The, so called, "free pendulum" time-keepers had a distinguished history. For example, Frank Hope-Jones developed his "Synchronome" clocks in the late 19th and early 20th century. By 1921, W. H. Shortt had greatly improved the Synchronome; primarily by relieving the pendulum from its burden of pulling around the count-wheel. Mr. Shortt's clock set the standard of time-keeping accuracy for the two decades prior to the advent of quartz-crystal chronometer.
Some of you may recall having seen the Matlock-Collins Clock advertised in various DIYer magazines in the mid-1970s. Here is a scan of the Matlock-Collins Clock kit, as offered in the 1976 issue of Caldwell's wonderful, Craftmanship Catalog.
This is a simplified version of Mr. Shortt's "free pendulum" clock. Of course, given that the pendulum is impulsed periodically it's not absolutely free, but that's another matter.
The electronic timer replaces the mechanical count-wheel that's used in the Synchronome clocks. In other words, the pendulum of the Matlock-Collins Clock doesn't have to drag around a little count-wheel in order to synchronize the tripping of the gravity arm. The idea is to let the gravity arm reset-impulse also reset the 555 timer. This timer (electronically) waits roughly one minute before dropping the gravity arm again to impulse the pendulum. But it's the pendulum itself that regulates the overall timekeeping of the clock. As simple as it is, I've never gotten around to building this clock.
A few years ago I picked up an apparently unused, hermetically sealed tuning fork assembly at a local hamfest. The tuning fork was invented in 1711 by John Shore (a trumpeter in Handel's orchestra). By the mid-1950's the maker of my fork, Philamon Laboratories, had advanced the technology to a high degree of perfection.
I came across this tuning fork again a few weeks ago while searching through my junk box for another component. Seeing it prompted me to build a tuning fork drive circuit that's roughly based on the scheme used in the Matlock-Collins Clock. It turned out to be a very simple and enjoyable project.
My tuning fork-based emulation of the Matlock-Collins design begins with a zero-crossing voltage comparator that's wired to one of the two electromagnetic coils housed in the sealed resonator. The comparator has a very high input impedance. The coil terminal labeled "1" on my tuning fork has a DC resistance of 855 Ohms, while pin "2" has a resistance of 2000 Ohms. I use the higher resistance winding as my "sense" coil.
The comparator outputs a 1600Hz square-wave; the falling edge of which triggers a 555 timer that's configured as a monostable multivibrator. The timer output promptly goes high and ignores any further trigger pulses until nearly 20mS has elapsed. At this time the timer output falls back to zero and the circuit awaits the arrival of the next falling-edged zero-crossing. Allowing the highly accurate tuning fork signal to periodically reset the timer drastically reduces the long-term accuracy required of the 555 timer circuit.
The timer is followed by a D-type flip-flop that's also configured as a multivibrator. Its purpose is to set the tuning fork drive-pulse to the optimal width; Msgr. Fourier's mathematical poetry in action!
The 555 timer in the original Matlock-Collins Clock operates at just under 60 seconds in order to produce a pendulum to drive-pulse frequency ratio of 30 to 1. The 555 timer in my tuning fork time-base is set to just shy of 20mS to affect a frequency ratio of 32 to 1. Thus, the drive pulses are regulated to 50Hz. The 2N7000 transistor insures that absent the drive pulse, no load is placed across the tuning fork drive coil. As with the original Matlock-Collins design, the brief drive impulses are the only external interference to the "free vibration" of the fork tines. Properly buffered, the output signal might be used to drive either a 50Hz synchronous clock motor or an electronic clock display.
As a practical chronometer time-base my tuning fork has one potentially fatal defect. The fixed resonator frequency is not easily adjusted. A simpler feedback oscillator arrangement offers the possibility of fine-tuning the oscillator frequency by varying the phase of the drive signal in the feedback loop. It might be possible to gain some control over the oscillator frequency by introducing a time-delay in my Matlock-Collins drive pulse. Simply varying the pulse-width of the drive might produce some slight change in the oscillator frequency as well. Although my tuning fork is currently operating at 1600Hz so far as my bench frequency counter is concerned, that measurement is far too approximate for time-keeping purposes. It remains to be seen how closely my Philamon tuning fork has been factory-adjusted to 1600Hz.
Aside from this worry, my Matlock-Collins tuning fork emulation appears to be working quite well. Adjusting the 555 timer potentiometer results in well-behaved discrete jumps in the frequency division ratio. I've watched the circuit work stably at an 80 to 1 ratio. Of course, as the division ratio increases so the accuracy required of the 555 timer circuit is increased. However, the operation is quite robust at the normal division ratio of 32 to 1.
By the way, if the signal output were taken at the resonator one could think of this circuit as a shock-excitation type frequency multiplier (similar to the frequency multiplier used in my Bell Ringer radio transceiver, for example).
Here's a sketch of my circuit.