the block digram of a typical
Rb standard helps. See Page 3 of [1]this Symmetricom White Paper .)
Let's start with some convenient oscillator at, let's say 10 MHz.
Multiply it up to 60 MHz and then hit a Step Recovery Diode to get the
114th harmonic at 6840 MHz.
Then difference between the 6834.. and 6840 MHz is 5.31738+ MHz. In
the standard Rb configuration, we apply a magnetic "C-field" to bring
the difference frequency upwards by 4.89 kHz to 5.31250000 MHz which
happens to be 5MHz + 5/16MHz. Back in the early days, we didn't have
nice programmable DDS chips, but simple dividers/multipliers could
make the 5/16 MHz "adder".
So what you are doing by tweaking the magnetic field to shift the RF
resonance of the Rb cell so that it matches the arithmetic "quirk"
that the 6834 MHz is almost contains the neat 5/16 MHz in the tail-end
digits.
Hope that helped -- 73, Tom
References
1. http://www.symmttm.com/pdf/Precision_Frequency_References/wp_mmrfs.pdf
More information about the Time-nuts_lists.febo.com
mailing list