The Correction Engine in Action — A Real Correction Cycle

The Principle

The clock had deliberately been set more than one minute fast and was then left to run without manual intervention. With the timing weight OFF, the clock's natural rate was slightly slow, typically around 3599 Beats per Hour (BPH), against a target of 3600 BPH. Over several days, the system observed the clock gradually losing its initial time advantage and approaching true time.

Why the Clock Needs a Reference Point

The pendulum can tell the Correction Engine extremely accurately whether the clock is running fast or slow, but the pendulum cannot tell the Correction Engine where the hands actually are. A clock could, for example, be running at exactly 3600 BPH while its hands were several minutes away from the correct time.

The Centre Wheel Hour Position (CWHP) provides the missing mechanical reference. A datum point is fitted to the centre wheel and positioned so that it reaches the CWHP sensor at precisely the same moment that the minute hand reaches 12 on the clock's dial (clock face). The datum point does not physically have to be at the 12 o'clock position on the centre wheel; its position is mechanically referenced to the minute hand.

The CWHP does not identify which hour the hour hand is displaying; it identifies the clock's position within each hour. For a correctly set working turret clock, this is all the Correction Engine needs.

Approaching True Time

At 22:00, the Centre Wheel Hour Position measurement showed that the clock had reached:

The clock was still running fractionally slow, so no correction was made. Instead, the Correction Engine continued to observe its movement. Over the following hours, the clock naturally passed through zero:

Simply becoming slightly slow was not, by itself, sufficient reason for the Correction Engine to intervene. The system also considers the direction in which the clock is moving.

By 02:00, the clock was three seconds slow and successive CWHP measurements had established that it was moving away from zero. The Correction Engine therefore determined that intervention was required and issued the command ADD WEIGHT.

The motor lowered the timing weight onto a timing tray mounted on the clock's pendulum, increasing the pendulum's rate, and the clock was allowed to respond.

The Correction

Adding the timing weight increases the BPH rate of the clock. With the weight OFF, the pendulum had been running at approximately 3599 BPH. With the timing weight applied, the measured rate increased to approximately 3605 BPH.

The effect can be seen in the following CWHP observations:

During the following hour, the increased pendulum rate recovered the three seconds that had been lost and carried the clock through true time to approximately 2 seconds fast.

Importantly, the Correction Engine did not immediately reverse the correction simply because the clock had crossed zero. Instead, it continued to observe the clock's position and direction of travel.

By 04:00, the clock had reached 7 seconds fast and was now moving away from zero on the opposite side. The Correction Engine therefore issued REMOVE WEIGHT, and the motor raised the timing weight away from the pendulum tray.

Returning to Observation

Once the timing weight had been removed, the pendulum returned to its normal unweighted rate:

The clock was therefore once again naturally moving towards true time, so no further correction was required.

Why Crossing True Time Is Acceptable

This correction caused the clock to gain approximately 7 seconds beyond true time. This is an acceptable part of the correction process rather than an error in the system. The timing weight is deliberately selected through practical experimentation rather than being calculated to produce the smallest possible correction.

For a turret clock, allowing the clock to cross true time by a small amount can actually be beneficial.

A clock installed in a tower does not operate under laboratory conditions. Its external hands and motion work encounter changing mechanical loads caused by factors such as wind and weather, while the pendulum and clock movement are also affected by seasonal and environmental changes.

The purpose of the Correction Engine is therefore not to force the clock to read mathematically perfect time at every hourly observation. Its more useful objective is to keep the clock within a small controlled band around true time, intervening only when its measured position and direction of travel indicate that correction is required.

In this example, the clock began more than a minute fast and was allowed to approach true time naturally over several days. It crossed true time and, only when it became slow and continued to move away from true time, did the Correction Engine intervene.

The timing weight was automatically applied, increasing the clock's rate. The clock regained the time it had lost, crossed true time and continued until it was approximately 7 seconds fast. Once the system detected that the clock was moving away from true time on the fast side, the timing weight was removed. The clock then resumed its natural rate and began returning towards true time — all without manual intervention.

Conclusion

A complete correction cycle also provides useful information about the effect of the correction weight itself.

With the weight OFF, this clock typically runs at approximately 3599 BPH. With the correction weight ON, its rate increases to approximately 3605 BPH.

This gives the Correction Engine considerably more corrective authority than would be required simply to trim a fraction of a second from the hourly rate. However, that additional authority may prove advantageous on a turret clock exposed to significant variations in mechanical load and environmental conditions.

Rather than continually attempting to force the clock to an exact mathematical zero, the system can allow the clock to move naturally within a small tolerance around true time and intervene when necessary.

As further correction cycles are recorded, the accumulated telemetry provides a clearer picture of how consistently the clock responds to the timing weight. This allows the correction process to be refined using the behaviour of the actual clock in its working environment, rather than relying solely on theoretical calculations.