Clock 4 pendulum measurements

Here are some measurements of the clock 4 pendulum, trying to get a handle on its performance issues.  Woodward is adamant that limiting count wheel friction was major concern in his designs.  He employed a number of countermeasures, including anti-friction rollers, a polished acrylic count wheel, lightweight stainless steel pallets, and the merest hint of watch oil.  I don't know that my situation calls for such measures, but I figured I ought to investigate.

The pendulum is a solid square black walnut rod about 2" on a side, and is 48" from knife edge to bottom.  It weighs 28.86 oz, which is fairly uniformly distributed along its length.  The pendulum was fitted with a crude bob constructed of short copper-clad steel rods bound together with a rubber band located 45" (on center) from the knife edge, weighing 26.75 oz.

I measured pendulum amplitudes as deflections from equilibrium. 

Provided the amplitude is greater than 2.5" (3 degrees) and less than 5" (6 degrees), the count wheel advances reliably.  The count wheel does not advance at all when the amplitude is less than 2.25" (2.6 degrees). Double counting occurs when the amplitude is greater than 5.5" (6.6 degrees).

Here are counts of pendulum full periods starting at 4" (4.8 degrees) and ending at 2" (2.4 degrees), which is basically a half-time.  Pendulum Q can be estimated from this by Q = 4.532 * number of periods to halve the amplitude.

• Unloaded pendulum: 65, 72, 68.  Median Q = 308
• Pendulum driving pull pallet and count wheel, but no backstop: 58, 54, 54.  Median Q = 245
• Pendulum driving count wheel normally: 52, 45, 50.  Median Q = 226.

This indicates a count wheel-only reliable run time of about 100 seconds, which I've confirmed approximately on previous days.  If you push it a bit, you can sometimes do better on occasion.

There definitely is a noticeable change in loaded Q caused by driving the count wheel, as Woodward warns.  But, the unloaded Q figures are probably the source of my trouble, though.  The unloaded Q is around the same as a marine chronometer's balance (and not a good one at that), and that needs an impulse every period to keep running!  (I already know that the clock can run if it impulses every second.... it's not supposed to do that, though!)

Although this is probably excessive for my needs, Woodward has a table that lists a "heavy seconds pendulum" at Q = 15 000.  I think I need a better resonator!

Triggering the escapement certainly consumes energy, possibly a large amount of energy.  But it's unclear how exactly to measure that accurately...