[time-nuts] Cesium Mechanical Chronometer

[Tom Bales]

>
> Many thanks for everyone's response on my cesium-synchronized mechanical
> chronometer project.  I'll keep the group informed of progress.  Some
> responses to your questions and suggestions:


   - My current plan is to use a chronometer that has been adjusted for
   rate and is close to dead-on.  Of course, they're never dead-on, and that's
   why keeping track of the rate is important.  I haven't dismissed the idea
   of actually disciplining the chrono from the CSAC, but since this chrono's
   rate adjustment involves screws on the balance wheel (it has a free-sprung
   helical balance spring without any means of adjustment), that would be a
   great challenge.  Perhaps the best would be for the CSAC system to maintain
   a "rate card" of the chronometer, so that when all the electronics fail,
   and the user is left with only the chrono, at least they would have a
   really good knowledge of its rate and variation.  Doing so would require a
   feedback signal from the chrono, which could be acoustic or photoelectric.
   The electronic rate card could be updated now and then by turning off the
   sync impulse to the chrono and listening for the ticks to move relative to
   the 1pps.
   -  The chrono would need a bigger box in order to incorporate the
   CSAC/GPS system, microcontroller, and--the biggest part--backup batteries.
   That's part of the fun, and I'm thinking to mount the gimballed chrono in a
   purpose-built enclosure that will hold everything.  Connections to the
   chrono will have to be very flexible FFC or silicone wire.
   - I thiink auto-winding is a definite requirement.  The engineering
   challenge is to couple a small gearmotor to the winding system without
   disabling the manual-winding system that uses a ratcheted key.  Winding
   noise should be short in duration, so it probably won't affect the time
   sync.  People have wound chronos by turning them upside down in the gimbals
   and poking a key through a hole for a long time without any  big issues.
   - Coupling to the chrono balance and escapement might be difficult, or
   it might be easy.  If it takes hours or days to synchronize, that's OK, I
   think.  My guess is that anything that generates a slight impulse in the
   direction of the escapement detent or a torque pulse to couple to the
   balance will work well enough--a tiny little linear or rotary
   electromagnetic actuator should work.  We'll see.
   - The Hamilton M21 chrono beats at 2Hz, so driving it with a 1pps signal
   will probably work fine.
   - The idea that TVB suggests of disciplining the CSAC to match the
   chrono is fascinating, indeed.  Hadn't thought of that one.  It would be
   the easiest way to keep the chrono and CSAC in sync, and it would still use
   the traditional rate-card method of keeping track of the actual time.
   Hmmmm.
   - Yes, the CSAC modules are expensive!  Over $5k currently.  It seems
   like during the evaluation period the company sold them "at a discount" in
   order to build applications and sales (or so the court documents say).  Let
   me know if anyone sees a gyneesium one on *bay.  I thought long and hard
   before buying it, but idle hands are the devil's cash register.  (Oh,
   oh!--now I have an excuse to buy an HP 53230!)  In my initial measurements
   of rate (after GPS disciplining for 24 hours) the CSAC freewheels within
   1.7ms/yr.  Probably good enough.  Needs longer testing.
   - The M21 has a single balance wheel consisting of an Invar spoke and a
   stainless steel ring (it was Hamilton's magical way of doing temperature
   compensation to match the balance spring.  The Russian knock-offs have a
   standard bi-metal split balance wheel.  The Hamiltons beat just about
   anything in performance.  Besides, "Murica!".