Queen's Prize 2018 Project: Table Clock Movement

Note: I've been putting off writing this up for a while now, since I already did it once for the documentation. But, I need to catch up before I have to write up my next project. So, this blog is basically just my final pictures and a web formatted version of my documentation.

Also worth noting, this project won the Judges' Choice award.


Documentation

Background

One of the first steps in this project was figuring out what the pieces inside the clock were called and what they were supposed to do for the clock to function properly. From my reading, all mechanical clocks have four categories of components: the power source, the escapement, the output, and some method to tie the first three components together.

Power Source

In order for the clock to go, it needs something powering it. In an older water clock, the power came from potential energy in the water. In modern electric watches, this is represented by a battery. In the Renaissance era clock I was trying to recreate, the power came from the mainspring.

The mainspring is a piece of thin metal ribbon that has been coiled into a spiral. By fixing either the center or outer edge of the spiral and twisting the coil, the mainspring can store rotational energy. This is used to rotate the gears in the rest of our clock. The mainspring may have been used in clocks as early as 1430, first appearing in a clock made for Philip the Good, Duke of Burgundy.

The mainspring introduces its own problems to accurate timekeeping. A spring’s force is function of how far it is stretched, so as the mainspring winds down, the force applied to the gears is reduced. During this time period, there were two devices used to try to even out the force.

The first of these is the stackfreed, as seen on the right. This device mainly appears in German clocks coming out of Nuremberg. The stackfreed uses a second spring and a helical disk to apply a counterforce to the mainspring. When the mainspring is fully wound, the secondary spring is pressing against the tallest portion of the helical gear. This position produces the most force on the gear and thereby the most friction resisting the force of the mainspring. As the mainspring winds down, the stackfreed moves down the helical gear until it is at its lowest point, providing the least resistance. Since this method relies on friction instead of mechanical advantage, it was a less efficient option to the fusee and eventually died out.

The other method of evening out spring force in this period is the fusee and is used by the clock on display. It is the tall spiral cone sitting atop the largest gear. When the clock is first wound, the cord attaching the mainspring to the fusee is wound up to the top of the cone. At this point, when the spring is generating the most force, it is pulling against the fusee where it has the least mechanical advantage. As the spring uncoils and loses force, it unwinds down the fusee to the larger diameter section, gaining mechanical advantage. Through this mechanism, the net force output of a correctly designed system remains about the same throughout the period between windings of the clock.

Later mainspring powered clocks worked around the issue of reducing force by having a much larger mainspring. By doing this, only a small fraction of the range of the spring was used, keeping the force applied relatively constant. This, coupled with advancements in metallurgy eventually allowed watchmakers to drop both methods and move to the going barrel where the first gear is directly driven by the mainspring.

Escapement

The escapement is what makes the clock tick. That is, it divides the constant force of the power source into steps. In modern clocks, a quartz oscillator is used to create a constant frequency that drives the clock. In this clock, we are using a verge escapement. There are three parts to a verge escapement: the crown wheel, the pivot, and the balance wheel. The crown wheel is called such because of the pointed teeth coming up off the top making it look like a crown, as opposed to a normal gear where the teeth extend out from the side. The teeth of the crown wheel are caught by two flaps attached to the pivot that extend out at a ninety degree angle from each other. As the crown wheel rotates, one flap is continuously caught in the teeth. As the teeth push the interfering flap out of the way, it causes the other flap to rotate back into the path of the teeth as the first flap clears the gear. This constant back and forth is slowed by the angular momentum of the balance wheel, creating a steady tick.

It is also important that the crown wheel have an odd number of teeth. If the number of teeth is even, they will be aligned across from each other on the wheel. This means that both the top and bottom will engage the flaps on the balance wheel pivot, binding the system instead of pushing the pivot back and forth as intended.

The image on the prior page includes a balance spring, which is the mainspring-like coil just below the balance wheel. This improvement began to appear near the end of the 17th century, and as such is not featured in the escapement I recreated. The spring force helps regulate the oscillation of the balance wheel and make it less sensitive to changes in force output from the mainspring as it winds down.

Gear Train

In mechanical watches, the power source and escapement are connected by a gear train. The number of wheels and the number of teeth per wheel varies based on how long the clock is designed to run for and what particular maker created the movement. In general, the gears are simply referred to by their number: first wheel, second wheel, and so on, with the first wheel being the one driven by the power source. There are some special names in use such as the center wheel, so named because it is designed to rotate once per hour and as such is located in the center of the movement where its pivot can extend through the front plate and drive the minute hand. However, the early movements I am studying do not have wheels in the center and do not have minute hands, so I will keep to the general naming conventions.

All of the gears in the train have an attached pinion. This is a smaller gear attached to the same pivot as a wheel that meshes with the prior wheel in the train. For example, the second wheel has a pinion attached to it that the first wheel meshes with and drives. If we suppose the first wheel has sixty teeth and the second wheel’s pinion has six teeth, then for every complete rotation the first wheel makes, the pinion will rotate ten times, which also makes the second wheel rotate ten times. We continue gearing up our motion through the clock so the clock will be able to run itself for hours or more with just a few turns of the fusee.

Output

In order for the clock to be useful, it needs something to translate the gear’s position into the current time. To do this, the output will tie into the gear train at some convenient point. In a modern clock, the center wheel, which turns once per hour, has a pivot that extends through the front of the clock to drive the minute hand. A gear with a 12:1 ratio is fitted onto that pivot on the outside of the movement to drive the hour hand, so that for every twelve 60 minute periods, the hour hand will move around the face of the clock once.

As I mentioned, the early clocks I’m recreating do not have a center wheel since they do not track minutes. Unfortunately, all the museum documentation of movements I found photographed the movement upside down. This makes sense as many of the interesting parts like the balance wheel, the mainspring adjustment, and the winding point are on the underside. But, this hides how the output ties into the gear train. I eventually found a museum’s project that took a CT scan of one of these clocks that showed the first wheel extending through the front plate of the movement and ending in a lantern gear that drove a gear attached to the hour hand. While this project was actually focused on determining whether the clock was a counterfeit or not (they determined it was), the fact they had to resort to a CT scan to figure it out made it good enough for me to go with it, especially given the lack of contradictory evidence.

Movement Design

Besides all the small intricacies of fitting things together, there were three major things I needed to figure out before I started construction work on the clock components: the gearing, the escapement’s frequency, and what direction it was going to rotate. Since I was copying existing movements, determining the direction of rotation was as simple as laying out the gear train and mapping how the gears interact. It was still important to determine this early to make sure the fusee’s windings went the correct direction so I could cast all my parts in one run.

The frequency of the escapement took a while, but I was eventually able to dig up a listing from a jeweler that was selling a 16th century clock that was very similar to the one I was trying to build. They happened to include a video of the clock running with the listing which I was able to use to count the beats and determine it was running around 120 BPM. The ticks were somewhat uneven, as it is a pre-balance spring escapement and these type of movements also tend to speed up with wear, but like with the clock output piece, I decided this data was good enough as I am unlikely to get anything better. Unfortunately, the listing has since sold and the jeweler removed the videos.

When I began designing the gearing, I actually wound up dodging a lot of the math in the name of exact accuracy. Instead of calculating out the gear train, I used the museum photos of clock movements to count the number of teeth in the gears. Since most of them did not have good angles, I was able to use the spokes on the gears to divide them into three or four segments, count that segment, and multiply to get the results.

I eventually found the photo on the prior page and was able to get similar counts from it. Working back through the gearing with the knowledge I wanted my escapement to tick at 120 BPM, the train I reverse engineered had the first wheel running slightly faster than one rotation per hour. I increased the second wheel’s tooth count from 44 to 48, which brought the overall train close enough to my targets that it would be accurate enough, assuming it ran.

 As I come back to it now, it appears I made a math error somewhere. With the same gear train, it could be corrected by slowing the escapement to around 98 BPM by adding more mass to the escape wheel, so it would not have been an uncorrectable error.




Pinion Teeth
Gear Teeth
Times Pinion Must Rotate for One Gear Rotation
Rotation Time @ 120 BPM

@ 98 BPM
Crown Wheel

13

6.5 seconds
7.96 seconds
Third Wheel
6
36
6
39 seconds
47.75 seconds
Second Wheel
6
48
8
5 minutes, 12 seconds
6 minutes, 22 seconds
First Wheel
6
56
9.3
48 minutes, 21 seconds
59 minutes, 13 seconds


Regardless of my math errors, my final plan for the movement is below.

Process and Materials

Early clock movements were originally made of iron, but near the end of the 16th century, they began to be made of brass instead. I chose to use brass for my components, as it is easier to work with my existing tool set.

Period gears would have been cut by hand with files. Originally the teeth would have been measured and plotted by hand as well, but later a specialized divider machine was made to plot gears consistently. Since this project was a learning experience and I did not want to invest the time to cut a gear by hand only to learn I messed up some part of the design as to render it useless, I chose to cast the gears. I was able to use a software tool to generate vector images of the gears I wanted. I was then able to take those plans and use Blender to create 3D models which were then printed in a castable resin. While I was creating the gears, I also modeled the fusee since I would not be able to fabricate that piece easily either.

While those parts were being cast, I cut additional pieces such as ratchets, the top and bottom plates, and supporting pillars from sheet brass. Once I cast the parts, I then cleaned them up and drilled out the center holes to prepare them for mounting. The various wheels had a piece of straightened wire soldered through the center hole to form the pivot, then pinions were soldered onto the pivot at the appropriate height to form the gear train.

The fusee had a ratchet soldered onto the base and a hole was drilled in the center of the ratchet so the pivot of the first wheel could pass through it. This made the two pieces act as one column in the movement while allowing the fusee and gear to turn independently. A lever and spring were fit onto the first wheel to form the other half of the ratchet, allowing the fusee to be wound up, then engage the first wheel and turn the rest of the clock as it unwinds.

A large amount of time was spent laying out the wheels on the top and bottom plates and getting them to spin freely. I made a mistake and mounted the second wheel about a millimeter lower than I planned. What this did was prevent me from having the gear on the front that drives the hour hand pass through the plate and be held by a clip on the backside, as would have been done by these early century clocks. Instead, what I wound up doing was soldering four posts on the outside of the movement, supporting a cut cross of calatrava. Clocks that only tracked hours did not typically have a supporting plate out front, but later clocks with more complicated gearings that also calculated celestial positions did have additional gears here.

After my wheels were in position, I was noticing some additional friction and binding between them. To try to fix this, I wound up going in with hand files at first and later a flex shaft to try to get them to run smoothly.

Once the wheels were positioned and altered, I focused on the mainspring. This required cutting and soldering the outer case together, as well as the center pivot. The spring itself was actually scavenged from a modern wind up clock, as sourcing the appropriate metal ribbon to make it myself would be prohibitively expensive and I did not particularly care to try to fabricate it myself. The spring has a pear shaped hole in the spring at the inner end of the spring and doubles back on itself at the outer end. To attach this, the center pivot had a small hook made from sheet brass soldered on that catches the inner hole and a post was soldered into the case on the outside that the doubled back section fit over. To assemble this, the outer edge of the spring was placed around the post, then the whole spring was wound into the case. The center post was then inserted through the case and turned so the hook in the center catches the other end of the spring.

The upper end of the mainspring’s center post was forged square so that a ratchet wheel could be fit over the end of it. This allows the mainspring to be brought up to tension after initial assembly. This post would not be touched by the final owner; winding is handled by turning the top of the fusee..

The outside of the mainspring’s case has a loop soldered onto it. In period, a catgut cord would be tied to this loop, wound around the mainspring’s barrel, and attached to the base of the fusee. When the fusee is wound, it would pull the cord off the barrel and fall into the groove on the fusee. Since my cat was not being overly annoying during this project, I chose to use a 1/16” leather cord instead which is a similar size to what I found.

Finally, the ends of the supporting posts for the case, the posts for the front plate, the ends of the mainspring pivot, and attachment points for the balance wheel’s holder all had holes drilled in them so they could have wire pins inserted to hold the parts in place.

Project Retrospective

Now, sadly, this clock does not run. If I remove the escapement’s crown wheel, the first three wheels will spin freely, but that last step of gearing seems to be pushing it too far. Unfortunately, I think I’ve learned enough that I’m going to feel compelled to try this again. At some point, I think I want to try to include hand cutting at least one gear, but that is in the far future.

If I were to attempt this project again, I would likely change the gearing so the first wheel rotated with a period closer to two hours instead of one. This is because most of the clocks were designed to at least be twelve hour clocks, so they would be wound in the morning and evening. In addition to their twelve hour runtime, they would typically have four to eight hours of reserve time. By comparison, my fusee has twelve windings on it and will unwind one per hour. So at most, my clock could be considered an eight hour clock.

I learned after designing my gears about the idea of backlash. This is essentially incorporating some play into a gear system so it does not have an exact fit. I tried to cut some into the gears after the fact, but that also altered my tooth profile significantly. While it is not desirable in some gear systems, since clocks move slowly in one direction, it is not a negative here and can help prevent binding.

Some form of grid system to help plot out the top and bottom plates would be helpful to ensure the pivots are square. I realized this too late and attempted to do it by sight and line things up by hand. Being able to set an origin and measure out coordinates in millimeters would be an improvement. Another possibility would be to stack both plates on top of each other and drill both holes at the same time so they are lined up and insert a bushing later, but I did not see bushings in the examples I used.

I used sheet brass for the supporting columns in the movement; in period they typically were more substantial and decorative. They also provided mounting points for arms that supported the internal gears. I would likely add these to my “to be cast” list.

Overall, I consider this project a success. My original project pitch was that I would have an “interesting failure,” which I think this covers. The initial research and understanding were somewhat painful to get together, but now that I have that, I should be able to focus in on the components on a later attempt instead of winging some things or relying on software tools.

Bibliography

“A Treatise on Watch-Work.” Google Books. Accessed September 8, 2018. https://books.google.com/books/about/A_Treatise_on_Watch_work.html?id=7DcDAAAAQAAJ.
“Found: The World’s Oldest Clock With A Fusée, At The British Museum (And Other Cool Stuff).” HODINKEE. Accessed September 13, 2018. https://www.hodinkee.com/articles/worlds-oldest-clock-with-a-fusee-british-museum.
“Image Gallery: Table Clock / Lunar Clock / Clock-Case / Astronomical Clock.” British Museum. Accessed September 13, 2018. http://www.britishmuseum.org/research/collection_online/collection_object_details/collection_image_gallery.aspx?partid=1&assetid=1613045729&objectid=55468.
“Image Gallery: Table Clock / Spring-Driven Clock / Clock-Case.” British Museum. Accessed September 13, 2018. http://www.britishmuseum.org/research/collection_online/collection_object_details/collection_image_gallery.aspx?partid=1&assetid=1332390001&objectid=55419.
“Is This The World’s Oldest Known Watch? A Peter Henlein Mystery From 1505 | Quill & Pad.” Quill & Pad, December 15, 2014. https://quillandpad.com/2014/12/15/the-worlds-oldest-watch-a-peter-henlein-mystery-from-1505-solved/.
Milham, Willis Isbister. Time & Timekeepers: Including the History, Construction, Care and Accuracy of Clocks and Watches, 1942.
“Nuremberg 16th Century Clock Watch.” Francis Jewellers. Accessed July 25, 2018. https://francisjewellers.com/products/nuremberg-16th-century-clock-watch.
“The Cyclopædia, Or, Universal Dictionary of Arts, Sciences, and Literature.” Google Books. Accessed September 8, 2018. https://books.google.com/books/about/The_Cyclop%C3%A6dia_Or_Universal_Dictionary.html?id=-XxMAAAAMAAJ.
“The Museum of Science and Art.” Google Books. Accessed September 8, 2018. https://books.google.com/books/about/The_Museum_of_Science_and_Art.html?id=480EAAAAYAAJ.
“The so-Called Henlein Pocket Watch - Google Arts & Culture.” Google Cultural Institute. Accessed September 12, 2018. https://artsandculture.google.com/exhibit/the-so-called-henlein-pocket-watch/CQLyeqppGOtCLA.
White, Lynn Townsend, ) Lynn White (jr, and UCLA Center for Medieval & Renaissance Studies Lynn White. Medieval Technology and Social Change. Clarendon Press, 1962.

Image Sources

“The Cyclopædia, Or, Universal Dictionary of Arts, Sciences, and Literature,” Google Books, 435, accessed September 8, 2018, https://books.google.com/books/about/The_Cyclop%C3%A6dia_Or_Universal_Dictionary.html?id=-XxMAAAAMAAJ.https://books.google.com/books/about/The_Cyclop%C3%A6dia_Or_Universal_Dictionary.html?id=-XxMAAAAMAAJ“The Cyclopædia, Or, Universal Dictionary of Arts, Sciences, and Literature,” 435.

“The so-Called Henlein Pocket Watch - Google Arts & Culture,” Google Cultural Institute, accessed September 12, 2018, https://artsandculture.google.com/exhibit/the-so-called-henlein-pocket-watch/CQLyeqppGOtCLA.

“A Treatise on Watch-Work,” Google Books, accessed September 8, 2018, https://books.google.com/books/about/A_Treatise_on_Watch_work.html?id=7DcDAAAAQAAJ.

“The Museum of Science and Art,” Google Books, accessed September 8, 2018, https://books.google.com/books/about/The_Museum_of_Science_and_Art.html?id=480EAAAAYAAJ.https://books.google.com/books/about/The_Museum_of_Science_and_Art.html?id=480EAAAAYAAJ“The Museum of Science and Art.”

“Is This The World’s Oldest Known Watch? A Peter Henlein Mystery From 1505 | Quill & Pad,” Quill & Pad, December 15, 2014, https://quillandpad.com/2014/12/15/the-worlds-oldest-watch-a-peter-henlein-mystery-from-1505-solved/.https://quillandpad.com/2014/12/15/the-worlds-oldest-watch-a-peter-henlein-mystery-from-1505-solved/“Is This The World’s Oldest Known Watch? A Peter Henlein Mystery From 1505 | Quill & Pad.”

“Image Gallery: Table Clock / Lunar Clock / Clock-Case / Astronomical Clock,” British Museum, accessed September 13, 2018, http://www.britishmuseum.org/research/collection_online/collection_object_details/collection_image_gallery.aspx?partid=1&assetid=1613045729&objectid=55468.http://www.britishmuseum.org/research/collection_online/collection_object_details/collection_image_gallery.aspx?partid=1&assetid=1613045729&objectid=55468“Image Gallery: Table Clock / Lunar Clock / Clock-Case / Astronomical Clock.”

“Image Gallery: Table Clock / Spring-Driven Clock / Clock-Case,” British Museum, accessed September 13, 2018, http://www.britishmuseum.org/research/collection_online/collection_object_details/collection_image_gallery.aspx?partid=1&assetid=1332390001&objectid=55419.http://www.britishmuseum.org/research/collection_online/collection_object_details/collection_image_gallery.aspx?partid=1&assetid=1332390001&objectid=55419“Image Gallery: Table Clock / Spring-Driven Clock / Clock-Case.”

“Found: The World’s Oldest Clock With A Fusée, At The British Museum (And Other Cool Stuff),” HODINKEE, accessed September 13, 2018, https://www.hodinkee.com/articles/worlds-oldest-clock-with-a-fusee-british-museum.