11-Nov-2004

Introduction

The HP 5065A, like many atomic frequency standards, can be fine-tuned in frequency with the so-called C-field adjustment. The hyperfine transition used in the clock is slightly influenced by magnetic field so this phenomenon can be, and is, used as a fine frequency adjustment. The name is historical: the earliest atomic beam standards used magnets labeled "A" and "B"  to state select atoms on either end of the microwave cavity and third magnet labeled "C" to maintain the magnetic field within the cavity. Hence C-field.

Since the HP 5065A is not a primary frequency standard the C-field is set by the factory or by the user so that the frequency standard output matches a primary reference. Given the low drift rate of the 5065A, once set, the C-field is typically not adjusted very often (months, even years, depending on the accuracy required).

The specifications of the HP 5065A state that one minor division of the knob changes the fractional frequency by 4x10^-12 and the full scale range (10 turns) is 2x10^-9.

To clarify, turning the knob one full turn, which moves the "hour hand" one digit changes the frequency by 2x10^-10. Turning the knob 1/10th turn, which moves the "minute hand" one digit changes the frequency by 2x10^-11. And each of the 5 "minor division" tick marks between digits represent a frequency difference of 4x10^-12. There are 50 minor divisions per turn and 500 minor divisions full-scale.

If you've ever taken a ten-turn precision wire-wound knobpot apart you know setting precisely to one minor division is not far from the limit of its resolution.

The short-term stability of HP 5065A vary. For this experiment I used one with a stability of about 1x10^-12 at one second. The nominal value of its C-field for 11 digit frequency accuracy was 290 (or 2.90 depending on your convention of labeling the C-field dial).

Measurements

To clearly show the effect of turning the C-field dial I made continuous frequency measurements and quickly turned the dial a precise amount each minute. The 1-minute per division, 9 minutes full scale frequency plots below show frequency steps each instant the dial is turned. The magnitude of the frequency steps is calibrated by looking at the y-axis scale.

Major Digits

Below the 10-turn C-field adjustment was moved one full turn each minute (CF from 200, 300, 400, ... to 800) for a total of six turns. The effect is almost exactly 2x10^-10 per turn, as expected.

HP 5065A C-field adjustment - one full turn each minute

Minor Digits

Below the C-field adjustment was moved one tenth of a turn each minute (CF from 220, 230, 240, ... to 280), for a total of six tenths of a turn. The effect is almost exactly 2x10^-11 per digit, as expected. The 1x10^-12 short-term stability of the 5065A is just noticeable at this level.

HP 5065A C-field adjustment - one tenth turn each minute

Minor Divisions

Below the C-field adjustment was moved just one tiny minor division each minute (CF at 280, 282, 284, 286, 288, 290), for a total of five minor divisions, or one tenth of a turn. The effect is almost exactly 4x10^-12 per minor division, as expected. It is difficult to turn the knob and land right on a minor division. The short-term stability of the 5065A is barely good enough to resolve each setting.

HP 5065A C-field adjustment - one minor division each minute

Full-scale and Back

Below the dial was moved full-scale, ten turns, from min to max. Each minute I made one full turn, paused a few seconds and made a second full turn. The result is 5 big steps with a bump in the middle of each. After a few minutes at its maximum setting I turned it all the way down to its minimum setting. The net frequency change is 5 divisions of 4x10^-10 for a total of 2x10^-9.

HP 5065A C-field adjustment - two turns each minute, full-scale and back

Linearity

Below are two plots, the first showing the dial turning ½ turn starting at minimum and ending at 0.50 and the second showing ½ turn starting at 9.50 and ending at maximum.

At the beginning of the dial a half-turn changes frequency by about 9.3x10^-11 (from -5.902x10^-10 to -4.971x10^-10). Notice the instability after the change was made. It's not clear if this is due to turning the C-field (all the other plots are clean), wiggling cables (one must be careful), or something with this particular 5065A (I've seen this sort of instability now and then; unrelated to the present C-field experiment).

HP 5065A C-field adjustment - one half turn starting from minimum

At the end of the dial a half-turn changes frequency by about 9.1x10^-11 (from +1.3161x10^-9 to +1.4071x10^-9). This suggests the full-scale linearity is on the order of 0.2x10^-11 out of 9x10^-11 over 9.5 turns, or about 2%.

HP 5065A C-field adjustment - one half turn ending at maximum

Measurement Notes

The measurements were made with a TSC 5110A time interval analyzer (with 0.1 ps resolution) and a CH1-75 active hydrogen maser for the reference (<10^-13 at t 1 s). Using a conventional frequency or time interval counter or using a cesium or another rubidium standard for the reference would have made resolving minor divisions difficult.

Conclusions

• The C-field works as advertised with a resolution of 4x10^-12 per minor division.
• There appears to be little or no delay between when the dial is changed to when the frequency changes.
• There is no noticeable settling time in the output frequency for changes as small as one minor division to as large as one quick full turn.
• The control, error, and 2nd harmonic meter readings remained constant throughout the experiment.
• Linearity across the dial is a few percent from one end of the dial to the other.
• Retrace appears excellent. No hysteresis is visible.
• Given short-term stability on the order of parts in 10^-12, a drift rate of parts in 10^-12 per week, and finite mechanical resolution of the knobpot, one ought not to worry too much about subdividing a minor division.