Back Up Next

Major design change from weight to spring-driven conversion, begin work and begin clock disassembly  - October 2020

A major design change is initiated to improve the functionality and practicality of the machine

Rex Swensen has avidly followed this project for some time, and has been involved with the creation of a series of clocks based on James Condliff's elegant curvilinear frame design from England c.1860.

He writes:  It is really too late for me to make this suggestion, but did you ever contemplate the use of Constant Torque Springs in your Astronomical Skeleton clock instead of weight driven? They have 27 active turns and do not require a fusee. Design life is 5,000 cycles, so for eight day running that gives a life of 100 years. In the overall design, I personally think they would fit in better and be more appropriate than the weights.

 We are using them in our replicas of the 1860 Condliff two barrel clock instead of the normal spring barrel and fusee. They come in a very wide range of torque specifications. The second photo is the same clock made by a colleague in the Sydney Clockmakers Society. He is about a year ahead of me, but then he had a set of water jet cut plates – but lots of edge filing needed.


The first photo shows the springs in the Rex Swenson Condliff replica. Here the regular spring barrel and fuse are replaced with one pair of constant torque springs used in the Condliff replica to power the strike train. Another pair is used on the opposite side to power the time train. Unlike a regular spring contained with a barrel, this spring is wound around a pair of spools. The length of the spring when fully wound is mostly on the upper spool and travels downward to fill the lower spool as it unwinds. The next illustration from a catalog describing a typical spring, here referred to as a motor spring since it has the constant force of a motor to drive the mechanism it is intended to power.

This idea came at just the time we had finished the orrery and realized that to get a good view of that assembly we would have to lower the entire clock by about four to six inches to get the best perspective. This reduces the weight drop and we would lose our intended eight day duration. The springs would eliminate those weights allowing us to have the clock at any height we wished.

I had my doubts about whether such springs would be powerful enough to drive the time train which required about 120 lbs (54 kg). But Rex and Buchanan were positive they would. At first I was reluctant to do away with so much beautiful brass represented by the weight shells. However, this idea of a spring-driven mechanism was contemplated by me before.


This drawing is dated July 24, 2005 and was one of several I had made contemplating different dial designs and other technical details of the clock. In this rendering I use conventional springs as well as fusees. (I was unaware of motor springs at the time). This was discarded since conventional springs would prove to be too large and if possible would contain an enormous amount of power that would represent a real danger of damage to the mechanism should they fail. I asked this same question about the event of a spring explosion with motor springs and apparently they do not pose this hazard. Had motor springs been considered it is quite likely this would have been the original design.

It is interesting how many details of the design from this very early conception were retained in the final machine.


Here is a view of the clock with its compliment of weights, photo taken in August. The stand has to be made of structural steel to support the clock's frame since it is not strong enough to support the weight set without distortion to the frame. The entire weight set is approximately 220 lbs (100 kg). The stand adds another 120 or so lbs (54 kg) and the glass case another 100 lbs (45 kg). By using motor springs we eliminate about 340 lbs (155 kg). The entire clock is about 180 lbs and with glass case, another 100 lbs is estimated to be 280 lbs (127 kg), so the entire setup would originally have been 620 lbs (280 kg) vs. 280 lbs. (127 kg) This change reduces that weight by 55% and makes the entire machine far more manageable to move. One can now place it on any suitable, yet sturdy table top. I think many people who have followed the project may be surprised to see this photo. Nearly all pictures taken of the clock have been from the level of the table top.


This is one of a very few videos taken of the clock with its compliment of weights. It was taken back in August 2016 before completion of the sun/moon and orrery complications. The planisphere is still a mockup here.

Rex writes: On the aesthetics aspect, personally I think that massive weights look out of place on a skeleton clock like yours. OK for wall clocks and long case clocks, where the case design is in harmony with the weight drive, but not for a skeleton clock potentially intended to live on a side board or similar.

I'd tend to agree with his comments. Buchanan saw the advantages of the motor springs early on, before I was convinced.

I write Buchanan in reference to the spring design: Just curious, do these things even have the torque to do the job? Or the length to last eight days?

Later in the day Buchanan writes: We need 1.8 kg at 10.5cm torque to wind the time train. This is 19kg cm or a single SR94 in the catalogue. Width 2 inches, largest diameter 3.34 inches, smaller than the largest diameter of our drum. We only need 24 turns of the 27 turns.

I need to check how we could fit them in to keep winding the same, with maintaining power (indicators). They are also stainless steel. I think we could do a clever, and cross two over between two winding arbours. Very little modification to the clock.  

Other advantages. : Much, much lighter. Far more portable. Much less strain on the clock frames and main arbour bearings. Completely sealed case. No lead to ship or cast. No danger of cables breaking. No Harrison balance type clock had weights. They are not unattractive either.  

The idea of a completely sealed case was also vey appealing. With weight cables one cannot have a sealed case.

The decision was made in about a week's time. I have to thank Rex for his contribution. 



Test rig to determine the torque characteristics of the motor springs. Next photo shows the sliver cut from spring hub to provide a constant radius at the attachment point.


Video of a sliver being cut from the spring hub with EDM machine to make a constant radius surface where the spring will be attached to the hub ensuring there will be no stress to the spring at the attachment area. The cut sliver will then be used to clamp the spring onto the hub.

Buchanan writes: I am working on the spring barrels today.

I built a quick test rig for the springs and they match the design specification well. So we have about 10% more power than the present weights.  This is looking very good. Third photo.

I have started to machine the spring barrel centre sections to have a lead in for the spring ends.  I have cut a constant radius curve so as not to create any stresses in the spring ends.

This is shown on the last photo where a sliver is cut from the brass cylinder blank and will act as a clamp for the spring attachment. Not even the manufacturer of the spring specs for this but it is best practice. Another example of Buchanan's “Gilding the Lilly” with respect to his work.

I realised that there is another advantages in these springs over against conventional springs: You can visually inspect them every time you wind them. If one was to start cracking I would expect it to happen over a few cycles.  There is a very good chance that one would see, or, hear something out of place if it started to crack. My guess is that if a crack were to develop it would split immediately without the chance to catch it via an observation. In fact I would hope my face would be nowhere near the spring at the moment of that event.

I also have the barrel flanges roughed out. I will be ordering the bearings tomorrow. I will be installing hybrid steel ceramic bearing as I go.

If all goes well, tomorrow I will make sure that I have a good video recorder working and will start the main strip down, as I need dimensions from the clock it’s self for the new spring barrel arbours.

At this point the machine is being taken down to its component parts for the final miscellaneous fabrication, polishing and finishing many larger parts including frames are being lacquered. In this video, Part 1, Buchanan illustrates the process of disassembly prior to shipping. This involves the removal of most of the clock's complications which have been designed as modules for just this purpose.

In this video, Part 2, Buchanan illustrates the process of disassembly prior to shipping and in this video the process of complete disassembly down to the individual components begins which is necessary before the final finishing process can begin. The three main modules, time, celestial and strike are removed from the base, (the strike has two trains in one hence four main drive barrels in the base). Clearly this is not a simple process, however, if this were designed as conventional clocks with all the train wheels between one set of plates , it would be near impossible.

In this video, Part 3, Buchanan illustrates the disassembly of the clock base. This contains the four main drive wheels and the state of wind complication for each. Besides the fact that this disassembly is necessary for the final finishing stage of the clock, the barrels must be re-engineered to power the machine with constant force motor springs in place of the current conventional weights. There are many advantages to this decision, the elimination of about 250 lbs. in weights and another 150 lbs for the substantial stand to hold the machine. That stand was specially designed to support the base frame as it could not handle the weights on its own without deflection. Now the machine can be placed on any substantial table top at any height and still retain a full eight day duration. The entire mass of the project has been cut by over 60%.

Notice the clever engineering that Buchanan used to make the frame look as if it were molded from one piece. No screws or attachment points can be seen throughout the base frame. That method could only work visually if the frame parts were machined to fit perfectly together. This is something few will ever see or comprehend when observed in the future. A puzzle for them!


Buchanan writes:

I also ran all the clock over the scale in pieces. It is as follows: main frame, carrying frame, time train and strike train weigh 154 lbs.  The celestial train is 19.8 lbs, so the bare clock as you will have to carry it including the carrying frame weighs 180lbs max. or 90 lbs per person.  It feels heavy but not dangerously so, as the carry frame is ergonomic.

Other components are; Bronze balance spheres 26lbs, bells 2.2lbs, Planisphere 4.2lbs, Calendar, 1.1lbs, Sun moon dial 2.2lbs, Tellurian 0.66lbs, Orrery, 1.75 lbs. 2 balance assemblies without balls 4.4 lbs for both, extras(thermometer and transfer gears escape wheels etc.  6.6 lbs.

This gives us a clock weight of about 230 lbs. of which 50 lbs is easily removable.



Buchanan writes: This is the clock in all its major sections. I added up the length of video time and the complete dismantle took 1 hour and 13 minutes of video.  With all the set up and stands to be made as well as downloading etc. it has taken more than a day.

These photos almost makes the machine look less complicated than it is. Everything fits onto a simple table. But obviously if one were to zoom into a close up of the individual components the complexity would come to the fore.


Buchanan writes: Attached are photos of a modification to the Robyn feed pawl. As it has a slight dog leg bend for clearance, it is slightly imbalanced so I made a counterbalance so that it lies in a neutral position on the c shape bearing and roller. I have, in other words lowered its centre of gravity to make it stable.

These two photos show the reworked feed pawl. The poising weight is a beautifully turned tear-drop shape and is attached to another well-shaped attachment to the pawl via a blued rod and is adjustable and secured by a tall-blued screw. The dog leg mentioned is the slight curve where the poising weight rod is attached seen in the second photo.


This photo shows the feed pawl posing weight assembly within the circled area and the weight indicated by the arrow.


This photo shows the opposite end of the pawl resting upon a guide fork (currently temporarily attached to the frame via super glue) and has a clevis at the end, partially hidden behind the frame upright, attaching the pawl to the remontoire assembly.

Buchanan writes: Today I designed the spring barrels.

As usual space is tight, particularly on the time barrels and our stop work, state of wind indicators and maintaining power all complicate things.

Each barrel assembly is completely different to the next, this is because the great wheels alternate in front and behind each other at the same time as we change spring sizes.  I also have to take the drive to the stop work through the idler spring barrel if the drive spring is in the front and the drive to the bevel gears has to go through the idler barrel when the drive spring is at the back. Then through it all there is the main arbour which carries the ratchet wheel at the back of the clock. So we have a triaxial drive at some point on each main arbour. I am also limited by the bearing sizes that are available as stainless steel ceramic hybrids. I feel like a juggler that has just finished a juggling 7 hour marathon. See CAD drawing above.

I am wondering if the weights would have been less work (I am being silly here, not complaining.) It will be very interesting to watch when the clock is wound.


It’s unfortunate that we will not be using 100% ceramic bearings, but they are nearly ten times the cost of the hybrids and more critically are not available in all the sizes we need. Still these bearings are far better in performance and longevity with very little lubrication and the metal cage rings are stainless steel; far superior in all respects to conventional bearings. Buchanan says that even if the oil dries in 20 years, there will be a coating that will perform well given the dissimilar materials of stainless steel cages and ceramic balls, compared to like materials. Furthermore, we are using shielded bearings that have integral dust covers and in many cases those are covered by a faux jeweled chaton for additional protection against contamination. When this project was first conceived sixteen years ago we had not even considered bearings that were not made of conventional metal throughout. Ceramics were confined to NASA and military applications and even the hybrids cost ten times what they are today.


Buchanan writes: I have the spring clamps complete. Today I am making the end caps and bearings for the idler or storage drums.  I have almost everything of the lower frame dismantled. Making those lead in curves for the springs is what Buchanan referred to earlier as a “constant radius curve” for the lead in of the springs.


Buchanan writes: Here is a photo of the new spring barrels with the extended length winding squares, (First photo). There are 28 bearing fits, 8 threads, 8 part fits and 8 squares in total as well as many length critical dimensions. I just have to cut 4 small key ways and they are complete. (It was decided some time ago that all of the winding arbors needed to be lengthened because the two inner arbors were too short; being adjacent and too close to the planisphere frame where the winding crank was inserted. The other two were lengthend to match.) Today I am working on the actual spring drums and spring clamps. I also have to order some more stainless steel for the drum parts.

The spring barrels are progressing well, second photo, I have all 16 flanges complete.


Tomorrow I will start the sleeves that carry and connect everything. Then we should see it all start to come together. When the barrels are assembled I will add turned decoration where applicable. Everything looks very square right now. I have also ordered the first batch of bearings. I have the stainless steel sleeves that carry the storage barrels complete, fourth photo, I have to make two more stainless steel spacers and 150 odd screws and I can start to assemble the barrels. I have ordered the first batch of bearings. They should be here soon. Coming from the USA.


The holes for the storage barrel flanges are now tapped and drilled.


A completed main wheel. Each has two barrels. The barrel’s main spring is the larger diameter with the screws that secure the “constant radius curve” section that secures the end of the spring. The smaller barrel is the take up, or storage barrel from the spring that will be spooled onto that barrel from the adjacent main wheel barrel, it too has a constant radius curve on the opposite side for the spring’s attachment point. Each barrel moves independently since the springs they attach to will spool at different rates over the short term, but over the entire eight day duration should all be unspooled to nearly the same degree. Not unlike how the going barrels or winding drums operate on a quarter strike clock where the time and strike trains all move at different rates and times, but over one week's time the weights are at about the same level when rewinding needs to occur. 

It is interesting to note the similarity in the pair of barrels of different diameters to hold the supply spring and the take-up spool. Look at the old barrels standing on the table below, these too were of a dual diameter design, but used in a completely different way. It is a serendipitous that such a dramatic redesign could be accommodated without having to dispose of the entire main wheel assembly.



The main barrels and take-up spools for the conversion to constant force springs are ready for finishing and reassembly. The four old weight barrels are seen in the background

Back Up Next