Now that I have a decent steel wire drive between the main-spring barrel and the fusee, I found myself pondering over the possibility of setting the balance wheel screws to improve the clock’s timing.

As reported before, my version of John Stevens’ skeleton clock has several issues of time-keeping. My very simple attempts at torque measurement revealed the cause of poor time-keeping. Although crude, they provided enough information to show a distinct fusee inaccuracy, a situation which I’m no longer in a position to rectify.

A few months ago however, I commenced a series of tests and adjustments, and graphed the results. My final plot below illustrates two factors.

The heavy line displays daily errors. I took these at 5 am each morning based upon the assumption that they were perhaps the least affected by (Melbourne’s) ambient temperature variations. I was not serious enough to collect night-time readings. The caterpillars crawling up and down the slopes are the results of hourly daytime variations.

Given that the fusee does not accurately compensate for the varying output torque of the main spring, my intention was to set the timing screws on the balance wheel to the best compromise. After several weeks of timing tests and adjustments, I believe that I have set the mass of the balance wheel to the best compromise. Here are my observations:

• Over 8 days the variation ranged between +74 s and -79 s.
• When the main spring is fully wound, the stop-work mechanism fails to function.
• As the main spring runs down, the arc (swing) of the balance wheel begins to decline.

Until about day 6 or 7, the balance wheel arc (swing) appears to remain fairly constant.

My conclusions.

• I suspect that as the fusee torque varies up and down, there is a corresponding change in the friction at the locking and impulse faces of the pallets.
• I also suspect that there is a change in the mark/space ratio during the locking and impulse phases of the escapement.
• The fusee profile fails to accurately compensate for the output of the main spring.
• Selective use of the `flat’ portion of the main spring is not possible.
• The escapement is extremely sensitive to variations in the fusee torque.
• Balance wheel temperature compensation needs to be properly evaluated.

I shall now be content to wind the clock when it needs it, and set the fingers to the appropriate time.

Best regards to everyone.

Sam (aka Dennis) Stones

Melbourne

Hi John,

Many thanks for your prompt reply. There is certainly no need to apologise. If anything, I must apologise for failing to provide references to previous postings on this subject. A scan of the relevant threads reveals that I spawned (or contributed to), no less than twenty (20) which are either relate directly or are slightly off topic with this skeleton clock. I would be quite happy to lump the references together in chronological order if it would help.

Here are a couple of pictures from my albums to start with –

The main-spring output was determined by calculation using the (approximate) radii measured off the fusee. A sort of inverse determination.

In answer to your question –

“Is the profile you used a published one or is it made up or derived in some way?”

Yes. It was a 2” radius profile, taken directly from the drawing thus –

For various reasons John, I’m no longer in a position to alter the fusee profile.

In terms of setting up the fusee and main spring, I fear that their combination fails to provide useful `access’ to the so-called spring flat-spot.

In other words, up to the point when the stop-work lever `kicks in’, the main spring is almost completely wound. Adding any more clicks to the barrel ratchet fully winds the spring, and the stop-work mechanism doesn’t engage. Removing barrel clicks and the clock stops before all of the fusee line has been used up. At the start of day 8, the arc (swing) of the balance wheel decreases, and the clock speeds up. See the dotted line between day 8 and day 9 on my graph.

John, I’m don’t understand your statements –

. . . the escapement is supposed to ration the unwinding of the mainspring with a time dependent element.

or

Now an increase in driving force by a well positioned finger on the driving train will cause a speed change. It shouldn't cause the train to free wheel. If your train is seriously sensitive it would appear to be effectively a freewheeling set up and not a time latched/unlatched process.

Applying extra driving force causes the balance wheel to spin through an increasing arc until the impulse pin travels around far enough to strike the outer edge of the escapement lever. I would need to understand what freewheeling means in this context.

Finally, from what I can see of the escapement wheel (under low power magnification), it appears to lock with no detectable twitching. As for adequate depth, I estimate about 0.01” (0.25 mm) for both entry and exit pallets. The geometry of the escapement as a whole, conforms well with the drawings, as far as I can tell.

I would welcome your further comments.

Regards,

Sam

Edited By Sam Stones on 27/10/2014 23:16:25

Thank you Michael,

At four score next year, I must say "That's a weighty piece of literature!"

With my long-gone workshop; another lump of brass; and a guaranteed 12 months of uninterrupted effort, I think I could just manage a new fusee or two.

Perhaps I should have gone the electronics route?

Regards,

Sam

Edited By Sam Stones on 29/10/2014 03:33:31