Back Up Next

Debugging continues, correction for seconds hand finished, world time dial corrected and finished  - May 2021

This month's installment sees the finishing of the seconds correction dial. Be sure to compare last month's video before the correction to the video this month after correction. It was fortunate that we had a spare main time dial now that true jump seconds were being read out, that replacement dial had a more accurate seconds spacing to match the hand. Debugging continues, an error in the positioning of some of the cities in the world time dial was found, corrected and finished. One of the remontoire counter weights were moved away from an adjacent wheel to give more comfortable clearance, center remontoire wheel slightly repositioned to run in the middle of its mating pinion (this was not a bug but done to look better). The tellurian was slightly repositioned to give a better fit, a minor depthing issue to the planisphere drive is addressed. Many other smaller bugs were dealt with.


The escape wheel parts now finished, next the seconds drive wheel and the finished count wheel, simply beautiful. 


The second photo more clearly shows the skeletonized spring barrel located between the count and drive wheels.

Wheel spoke artistry simply doesn’t get much better than this, note Buchanan's signature sharp inner corners and where the connecting points to the outer toothed rim nearly disappear. The last wheel!


The components of the pallet fork now finished.

The finished pallet fork. Note the paired and unequally sized jewels, this design is key to the conversion of of each two-second input into a pair of one-second outputs.


Buchanan drew a set of 60 equally spaced lines and printed them out on paper and then placed the paper dial on the enamel dials; the one with no bezel is the dial with the two minor cracks that we discarded (first photo). The other is on the clock at present. It shows the bad spacing of the seconds lines, (second photo), compared to the spare.

It was fortunate we had two dials made by the enamel artisans in China. The one we had thought we would keep had no imperfections as far as the enamel work. However when Buchanan made his drawing showing the equal spacing that the seconds hand will now produce thanks to the revisions made last month, we see that they do not line up well. The dial we did not use, (below), has much better spacing and only had a very small crack and blemish near the thirty-two seconds mark, this was very minor and will be mostly covered with the bezel. Given how much better the spacing is on this dial we decided to make the switch.

The dial blemish is very small and demonstrates the high standards for this to be the second choice dial.

The blemish nearly disappears when placed behind the dial bezel. And to be fair, a very small crack almost always appears somewhere on large antique enamel dials, in particular those that take the shape of a ring which are more fragile than the conventional disc shape. Also note the visually tight and accurate bezels around each individually moving dial ring between the mean solar, and sidereal hours and minutes ring.

Note the inconsistencies in the rotation of the wheel shown by the arrows. This wheel used to be the output for the seconds hand. This is now buffered by the spiral spring (within circle), allowing the indexing seconds count wheel to accurately display jump seconds from the former erratic two seconds output wheel caused by the remontoire reloading cycle. By creating this indexing wheel and spiral spring buffer we have achieved two goals: First was to have a reliable seconds hand output. However, before this change the pendulums as they are designed have a two second period (4 second cycle) meaning that the second hand, even if not erratic, would jump every two seconds. However, the feedback from the remontoire reloading cycle was also causing the hand to act erratically, floating, even skipping a second or going backwards! Not an acceptable situation. The spiral spring and what looks like an escapement wheel, but is actually an indexing wheel, is now the output for the seconds hand achieving a reliable period, but even better, changing the period from a two second to a true one second jump, making the time dial look like a conventional regulator, even though it has a two second pendulum. We chose the two second pendulum because it has a better visual, almost hypnotic effect on the viewer compared to a one second period and I was content to have the seconds hand jump every two seconds. But with this design change we have solved the erratic feedback problem and have given the time dial seconds hand the advantage of a one second readout without losing the two second pendulum period. A rare win-win in engineering design.

In this video one sees what looks like an unusual escapement. This is not an escapement but a count wheel. The pallet fork is controlled by the two second compound pendulums so it takes four seconds for the fork to make a full back and forth swing; since the fork has a pair of two separate pallets, these four pallets divide the four second pendulum cycle into four individual contacts with the count wheel, that wheel is connected to the second hand resulting in a true jump seconds from a two seconds pendulum swing (one half cycle). Think of this as a 'reverse coup perdu' device since a typical coup perdu creates a one second dial readout from a half-second pendulum, the words are French for "lost beat" since only every other beat is recorded on the dial. In this case we have a "gained beat" perhaps the French translation could be "battement gagné"? A complete explanation can be found in last month's April installment Look carefully at the movement of the seconds hand in comparison to the twin Harrison grasshopper escapement wheels to the right of the dial and one sees how two jump seconds are generated on the dial for each two second long beat of those escapements. Also turn up the volume during the last frame of the video! This is the first video where one can clearly hear the many different sounds the machine makes, when not striking. It now has what one would expect as a regular 'tick tock'. The grasshopper escapement was always virtually silent. The video was taken close up when the background noise was low so some of the clicking and whirling can be heard.


These two photos show the beginning of Buchanan using a Microset timer to get the count wheel properly calibrated, the second photo shows the improvement from the first.

Now Buchanan turns to the world time dial


The first photo shows all the errors on the international time dial. Buchanan printed out a paper with lines on the correct position of the hands for each city and placed the faulty dial above it, (first photo). Paris Moscow and Abu are correct. Bangkok is hopeless. Sydney, Los Angeles, Chicago and Rio are all an hour out.

We decided to drop Bangkok, Shift Sydney and Beijing anticlockwise a bit and this will get Sydney into the correct place and Rio, Chicago and Los Angeles a little clockwise, Honolulu was placed in the space at the bottom of the dial to keep reasonable spacing. Honolulu only having one more letter than Bangkok, (second photo).


The new design is seen in the first illustration, next the that design fresh off the mill.

The finished international time dial. The city depiction ring is plated in French silvering.

This photo shows one of the myriad of debugging problems. The remontoire counterweight in the time train was too close to an adjacent wheel, here Buchanan has reworked the position of the weight to give a better clearance.



The machine is getting very near completion!

Case construction subcontracted in the United States

The brass trimmed case is being made locally. Fred Widman is a skilled model engineer located about an hour from my home. Each glass pane has a brass surround which fits into the main frame and can be secured with the knurled thumb screws, but is still fully removable from the main frame allowing access to the entire machine.


These two photos show the long brass stock securely held before milling to shape can begin. This is necessary to keep the stock from curling during the milling operation.


First photo shows brass strips that hold the glass within the brass channel frame surrounds. Next the exterior frame skeleton is milled.

This photo shows the cross section of the main skeletal frame. The brass glass surround frames fit into the recess.


The first photo shows a portion of how the window frames are constructed. The corners are secured by the angle piece with threaded holes that hold screws through the channel, the same type of piece is shown installed in the corner to secure the channel. The empty area is where the glass is inserted, and the remaining area along the channel has strips of the same width that is occupied by the corner piece securing the glass. Next an interior shot of the corner construction of the case. One can see the myriad of parts used in this design to make the glass fully removable on all surfaces.


The first photo shows the glass frames installed. The knurled knobs on the top screw into the frames to lock them in. The bottom of each pane is secured by a pair of pins located in the case bottom rail. Next a view of the interior corner bracing.


Next a view of the interior and exterior joining work for the complex corner design. All this is necessary since the frame must remain rigid to accept the heavy individual brass-framed glass panels.

The case near completion.

Total parts count to date: 7951. This count now includes the parts for the case. Additional parts will later be added for winding and setting keys.  

Back Up Next