The Shortt-Synchronome Synchroniser 

applied to a Gents Clock

G4OEP Homepage

Shortt Clock

Synchronome in action

Gents Clocks

Synchronome Company Leaflet

This page should be read with reference to Phase-Locked Pendulum Clocks which describes a pair of  pendulum clocks having electronic escapements which are phase-locked together, and to an ovened crystal oscillator which acts as an external frequency reference for both.  The current page describes how a Gents electro-mechanical clock was added to the phase-locked system by adapting a mechanical synchroniser originally designed for use with the Shortt Free-Pendulum clock.

A full description of the Synchronome, the Shortt Free-Pendulum Clock and of the Shortt Synchroniser is given in the copies of an original Synchronome pamphlet, and for a full understanding of these mechanisms, that material should be studied.  For our present purposes, however, the following extracts are sufficient:- 

Fig. 1.  Shortt Synchroniser.

The diagram of the device seems understandable, but to my mind the explanation lacks rigour and detail.  Notwithstanding this, I set about making my own version of the synchroniser, and the result is illustrated below, shown at various stages of development.  It consists of a stripped-down relay movement with a small rubber block attached to the contact-actuator bar. Referring to Fig 3, the air-gap of the relay is at the left, while the armature is pivoted at bottom left.   A foil spring, visible as an L-shaped yellow line keeps the air-gap open when the relay is not activated.  When the relay is activated, the air-gap closes, and the bar across the bottom of the relay rotates downward, bringing the black rubber block into the path of the leaf-spring which is attached to the pendulum.  If the relay is not activated, the top edge of the spring passes under the block without touching it.

Fig. 2.  Synchroniser fitted to Gents Clock.

Fig.3. Detail of Synchroniser.

Click for video

This device differs in several respects from the Shortt version. Firstly, the synchroniser is activated on a 2s period, rather than the 30s period of the Shortt.  Secondly, simply as a first attempt to be refined later, I left the shapes of the contacting parts on the armature (the rubber block) and the leaf-spring largely as they came to hand.  These parts are shaped in the Shortt version so that the armature can only contact the spring on its vertical face, whereas in my version the rubber block can contact the spring either on its vertical face or on its top edge.  

Initial trials with the relay operated from the 40ms pulse derived from the crystal unit of my own clocks, and with the Gent set to a losing rate of about 1s per hour were immediately successful.  I found that the pendulum came into phase lock with the relay rapidly and reliably with the mode of contact
differing, depending on how the synchroniser was initially set up (lateral position, vertical position, degree of losing rate, etc).  In some settings, the synchroniser changed from one mode of contact to the other at different stages of the 30s cycle of the Gent, changing from top contact to lateral contact as the amplitude of the swing declined during the 30s cycle.  I eventually found an adjustment that ensured that the top-contact mode continued throughout the cycle; Fig 2 illustrates this mode of operation.

How does it work ?  I found great difficulty in explaining this to myself, but eventually reached the conclusion that I could only give a convincing explanation if I adjusted my mind in two respects.  Firstly, I must abandon all attempts to visualise this synchroniser as a mechanical analogue of my purely electronic system.  Secondly, I eventually realised that there must be at least two distinct modes of operation which are radically distinct, one characterised by the relay operating while the spring approached the synchroniser, the other with the relay operating while the spring recedes.  This was a major cause of mental strife, but is, I think a distinct advantage of the device, since it allows great freedom of adjustment.  If initial adjustments are not conducive to phase-locking in one mode, the system automatically reverts to lock in the other !

Mode 1: Relay activated while Pendulum approaches Synchroniser.  This is illustrated in Fig 4. (right).  The vertical axes of the graphs (x) represent the lateral displacement of the leaf spring during the swing of the pendulum; this is shown by the curved purple line (part of a sinusoid).  The horizontal axes are time, represented as on an oscilloscope screen as a repeated 2-second sweep (only part of which is shown).  The vertical red lines show the times at which the relay is activated (left line) and released (right line). These lines remain fixed relative to the time axis since the relay-activating pulse is regarded as a time reference.   However, the sinusoid representing the motion of the leaf-spring moves progressively to the right, i.e. to later times, on successive 'sweeps' of the display (progressive retardation of phase of the spring relative to the relay-operating pulse).  This is because the pendulum is set to a losing rate.  The top graph shows the spring engaging with the relay at a relatively early phase of the pendulum, while the lower graph shows what happens when the relay activates with the pendulum relatively retarded.   Clearly, the spring is deflected more in the former case.  Since the deflection of the spring represents energy removed from the pendulum, it is evident that removal of energy depends on phase difference, being greater when the phase of the pendulum is later.  Removing energy from the pendulum causes it to swing with smaller amplitude, and this reduces its period (circular error), causing it to speed up, reducing, or even reversing the losing rate of the pendulum, and the right-ward drift of the curved red lines.  As the curved line then moves to the left, the energy loss per collision is reduced.  Energy loss, and reduction in period continue at a steadily falling rate until a steady state is reached where only sufficient energy is remove for the period of the pendulum to become exactly that of the relay.  At this point phase-lock will have been achieved.

Fig 4.  Phase locking in Mode 1.
Mode 2: Relay activated while Pendulum recedes from Synchroniser.  This is illustrated in Fig 5. (right).  In these diagrams the horizontal blue line represents the X position of the rubber contact block when the relay is energised (position of relay when down).  In the upper graph, contact between the synchroniser and the spring is made at the moment the relay descends, but contact is broken not when the relay is released, but at point d where the x position of the spring is less (i.e. more toward the centre line of the pendulum arc) than that of the relay block; at this point the spring has moved away from the synchroniser and is out of range. The corresponding deflection of the spring during the engagement is a.  In the lower graph (pendulum at a later phase) the synchroniser disengages from the spring at the moment the relay is released, as in Mode 1.  The deflection of the spring is now distance c.  Clearly c exceeds a, even though c is less than b. Once again, as in Mode 1, the deflection of the spring increases with retardation of phase, deflection varying from a (which could be zero with an earlier phase) to a maximum of c.   As before, the conditions are correct for phase lock to occur.

Fig. 5. Phase locking in Mode 2.

When I had originally completed this part of the page, I concluded ... "I feel that there is probably more detail to this synchroniser than I have described here, but at least I feel that I have established some of the broad outlines of its operational characteristics." In my mind was an awareness that when the leaf spring engages with the relay, it applies a component of lateral force which either assists or counteracts the restoring force due to gravity.  The system is then no longer a simple pendulum; it assumes some of the characteristics of a spring-mass oscillator.  I assumed that this was true for such a brief moment during the complete cycle of the pendulum (~20ms every 2s) that it could be ignored, and that the main mode of operation was as described above: it depended on the abstraction of energy from the pendulum, and the period then changed as a consequence of the circular error.

However, while adjusting the mechanism of the Gent, I made an observation which was of the nature of a crucial experiment which caused a mini Copernican revolution - a definite paradigm-shift in my conceptual system !

The Gent mechanism includes several adjustments , and is in fact a knob-twiddler's paradise.  One of these adjustments is the back-stop of the magnet armature.  This controls how far the gravity lever falls down the cam (which is fixed to the pendulum rod) before the contacts engage, and the gravity lever is restored.  It therefore acts as a fine control over the energy which is imparted to the pendulum each 30s cycle, and thus controls the amplitude of swing when the clock is operating in its normal non-synchronised mode.  

However, according to the theory described above, when the clock is synchronised there can be no change in amplitude.  Any amplitude change would imply a change in period, and this is not possible when the clock is synchronised.  Any change in energy input due to a change in the setting of the back-stop would be counter-acted by the synchroniser mechanism according to the principle described above, and the net energy stored in the pendulum (and therefore its amplitude) must be constant.  I was therefore amazed to find that the back-stop adjuster did, in fact, change the amplitude when the clock was synchronised.  This observation is completely incompatible with the theory based on circular error which I had so painstakingly constructed; the synchronised pendulum was not behaving according to simple pendulum theory where circular error applies.  Constant period with change in amplitude can only be explained if one incorporates a variable element of spring-mass oscillator theory.  The explanation given above is thus at best only partial, and fails to explain this new observation.  A new theory is required which is based on the idea of a gravity pendulum compounded with a spring-mass oscillator, the spring component of which is variable, dependent on the relative phase of the pendulum and the synchroniser.