Hi

The “classic answers” to this gizmo:

1. Rotate in one axis and re-mount the part for each pass. Come up with a 3 axis “quick change” mount for the OCXO.

2. Design it with “clicks” that index the rotation every (say) 360/12 = 30 degrees. Interpolate from there. It is a “smooth curve ….

3. Your device may have a “zero sensitivity” axis. X, Y or Z might essentially have zero G sensitivity. This is not all that unusual. A two axis setup would work (once you know which axis is zero …).

If you want to automate things stepper motors are your friend. Since 3D printers use them, there are a lot of three axis driver cards out there. Mate them up with an Arduino and you have the “drive hardware”. The size of the stepper motors likely scales significantly between the “outer” and “inner” axis. The outer motor also is rotating the two other motors plus mounts plus OCXO. The inner motor just rotates the OCXO.

One advantage today is that you likely would 3D print the parts. Back long long ago, we machined them out of aluminum.

Fun !!

Bob

Now I have a vision of a quartz oscillator tourbillon. (!)

Maybe more practical (it seems odd to say practical at this fiddly
level, but who knows, this might be useful) would be a 3-axis
accelerometer (they're cheap, used in every smartphone) detecting the
direction of gravity on the crystal and a microprocessor (with that
sphere map) calculating a frequency correction signal, similar to how
a tcxo does for temperature. Perhaps I should apply for a patent.

On Fri, 20 Feb 2026 at 10:23, Tom Van Baak via time-nuts
time-nuts@lists.febo.com wrote:

Hi

Indeed folks have been combining accelerometers with precision oscillators for a few decades.
The process does work to an extent.

The gotcha is that most customers are after compensation that works for dynamic situations. A
simple / slow / gentle tip is not of a lot of interest. This turns the measurement into “vibration
compensation”.  The same “match up an accelerometer” process gets used. The gotcha is matching
the phase / delay of the accelerometer output to the quartz.

Net result is that 10 Hz is pretty easy, 100 Hz is a bit more of a challenge and as you get to 1 KHz
(or beyond) it gets crazy.

Bob

This made me think of a couple of things that aerospace folks use for inertial guidance systems. One is a floated instead of gimbaled inertial measurement unit. This was tried several times, but the early ones had high maintenance issues (the C-5 IMU had a floated ball with the gyros and accelerometers in it). The most successful one has been the advanced inertial reference sphere (AIRS) used for the Peacekeeper missile. It uses a pumped liquid system to maintain its orientation relative to some vector the missile uses. It’s a mostly beryllium sphere floated in a fluorocarbon fluid. Power is transmitted to the sphere via pickoff rollers.

The other thing I’ve seen done is for a multiaxis ring laser gyro made into a tetrahedral configuration. The rotations of the beams in different planes and the resulting Sagnac effect shifts can be converted into a cartesian (or other) coordinate system. The tetrahedral configuration (also used with some mechanical gyro systems) can be used when redundancy is needed. Loss of a single gyro could be compensated for.

Either of these (neither particularly inexpensive) could work to sense the axis of the quartz crystal. With MEMS accelerometers and gyros,the change in position of the crystal’s axis could be determined and the data used to correct for position changes.

Sent from Proton Mail for iOS.

-------- Original Message --------
On Wednesday, 02/25/26 at 19:40 Ben Bradley via time-nuts time-nuts@lists.febo.com wrote:
Now I have a vision of a quartz oscillator tourbillon. (!)

Maybe more practical (it seems odd to say practical at this fiddly
level, but who knows, this might be useful) would be a 3-axis
accelerometer (they're cheap, used in every smartphone) detecting the
direction of gravity on the crystal and a microprocessor (with that
sphere map) calculating a frequency correction signal, similar to how
a tcxo does for temperature. Perhaps I should apply for a patent.

On Fri, 20 Feb 2026 at 10:23, Tom Van Baak via time-nuts
time-nuts@lists.febo.com wrote:

time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com

Hi

If all you are after is static compensation, some sort of accelerometer IC is likely to be plenty good enough.

If you are after vibration compensation, the time lag between the sensor output and the crystal “seeing” the vibration is the big deal. Again, accuracy of a fairly simple device is good enough.

Time lag gives you a phase shift. A phase shifted signal can not fully cancel out the vibration. The greater the phase shift the worse the problem.

What is “good enough”?

The oscillator wanders around all on its own. A phase plot will show this. ADEV is one of many ways of trying to measure this. All oscillators “wander”. Taking the vibration or acceleration numbers to that “noise floor” is indeed good enough. Since the compensation is done by feeding measured data into the compensation MCU, that also forms a floor to how good you can do.

So, some math. (Yes there are assumptions made):

Start off at 1 ppb for a static setup.

Floor is at 1x10^-11

Net, you need a 100:1 sort of range. A device with 8 “valid” bits of readout should be plenty good enough.

Bob