* GadgetBuilder.com *
Last Modified: 18 August 2010 01:18 MST
Click to Enlarge
After reading Dr. Philip Woodward's "My Own Right Time", I was inspired to build a copy of his gearless clock. John Wilding's construction book for this clock provides details that simplify construction although (as usual) I made some "improvements". The minimal number of parts and their unusual construction made this a delightful project, enhanced by the fact that this clock obviously wanted to run.
The picture above (left, with yellow string replacing the stylish pink drive string) was taken during a trial of the weight/drive system. The temporary dial is paper, just taped in place for evaluation. The pictures above right show a practice case made from poplar, done to improve my (non-existent) woodworking skill. The clock isn't complete but it is getting closer, lots of little details left.
Dr. Woodward's gearless clock design is novel in many ways. It is a modular design where the various functions are easily recognized and connect to other functions in a way that is ingenious but clear on careful inspection, leaving the observer with a "wish I'd thought of that" feeling as each facet of its operation is understood. Some of the connections between functions are made up of wires with counterbalance weights so gravity returns them to their default position - a most unusual implementation method in a clock. This design strips away the mechanical complexity of a typical clock, leaving the essential details implemented in an unconventional but very functional way; it's entertaining to watch it work and a challenge to understand the operating details.
The pendulum in this clock connects directly to a ratchet wheel via a pawl which indexes this count wheel for each swing of the pendulum; a backstop pawl ensures the wheel indexes in the correct direction at each swing. There are 48 teeth on this count wheel, one of which is deeper than the others; this deep tooth causes the pawl to move a deflector lever connecting to the minute module.
The minute module has a hook connected to the pendulum, where the deflector lever noted above causes this impulse hook to engage a pin on the minute wheel. The timing is such that the hook first pulls the minute pin wheel backward a half minute, releasing the escape. This allows the pin wheel to then pull the pendulum forward by 1.5 minutes, limited by the escapement; pendulum inertia carries the hook a little farther, releasing the hook. This hook and the lever which originated the pendulum impulse are both counterbalanced so they assume their disconnect position until the count wheel completes another turn. This gearless clock is unusual in that it impulses the pendulum only once a minute; the interaction of the impulse hook and the escapement is most ingenious, producing a sound similar to that from a common pendulum clock arming to strike.
The escapement remained a bit of a puzzle to me until I built it and saw it operate. Each minute the pin wheel is pulled backward slightly; this releases the pin captured by a lip on the counterbalanced escape so the escape pops upward, landing against the next pin; this pin then slides along the shelf as the pin wheel rotates until it drops through the slot and is captured by the lip, completing the cycle. The escape is shown in a picture below but it is best to see Dr. Woodward's book for a good explanation - or just build one...
Power to drive the clock is supplied by a weight attached to a string wrapped around a pulley on the minute arbor. This clock isn't "wound" in the strict sense - the drive weight is simply lifted and the resulting slack in the string is taken up by a second (jockey)weight. This second weight is a fraction of the weight of the drive weight, causing the string to bind in the pulley, much as a bollard is used to amplify friction when controlling a ship at a pier. Pulleys are used to direct the drive string path. While winding by lifting the weight is quick and easy, lifting the weight too rapidly can pick the string off the pulley so one must proceed gently while "winding".
The daisy wheel motion work module is driven by the minute shaft and is easily removable by taking out the retaining taper pin on the minute shaft.
I'm still working on details, some of which are major (dial, weights) and others minor (screws). The biggest thing left, assuming it continues to run, is to build a case for it. This is a work in progress; some pictures and text follow, more will be added later.
As with many of my projects, this clock was made mostly from scrap plus a bit of purchased material. I had a nice 4" long piece of 1.25" polished stainless tube so I filled it with chopped up lead; weight was 2 pounds - I used this initially as the pendulum (shown in some test pictures) even though it was only half the weight specified by John Wilding. In many clocks pendulum weight isn't critical but here it is because this clock uses larger, less frequent impulses to keep the pendulum moving. With this lighter pendulum storing and dispensing this energy, the single impulse each minute causes a larger than acceptable change in pendulum swing. When the swing is too large the pawl gathers two teeth on the count wheel for a couple cycles after the impulse causing the clock to gain time. Reducing the impulse enough to eliminate this makes the clock unreliable. So, I found a piece of 2" steel round and made a pendulum to John Wilding's plan - makes operation reliable and tolerant of reasonable variation in drive weight. With the original light pendulum the decay in pendulum swing between impulses was easily seen but it isn't obvious with the heavier pendulum.
The pendulum rod is fiberglass composite with the glass threads parallel to the length, another treasure from the scrap pile. I assume its coeffecient of expansion with temperature is less than a steel rod but can't be certain pending experiment. The parts holding the count and impulse pawls were from aluminum except the counterbalance for the impulse hook is brass - my notion was to minimize the weight above the pendulum which the fiberglass rod and aluminum parts help accomplish. The aluminum pawl holders didn't look good on the pendulum rod so eventually I replaced them with brass holders. In replacing the count wheel pawl I used 0.025 music wire and added a partial counterbalance to minimize the sound as the pawl drops.
My assumption was the pendulum period will lengthen with increasing temperature so I drilled the top half of the pendulum to just pass the rod and drilled the bottom half larger so an aluminum tube surrounding the rod could be installed. A pin through the pendulum rod below the pendulum supports the aluminum tube; by supporting the pendulum at its center, expansion of the pendulum itself with temperature should be neutralized so the aluminum's expansion need only cancel the suspension spring and fiberglass rod expansion (it says right here :-) The thin aluminum tube was filed to length (1 thou = 2.8 seconds) to set the rate a few seconds a day slow; adding a coin (or other small weight) on top of the pendulum will raise the pendulum's CG and speed it up slightly - easier and more precise than a rating nut. Similar to the way Big Ben is adjusted...
The escape pin wheel is the most difficult part to build in this clock. Fortunately, my mill's DRO has a "bolt circle" function which made it straight forward, if tedious. Here, the 60 holes (one for each minute) are being spotted using a shop made 0.022 spotting drill, spade type. This was followed by drilling with a 0.029 twist drill.
Click to Enlarge
The escape wheel was pinned using a shop made tool having a hole 2 thou larger than the 0.031 pin. The magnet on the side of this tool holds the pin in the tool until it is set. When the pins were pressed through the wheel, the steel block stopped them at a uniform point so cleanup of the back per Wilding's book wasn't needed. I cut all the piano wire pins to length in the "whack-bang wire cutter" from my prior clock project, then held them in a pin vise mounted in the 7x12 chuck, filed the ends flat and chamfered them, (the insert end more than the other) with a diamond file.
The completed escape pin wheel.
Escapement test - the escape is gravity operated so moving the pin wheel by hand cycles the escape. The count wheel/backstop and pendulum connection were tested previously (no picture) where the pendulum will operate the count wheel for a little less than 2 minutes.
A test driven with weight hung from pin wheel, runs for 15 minutes - until the weight hits the bottom support. This simple arrangement (a nut, washer and bent paper clip) allowed adjusting the basic operation of the clock without the complexity of the weight drive system. This test weight is also a convenience while threading the drive string through the works, keeps the escape from jiggling around - but isn't needed once the sprag is added.
Ran the drive section for a couple days without the motion work, then added it as shown below. Surprisingly, the motion work didn't seem to affect operation - pendulum swing didn't diminish due to the added load, as I thought it might.
The weight drive required some tinkering to get it right although the weights per Wilding's book provide a good starting point. I used 1.5 pounds for the drive weight and 120gm for the jockey weight. The nylon string slowly slipped on the pulley so I increased the jockey weight but it needed about 1/3 the drive weight to stop slippage. Winches on sailboats often have roughened surfaces in their scaled up application of this principle so I roughened the sides of the "V" drive pulley to improve traction, which worked well. This was done by mounting the pulley loosely on a small round then rolling the pulley on this shaft using pressure on a needle file in the "V"; this left an imprint of the file teeth on the sides of the "V", not deep but enough to improve grip. Fine carbide paper was then used to ensure there weren't any sharp snags left by this operation.
The clock takes some time to stabilize operation after startup. Initially, one deflects the pendulum so the impulse hook is moved one escape wheel pin space from rest and releases it. It then takes several impulses (minutes) for the pendulum swing to reach equilibrium. Generally, after 3 or 4 impulses one can tell from watching the count wheel pawl and backstop pawl vs the teeth whether the impulse is correct - this from several hours experience watching how it works, of course.
This picture shows the daisy wheel motion work assembled with the temporary hands in place. This emphasizes the modular nature of this motion work, where taking out one taper pin allows it to be removed as a unit. A video of the daisy wheel motion work.
Below is the daisy wheel motion work partially disassembled. On the left, the minute collet with the cam and spacer; I lightly press fit the cam rather than using a setscrew. Next is the daisy wheel - the stem was soft soldered in place. The hour collet with the pin wheel, again pressed on. The hour hand collet and finally the nut for the minute hand. The aluminum hands for testing just press in place so they're not really secure. I added a spacer to take up slack between the hour collet and the minute nut, slack caused by mis-reading Wilding's dimensions for the hour collet. Never having seen a daisy wheel motion work previously, it was an education watching this one operate. While it looks simple and there aren't many pieces, I made 5 cams to get the throw right before filing the daisy petals to fit... as the world's slowest machinist, it took over a day to make these few parts. Again, the DRO's bolt circle function was a big help.
A nice feature of the daisy wheel is that time can be set forward or back simply by turning the minute hand, unlike some clocks where time must always be adjusted forward. Little force is needed to drive the daisy wheel so the light clutch friction to the minute shaft is easily overcome while setting.
Initially there was an issue with the daisy wheel binding occasionally so I filed as required to eliminate these binds when they occurred; there were 4 stoppages over 2 days. When turned by hand the daisy wheel never bound up so this seemed the simplest way to find and eliminate the binding. I had visions of months of intermittent problems but it has now run for over 2 months without stopping unexpectedly so it seems to be settled (I have stopped it for short periods to work on it during this time, of course).
The pins that hold the daisy stem vertical while allowing it to move up and down were set slightly too far apart. This allows the escape release to wiggle the hour hand more than necessary so I Loctited 1/16" brass tubing over these pins to take up the excess space - not obvious that it is a patch and it reduces the wiggle considerably.
A maintaining work is needed to ensure the clock operates normally while being wound, where winding removes the driving force temporarily. The escapement used here, like Harrison's Grasshopper, will malfunction if drive is removed while winding. This maintaining work can provide one impulse so winding can proceed without regard to the impulse time; it will provide a second impulse but this isn't sufficient so the count wheel will stall about 45 seconds later.
Initially, my maintaining work wasn't reliable, stopping the clock after an hour or two. I added a ball bearing but it didn't help so I changed the spring from 0.040 to 0.032 music wire. (The ink drawing of the spring is because I installed it backwards... more than once.) 
This worked longer -- the sprag clutch scheme activated and released as expected but stopped within a day. The underlying problem was the pulley slowly migrated on the shaft until the sprag wheel contacted the sprag holder, adding friction; a spacer between the drive pulley and pin wheel resolved this. Unlike most clocks where the maintaining work activates only when winding, this clock also activates it once a minute (each time the pendulum is impulsed) so it really gets a workout.
Dr. Woodward used some of John Harrison's concepts while designing this clock but used a different, more complex, maintaining work. My design is closer to Harrison's maintaining work, consisting of this spring plus a sprag clutch (in place of Harrison's ratchet and pawl) that prevents the drive pulley from turning backwards as the clock is "wound". This required separating the drive pulley from the escape wheel on the shaft to drive the main shaft through the spring shown. The large diameter section is the bearing surface for the single sprag which is held by a short steel pillar with a flat on the side (see pictures below); the sprag is the small cylinder which rolls between the two. The flat angles toward the large diameter section so the small cylinder sits between them; when the arbor rotates clockwise it lifts the sprag upward slightly so it rotates in place with little drag. When the arbor attempts to rotate counter-clockwise the sprag wedges between the flat and the large diameter section, preventing reverse movement and allowing winding. (Both the sprag and the large diameter section were rolled on a fine file to add a little texture for improved traction.) A ratchet and pawl used in this type of application is often called a click; this sprag clutch is... a clickless click ;-) I added a retaining clip (not shown) over the sprag because it falls out when the clock is laid flat to work on it.
My design hides the maintaining spring, unlike Dr. Woodward's where all of his works, including the maintaining works, are visible. John Wilding used a manually activated maintaining work while I prefer automatic. My notion is that this maintaining work fits well with the elegant simplicity of Dr. Woodward's clock... this from an engineer who got a D in Art Appreciation :-)
The weights suggested by John Wilding are about right for my clock when adjusted for my pulley setup. I used 3 lines to the moving sheave to reduce the weight's travel to 1/3. All the pulleys have shielded ball bearings. The jockey weight is 1 pound so the force is 1/3 pound. 
The drive weight is 1.5 pounds (for force, subtract the jockey force) with a payout of about 19 inches per day or about 22.3 inch-pounds net per day, about normal for a regulator clock. The drive force is about 4.5 times the jockey force - the jockey weight might be decreased by experimenting but is acceptable as is and provides a margin for change over time.
I tried several diameters of monofilament fish line as well as a braided fish line but the light line slipped through the drive pulley unless the ratio of jockey weight to drive weight was increased dramatically. Heavy monofilament (60 pound test) didn't slip but held so well that it wasn't possible for the jockey weight to do its job. The pink nylon string is 62 thou diameter and far stronger than necessary. My preference is braided rather than twisted line because after running for a few days when tension is released it tends to twist itself up making it awkward to handle.
I found #18 braided builders line by Starrett at Sears. It was available in bright yellow or garish pink; these colors are deliberately bright for visibility at job sites but the yellow goes reasonably well with brass, see picture at right. This line is flat rather than round but grips well in the drive pulley. The picture shows my classy test weight (would go well with the pink string :-), a peanut butter jar containing 4.6 pounds of steel. Both the drive and jockey weights have 3 sheaves; fall is about 6.5 inches per day. A drive weight has since been turned from the same steel round which supplied the pendulum.
The pulley holders on the weights are 1/2" steel round; the pulleys are smaller than the more traditional ones used by John Wilding, closer to those used by Dr. Woodward. I didn't think through implementation of my design very well: the slot to hold the pulley is 0.200 wide and 1.44 long. The only slitting saw I have with appropriate diameter is 3/64" thick so it took 4 passes at 10 minutes per to make each slot.
I expected to use monofilament for the drive line so didn't cut the pulleys very deep. The pulleys work fine with monofilament or the round (pink) string but the flat yellow string ran off the pulleys so I made a mandrel to hold them and recut them deeper.
The blue tape above the pendulum holds a reflector for use with the Clock Watcher's optical sensor.
I used the Clock Watcher's setup mode (displays pendulum period in microseconds) to quickly tweak the rate within a couple minutes per day using shims under the pendulum. The original Clock Watcher software was written for my earlier (unsuccessful) clock which had a 1 second pendulum. I took advantage of that one second pendulum to simplify the software and this came back to bite me -- I had to change it to handle this clock's 1.25 second pendulum. Fortunately, it was easy to revise the program so pendulum period is a parameter - the Watcher's display is affected in that it updates on each pendulum swing so every fourth swing it updates by two seconds.
I synchronize the Watcher's display with the radio controlled clock, then the gearless clock's hands are set to the minute and the seconds are synchronized by lifting the count wheel backstop to stop the count wheel without affecting the pendulum. When everything is sync'ed I watch either the Watcher or radio clock's display and listen for the gearless escape to trigger on the minute, this sound makes it easy to get everything set properly.
I've run this clock with the Watcher for a some time now and it seems reasonably stable, the rate rambles up to +/- 2 seconds per day. Not clear yet what the daily average will be since I'm still working on the clock. The pendulum isn't in final form (I'm still fiddling with temperature compensation) so this may be affecting stability. There is a 1+ second per day cyclical rate change superimposed where this variation occurs hourly and seems to be from a slight bind in the daisy mechanism; removing the daisy eliminates most of this cyclical variation. Here's an example Watcher chart; the big jump at 11pm is where I added an adjustment weight to the pendulum the day before.
In a way, the Clock Watcher provides more information than I wanted to know. The pendulum seems to be disturbed considerably by the large, once per minute, impulse: the impulse cycle is about 70ms shorter and the immediately following cycle is about 12ms longer than the average. Following cycles alternate being longer and shorter than the average much of the time between impulses. It acts as if there is a sub-harmonic superimposed on the pendulum's natural frequency, not an obvious operating mode. This is gleaned from observing the Watcher's display of raw data so perhaps it isn't an accurate conclusion and could be a problem in the Watcher software, although it didn't happen with my earlier clock which impulsed on each cycle.
Note: my Clock Watcher is cobbled together from parts and a micro-computer board that are not readily available so it isn't easily possible to make a copy. Brian Mumford produces a commercial clock watcher. This might be a software alternative, although I haven't tried it.
Having read Dr. Woodward's "My Own Right Time", I tried to proceed along similar lines in construction, using John Wilding's book for basic information and measurements. This lead to several minor differences between my clock and John Wilding's interpretation of Dr. Woodward's clock. John Wilding has written a number of books on building traditional clocks using traditional techniques so Dr. Woodward's clock was a departure from his previous experience, as it would be for most. You can see this in the way he mounted stops for the counterbalances on sturdy arms. So my version differs from John Wilding's in minor details like this as well as the quite different maintaining work.
I used Lucite for the count wheel per Dr. Woodward's original clock. This made it easy to cut the wheel plus it allows adding a seconds indication very simply: numbers on the side of the count wheel. I used a spin indexer with a fly cutter in the mill to make this wheel; since the spin indexer only indexes by degrees, my count wheel has teeth at 7 and 8 degrees alternately - ratchet wheels are less fussy than gears in this respect.
An observation on the count wheel is that the teeth are deeper than necessary. The modest ticks from the count and backstop pawls dropping could be quieter (not that they're noisy) if the teeth were cut to about half depth. In addition, this would make it easier to form the deep tooth which triggers the impulse.
I press or friction fit most things in this clock. In particular, I knurled the arbors lightly as needed with a fine straight knurl so the 1/8" reamed collets fit but their position on the arbor could be adjusted; this was helpful in getting everything to line up nicely. Using Locktite would have made position adjustments far more difficult. Adjustments were necessary to accommodate my maintaining work - I didn't draw it first, just winged it as I went along.
I used the mill's DRO to drill the holes in the pin wheel; fairly fast, very accurate and no doubt about the result. Watching it work, I'm glad the wheel is accurately made because clearance with the escape in operation is not large.
0.031 music wire was used for the pins in the wheel and for wires on the counterbalanced parts. 0.031 stainless safety wire was used for other wires where ease of forming was helpful. In particular, the backstop pawl, and the stops for the escape and the deflector. These stops were mounted by drilling parallel #60=0.040 holes through the mounting pillars, then making a "staple" from safety wire with one side long to be formed as needed for the stop. These staples were pressed into the parallel holes where the slight mis-alignment inherent holds them firmly in place; the wire is bent as required for the stop. Note that the mounting pillars were first put in place and marked so the staple holes are approximately in line with the expected stop position; these stops are visible in this picture but are obvious only if you're looking for them. Adjusting these stops is done with finger pressure since this wire is fairly soft.
Pillars were made from recycled line printer shafts; I didn't use brass or cut the tapers specified --this makes it easy to install and remove them using a drill chuck as a grip. Depending how it looks when completed I may eventually make tapered brass pillars.
The pillar on the jockey weight side of the drive wheel is 0.2" longer and the pillar on the drive weight side is 0.1" shorter than called for in John Wilding's plans. This was done to reduce friction between the lines at the crossing point; not clear whether it is useful but it seemed like a good idea at the time.
The mill's DRO also simplified making the daisy wheel. I programmed the DRO for 22 bolt holes and set the diameter for the bottom of the V between petals, then drilled the even numbered 11 bolt holes with a #56=0.046 bit. I then set the diameter for the center of the petals and spotted the odd numbered 11 holes. These spots were used when scribing the petals prior to nibbling and then filing the daisy to shape. The pin wheel center and 4 holes were also spotted and drilled with the DRO's bolt circle function and their accurate location was helpful in fitting the cam and then final fitting of the daisy.
I built a practice case for the gearless clock from inexpensive poplar to see if I could do it... my first cabinet making project. It took longer to build the case than the clock because I am low on that learning curve plus I had to make jigs and learn how to make joints with a table saw and router. I have some walnut for the actual case, once I recover from building this case.
My case design has a door to allow setting the time easily. Winding could be done by lifting the weight but I've found it easier to grip the string and use that to lift the weight - better control so the string doesn't get lifted off a pulley. The case bottom is split to allow the pendulum and strings through plus it allows easy removal of the case. There are two pins in the top of the case that engage the plywood back, so the case simply lifts off. The sides of the case have windows to allow viewing since I expect the works will be of interest to visitors. The clock is mounted on an outside wall (not the best situation) so I added spacers to hold it 1/8" from the wall to reduce heat transfer.
This page was last modified
18 August 2010 01:18 MST
. If you have a comment on this site or its contents,
click here scroll down and click again.
