Back Up Next

Debugging continues, begin timer trials  - July 2021

This month the clock begins timer trials using a Microset timer system. Debugging continues apace. A change is made in the Robin remontoire chain, and a winding key refinement. A tight spot in one of the Wagner remontoire cages is fixed.


In addition to the Microset timer one to three cameras are stationed around the machine to watch for any anomolies throughout the day and night.

First debugging: Robin remontoire detent feed skipping a tooth. Buchanan's explanation shows the usefulness of keeping a constant video surveilance.

Buchanan writes: We had one glitch in the whole night and an explainable one. I am sending you a video link.

If you look at 20 min and 52 seconds into the video you will see the blued S arm of the Robin (top right hand corner) do a small backward jump. This is when it did not feed a tooth on the unequal cam.  If you watch the upper position of the centre weight sprocket, you will see its upper most point in the remontoir cycle is higher after the skipped feed than before.

I start at the beginning of the video and mark any reference positions with a fine felt tip pen (make sure you can remove it off the screen afterwards!) photo 337. Photo 452 is the same screen with the video playing.

I also mark the position of the two knobs on the poising weight to check the Wagner relationship, as well as the feed ratchet tooth position.

In the lower centre you can see the two escape wheels and the large bevel wheels on the escape arbour. You will remember that the escape arbour is two concentric tubes. The rear escape wheel and front bevel are attached to a tube between the front escape wheel and the rear bevel. If you watch the rear escape wheel and the front bevel gear, you will see them rotating in the same direction but the escape wheel recoils when the bevel remains stationary. You will also see the shock of the remontoir reload being absorbed but the cushion spring in the hub of the bevel gear, this the  other shock absorber spring that we were speaking about on the phone call yesterday.

All that happens is that the curved blued arm in the top right corner doesn’t fit in the slot in the unequal cam and rotates backward by less than a degree at that point in the video at 20 min and 53 sec. (It helps to put a post it note, blue tack, or some temporary marker on the screen to get a stationary reference on the curved arm.)  This means that it looses a release cycle and the Robin weight is now stopping higher each cycle. If this happens too many times the safety clutch will operate and the celestial; train will stop and be out of phase with the clock.

Here is another example of serendipity or maybe just luck. The Robin remontoire has a built-in safety clutch for just this type of error; to prevent the jamming of the remontoire causing an additional feedback jam to the time train. Now to be fair, we could probably have eliminated a Robin remontoire and fed a simpler unlocking to the celestial train, but it still does not eliminate the possibility of the celestial train jamming. It has to also be noted that in the original weight-driven design both the time and celestial trains ran off a common gear connection to both weights of the time and celestial trains. This was necessary as the weight for the time train was five-times that needed for the celestial train. By sharing the weights, we could make each weight more proportional. When the switch was made to springs in October of 2019, this was still necessary as the sharing of the springs resulted in the same proportionality. Getting a single spring to carry the entire power requirement for the time train could not have fit into the existing barrel design. Again the fact that each train has a state-of-wind dial will alert the user quickly should the time and celestial train drift; another example of  serendipity since these were completely unnecessary under a weight-driven design, but essential when the machine was redesigned to spring drive.

Second debugging: Robin remontoire drive sprockets.

Buchanan writes: I had noticed a little sluggishness on the Robyn’s operation. I spent some time yesterday investigating and found that I had lacquered all the sprockets as I thought that the chain had plenty of play on the sprockets, the lacquer had made the sprocket  surface ‘non slip’ and the sum of 4 non slip sprockets was absorbing a fair proportion of the available power. So I removed the sprockets from the clock. I found that it now takes less than 11 minutes to go from a running clock to having both balances out. I cleaned out the lacquer from the teeth and groove in the sprockets and everything is very lively again. I also removed the chain from the clock to give it a once over, it is in perfect condition. 

A change to the Robin remontoire chain.

New coupling screw, yellow arrow.

The chain is parted and coupling screw, first photo. In the next photo the copper wire which will be used to thread the chain around the Robin remontoire pulleys.

Buchanan writes:

The removal of the chain raised two questions.

1. What to do in the transport of the clock. I have not thought of a practical way to lock the Robin weights.

2. It requires a full centre celestial section dismantle to remove the remontoir chain.

In the end I popped a rivet apart on the chain to remove it from the clock.  It was a pleasant surprise to find out how difficult it was to separate a link. We have a robust chain. This raised the question of how I was going to reconnect the chain when I put it back in the clock. So I inserted a screw instead of a rivet in the open link so removal is now no problem.   

If when I ship the clock I remove the chain and leave in its place some copper wire for rethreading through the mechanism, then safe transport is not an issue. The only disadvantage is that you will need to reinstall it. It is not difficult to do, just a bit of a fiddle, when the balances are off, as the clock will be when you get it. I think if I find a method it will be simple. I thought it would be very awkward to replace the link screw but that was easy, as you rotate the double link like in photo 326, insert the screw through the untapped hole then place the single link over the screw and then rotate the last link to align the threaded hole with the screw. You are only dealing with one part at a time, not trying to hold three links in alignment, while threading a small screw through three holes.

Further key refinement.


The first photo shows the curvilinear winding key, the new design with ivy-type spurs that match those on the clock frames. The handle will later be replaced with Mammoth ivory left over from the creation of the tellurian Earth globe.


The first photo show some of the lead shot used as mass within the Wagner front hanging front weights shells. This was done to decrease the excessive swing of the pendulums from 9 to a more desired 7.5. This avoids too much expansion and compression of the pendulum modulating springs.

 The second photo shows a jig that Buchanan has created to accurately locate where in the front glass holes must be located to accommodate the various keyways and access ports.

This drawing illustrates where the access holes are positioned using the locating jig. It should be noted that the hole diameters are NOT those for the glass but for the INSIDE diameter of the keyways. This is supplied to the case maker. Once the keyways are made by the case maker, he will supply the the correct diameters for the outside diameters of the keyways, and the holes can then be drilled.

The difficulty here is the vertical location of the holes. The diagram shows the vertical measurement to the theoretical table top. But in reality one must locate how high off the table top the bare glass is from the table top and subtract those measurements to correctly get the hole positions. Once the case is delivered next month, I will be double check that measurement. This is more complicated than one might expect as the measurement is made up of where the glass is within the separable glass brass surround in addition to the skeletal frame that all of the glass frames fit into. This design allows one to individually remove any section of glass as needed for access to the machine as well as cleaning.

Buchanan will also make a template from plastic as a backup. A second, oversized template will be made so if there is an error, the template can be trimmed to fit the brass glass frame on site to match up with the machine. In this way, even if an error occurs, we can overcome that with the secondary template. This will also be needed if the actual glass is broken. I can say that another project Buchanan had completed for me in 2007 had such holes in the glass and it did get broken. The plastic template Buchanan has made at the time was invaluable in the recreation of the front glass with the keyways. 

The rating of the clock begins.

Buchanan writes: The blue line now represents how much the clock is losing or gaining. In interesting observation is that the slope of the line is getting shallower. I take it that the clock is settling down a little after a day’s running.

The bar graph (blue) gives us the length of each tick. 2 seconds is our target. Again you can see that it seems to be stabilising. The clock is starting to gain a little the last 6 hours. The green time is temperature, read of the scale on the right in Degrees F, we had a cool day yesterday.

The next day Buchanan writes:

The clock is still running perfectly. Attached is the record so far. When I checked the time against my phone it is 29 seconds slow, the 29 seconds is strange as the timer says that it should be gaining. We will keep up the monitoring.

This graph shows the rating as of the last day of July. The tests will continue through August.


The video above shows various close ups of the machine running.

Back Up Next