Weekend Letter

When I started this project, I really had no idea how long it would take.  Without a master plan or not having made one before just like this, I just took it as it came.  That and also having to make a living in between, opportunities to work on it extended the project beyond the 2 year window I had in mind when I started this in February of 2015.

One cannot rush a project like this as you just have to deal with things as they come along.  Kind of like life when you think about it.  So, today it is running and has been for a while.  I think I have gotten most of the bugs out of it since June of last year, which was my last post.

Changes I made since last posting:

• Replaced 15-30-45 minute bell with a larger one which sounded better.
• Added a latch to the hour strike to give it a more paced and consistent ring.
• Modified the ball diverter that replenishes balls used for the hour strike.  I did this three times before coming up with a design the worked every time.
• Added dampeners to the weights used for the hour selector and 15 minute reset to cushion the drop at the end of the cycle.
• Replaced some vertical links with chain for aesthetic reasons.

Because these things were not part of the main construction, I did not photograph them.

In October 2017, I started building a model 2 cylinder steam engine.  As I worked on the little engine, the clock was still on the bench so I could keep an eye on it for anomalies.  When one would show up, I would turn my attention to the clock and work to find a remedy for issue.  With each incident things got better and it ran for a longer interval without needing attention.  That may sound funny, but remember these mechanisms were built from theory and not for a set of proven plans.  My ideas did not always work as intended and had to be modified or in some cases re-engineered.

Like most ball clock, you tell the time by counting the balls.  Bottom row is hours.  Middle row is 10s of minutes.  Top row is minutes.  So on the photo, the time would be 3:06.

The clock will strike the smaller bell once for quarter hour, twice for half hour and three times for three quarter hour.

The large bell on the top is struck the number of times for the hour.

Both bells are from old National Cash Registers that I salvaged many years ago in the later 1960s.

I have attached a set of recent photos to this posting.  4 without the dome and 2 with.

Hope you like them. Jon  (longjon582@gmail.com)

One of the things I had to consider with the design of the Hour Selector mechanism was what to do with the balls that accumulated one every hour.  By the time 12 o’clock comes around the track holding these balls was filled to capacity.

Something had to be done to release (dump) these balls before 1 0’clock came around.

The trick was to find something in the works that only happened at 12 o’clock and no other time.

Then be able to trip another cycle to allow the 12 balls in the rack to exit.

Also and very important, the balls could not exit at the same time the clock was striking 12 since they used some of the same exit path back to the screw lift.

This could have caused a ball jam from balls hitting each other running down the same track.

When the hour selector is stopped by a ball in the rack, it engages another cam that opens a gate to allow that selected number of balls to move to the hour hammer trip area.  Because this happens immediately at the start of the selection at 12 o’clock, it is the longest rotation the cam makes.  At the end of the cycle to provide power to return the cam to its home position a chain is attached to it and a weight.  Now because this is the longest travel the chain will make, I had a spot that I could monitor.

By making an arm with a slot that straddled the chain and putting an adjustable collar on the chain, I was able to have an action occur only at 12 o’clock.  The collar contacts the arm and moves it up rotating a gear and turning a shaft that goes to the other side of the clock.

On the other end of that shaft is an arm that releases a clutch pawl to engage and clutch.

The clutch engages a cam, that when it turns, moves an arm attached to the gate holding back the hour balls.

The cam that moves the release arm for the balls is 285 degrees around from the engagement point of the clutch.  Because the shaft that the clutch is on turns are 2.5 rpm, the minimum amount of time before the gate is opened is 20 seconds and a maximum of 44 seconds.

It takes 18 seconds for all 12 balls to strike the hour. This allows a clear track for the 12 balls to exit when dumped.  And yes that was fun to figure out.  I wonder sometime what I got myself into.

For the next 12 minutes balls on the supply lift for striking will be diverted to fill up the lower track of the hour selector since twelve were used for the strike.

I soon discovered that I needed a way to keep pressure on the cam to keep the clutch pawl engaged.  I solved this by making a small spring loaded arm with a bearing to ride on the cam.  I cut an index slot on the cam to create a home position.  This made the engaging of the clutch dependable.

Now since this was the last major mechanism, it is time to do some base work.

March was pretty much taken up with fixing some things that I had been putting off for a while.

I knew it would require a fair amount of disassembly to get to the areas that needed attention and then the arduous task of cleaning and putting it all back together.

I wasn’t satisfied with a couple of the parts that I had made early on in the project and wanted to replace them.

It is like anything else you do where after getting familiar with processes and developing a technique, it then becomes easier to see how they can be better and gaining the skill to do it.

My silver soldering has gotten better after two years of working on the clock along with realizing that a different approach to an issue will produce a better product.

I needed to drill and tap some holes for the next stage and couldn’t do it with parts in the way.

This also gave me the opportunity to correct some issues that had reared their ugly head every now and then.

I’m not an engineer or purport to be one.  So many of the things I do on this clock are from theory and then verifying it works by making the pieces.

Now and then I discover a situation that I had not planned on.

Things like a ball stopping on the track say maybe once every other Tuesday.  Trying to figure out why that happens can be perplexing.  Is it just a dirty ball, dirty track or some harmonic causing the hesitation enough to stop the ball.  Usually it is dirt of some sort.  The dome will be important to keep things clean.  Once finished I will replace all the balls with new and only handle them with white gloves to keep the salt and acid for skin from etching their surface.

Also a part that worked fine in the beginning, was possibly bent from other work and is now causing an issue.

Anyway, a number of things were waiting to be addressed and it was a good opportunity to address them.

The photos show how far I had to disassemble the clock to get to the section that needed the holes.

Taking it apart and re-assembly, which that ability was planned from the beginning, then making the changes, took me the better part of March.

The upside here is the photos will give you a chance to see a little deeper into the clock and maybe if you have been following the blog, answer a question or two.

Regardless, the project is now better for the time spent and changes made.

Now it is time to get those 12 o’clock balls released so they get out of the way of the 1 o’clock ball arrival.

For the last couple of months I have been operating the Hour Ball Selector by hand to test it.

The time has come to power the main shaft of the selector.  These are the things to be considered for the drive:

• The rotation of the shaft has to be triggered by an hour ball when it drops into the upper rack of the selector mechanism.
• The clutch for the shaft has to make one complete revolution and then disengage.
• It clutch has to be in sync with the main screw lift which turns at 1 RPM so the release of the hour ball can coincide with the alignment of the clutch.
• The clutch shaft has to be aligned with the main selector shaft which is not perpendicular with the base.
• The clutch has to maintain a home position when disengaged.
• Power for this must come from the main drive, yet not slow it down.

To begin I used the same thickness of brass plates and spacers that matched the main drive.  

The gears were brass from Boston Gear to match the others.  The biggest issue was matching the size, pitch and ratios of the gears so that it could be mounted in line with the selector shaft and still mesh with the main gears.

Ball bearing we installed on all shafts to reduce wear and friction.

A clutch had to be made that would engage the upper shaft going to the selector with the lower drive.

An index was ground on the upper clutch and a spring driven roller arm was made to keep it in home position.

The alignment was a bit tricky so I used a small universal joint to make the connection.  There still was slight binding in the shaft so a slip coupling was made to allow a small shift and flexibility in the alignment of the two shafts.

All in all there are 10 ball bearings in the mechanism.

When a ball comes down from the lift on the hour, it trips a lever that allows the clutch to engage.

The clutch then makes one complete revolution until the trip level pushes the clutch pawl out of the lower drive.  

It stays in that position because of the homing arm.

One of the biggest challenges was to get the clutch pawl just right where it would stay engaged with the right pressure, but still release at the end of the cycle.

At first there was a tendency for the pawl to cam out of the lower drive and disengage while part of the way through the cycle.

I am tempted to make these parts out of steel to cut down on wear.

Right now, I need to keep moving ahead to the next challenge.

There should be time for some cleanup later.

Well at least I hope so.

Not everything will go as planned in life and none truer right now than this project.  Back from Christmas vacation and back at it.

My original design for the hour striker was pretty simple.  The weight of a ball would cock the hammer and a cam would release it at an interval to provide a time space between strikes. See Photo.

The problem I had to work out was a balance between cocking the hammer and enough weight to make a good strike when the hammer was released.  I tried adding weight to the hammer and even tried a bigger bell.  All the tings I tried worked to some degree, but not enough to make the statement that I wanted.

I had to “Bite the Bullet” as they say and accept the fact that a big change had to be made to the mechanism.  I felt that I could keep the lower mechanism that cocked the hammer in the original design and use it to set up a sequence run by the upper shaft.

Using a spring to power the hammer seemed like a better idea as it could be adjusted.  I could also make a better stop adjustment for the hammer and mount it on bearings.

So the idea was to release the hammer in a way that a cam on the upper shaft could cock it.  At the end of the cocking cycle it would be released by a pawl, which was also adjustable, to strike the bell.  The cam would provide a positive time interval between strikes regardless of when the mechanism was tripped. 

A latch was made to keep the mechanism tripped until a strike occurred. Since there is not fixed relationship between the balls being released into the trip mechanism below and the setup, the latch verified a strike and allowed the process to continue.  There is a relationship between the latch and accepting another ball in the mechanism below.

The first cocking cam had 4 lobes which was a bit slow at 6.0 seconds.  The second version had 5 lobes allowing a strike every 4.8 seconds.  It was important to find the right interval and still not be too fast.

There was another thing to consider also.  When 12 balls were released into the striker trip mechanism, there was only room for 7 balls that were clear of the gate that released them.  The way the hour Ball Selector Cam was made, there was only 27 seconds before the gate would close.    That meant that the striker mechanism had to consume 5 balls at 12 o’clock within 24 seconds.  The smaller hours weren’t an issue as there was more time to clear the gate.

Anyway, the larger bell and spring loaded hammer provides a clear strike.

The Ball Selector needed to return to its home position at the end of the cycle.

I chose to use weights rather than springs for this because springs will increase tension when expanded and weights are a constant force.

The main shaft of the selector mechanism makes a full revolution during the process.  The Ball Selector must return to its home position at the end of the cycle.  This occurs at different time depending how many balls are selected.  Longer at 1 o’clock (350 degrees) and shorter at 12 o’clock (185 degrees).  One of the weights is used for this.

In the photo you can see both weights used for this.  I chose round chain for its strength and the way it fit into the pulleys.  It also looked pretty good. 

When a proper number of balls are selected, a latch pawl engages with the Release Plate that looks a saw blade.  When this plate is engage and carried along with the rotation of the main shaft, it will open a gate to allow balls to roll into the hour striker mechanism.  The second weight is used to power the return of the Release Plate to the home position.

When balls are used on the lower rack of the Hour Selector they must be replaced.  The source of the balls for this comes from the track used to replenish the quarter hour strike.  A feeler was installed in the track going to the lower rack.  When the track filled with balls to that point, the feeler was depressed and latched a diverted allowing the balls following every minute to end up in the return area.

There are two independent ball tracks in the clock.  One is used to supply balls for the time racks and the other to supply balls for the striker mechanisms.

The next part to make was a ball stop for the upper track.  It had to align the balls on the upper track with the balls on the lower track for the selector to work right.  The balls will accumulate one every hour until 12 are in the rack.  I will then need to find a way to dump them before 1 o’clock.  That is for another day.

Note that the photos are of the mechanism mounted on the jig.  There just wasn’t enough room the work on it inside the clock.

Next up, reworking the hour strike.  It is just too weak.  I feel that way sometimes too.

After some testing it became apparent that an Index Plate was necessary to hold the selector arm in place once it has been extended.

It was important that the Index Plate was in line with the position of the balls in the upper & lower track.  This is so the vertical piece on the selector will pass into the appropriate slot on the Index Plate.

Tabs were soldered drilled and tapped (2-56) onto the lower track to provide mounting for the Index Plate.

I didn’t want to solder the Index Plate directly to the lower track incase I needed to change something on it.

If there is one thing I have learned during this project it is that you have to keep your options open.  It’s kind of like life, never box yourself in.

The index was necessary to establish the location of the lower gate to release the balls into the hour striker section.

Once mounted and tested, the assembly was reinstalled into the clock to check for clearances.

Things are getting tighter in that area and it is important to stay on top of it.