Tag: CNC

This morning I made some temporary clock hands so that I can tell how long the clock runs before stopping and whether the clock is running fast or slow.   I then followed my standard practice for debugging the clock.   Run the clock until it stops, looks at the teeth engaged when it stops, sand any imperfections and mark the teeth with a piece of tape, re-start the clock and repeat the process. 

If the clock continues to run past the teeth that I have marked I remove the tape marker.   Hopefully, over a few hours the number of teeth marked keeps reducing until none are marked and the clock keeps running.   Unfortunately, by lunchtime the clock kept stopping, there were no obvious issues with the teeth, and I realized that there was another problem.

I started tweaking some of the other parameters that I knew could be the cause of poor running.   I changed the weight on the remontoire arm, no joy.  Then, I changed the fore/aft location of the intersection of the pallet stem and the torsional spring form.   As I reduced the effective length of torsional spring fork the clock began running better.   I judge this by the “strength” of the tick-tock sound. 

Pallet stem position at the start of the day

I kept adjusting this length until I had a strong “tick-tock” and the clock was running well.  I then went back to the process I use correct problem teeth.   By mid-afternoon I had removed all the tape marking problem teeth and at the time of this post the clock has run for 4 hours and is still going. 

During this process, I have been changing the weights on the torsional pendulum to adjust the clock speed.  The clock is now keeping great time for this stage in its development.

If the clock continues to run, I have two tasks for tomorrow – Cut the actual clock hands and construct the clock speed adjustment mechanism.

With the clock running intermittently, the task becomes finding the teeth with small imperfections and sanding them smooth.   I started by sanding each tooth on the great wheel with 400 grid sand paper to smooth off the cutting tool marks. 

I then assembled each pair of meshing wheels in turn and rotated them by hand to see if I could feel any problem teeth.  I found several teeth that needed minor sanding.   I then assembled the going train of the clock and started it running.  When it stops I inspect the meshing teeth and sand any that looked suspect.   My experience is that I will need to continue this process for a few days, each day the clock will run longer before stopping, but once the last imperfection if smoothed the clock will run continuously with no problems.

Tomorrow – More tooth sanding, testing, tooth sanding, testing…

Started today by assembling the clock without the pallet so that it would run freely and rotated the great wheel by hand.   Everything ran freely.   I added a temporary weight to the remontoire arm but did not connect the battery and set the arm at the top of its arc and let go.  The clock turned under the power of the weight.  See video below or at my website (https://cedarclocks.com/the-clock-runs-for-the-first-time) if the video does not transfer to Facebook.

I then connected the battery so that the winder motor would run when the remontoire arm reached the bottom of its arc.

The only parts remaining to be constructed before I could test the clock, including the torsional pendulum and escapement, were the pallet stem and the torsional pendulum weight.   The pallet stem links the pallet to the torsional spring fork.   The fore / aft location of the pallet stem (the location of the intersection with the torsional spring fork) is a critical dimension as it sets the rotational amplitude of the pallet, effectively the point in the cycle that the pallet gives an impulse to the torsional pendulum.  I have found, through trail and error, that if this dimension is not correct the clock will run very poorly or not run at all.   I set this dimension based on my experience on previous clocks.  

I always make a temporary torsional pendulum weight holder so that I can easily change the weights to determine how much weight is required to get the clock to keep time.   Nothing left but to see if the clock runs.   After a couple of minor adjustments, it ran…  not particularly well… but it ran for a few minutes.   

I had decided not to sand the great wheel teeth as the surface finish looks reasonable and I was impatient to see if the clock run.   The clock appears to be stopping on teeth that that I can see are rough. 

Tomorrow’s tasks: Sand the teeth of great wheel to see if I can get the clock running smoothly, make some temporary hands and adjust the torsional pendulum weights so that it keeps time.

I was in a hurry to publish yesterday’s post and did not include photos of how I ensure the clock is hung vertically and the lower clock supports so I will start today’s post with them.

To make progress I had no choice but to cut the remaining 20 pinion rods and polish all 67 of them.  This is my least favorite tasks.

On previous clocks the pinion rods were a push fit in their holes.   On this clock the brass rod diameter is very slightly smaller than the nominal 0.125” and I suspect the holes are minutely large so that the pinion rods are a sliding fit.  Ultimately, this will not be a problem as I will glue them in place, but I want to run the clock to check it runs well before I remove the pinion rods to finish the clock.   I decided to dip one end of each rod in the Watco lacquer that I will use to finish the clock before inserting them in to place.  This should be good enough to hold the pinion rod for initial testing.

With any luck, tomorrow I will assemble the clock and see if all the wheels turn freely.  If they do, I will make the final few parts required to “set the clock to beat” and see if it will run.

After a day of lathe work, the clock (as is) is hanging on the wall.

It is amazing how much better the lathe is running after yesterday’s adjustments and I turned all the brass parts that attach the clock frame to the wall with no problems.

When I designed these parts I was very conscience of an experience from about 40 years ago when I built some shelves for my first house (built in ~1815) in Derby, England.   The shelves were to be located in the corner of my bedroom and I worked hard to ensure that they were square, all the corners were right angles.  I was very happy with the results until I tried to hang it on the wall only to find that there was a large ugly gap between the shelves and the wall.  The shelves were square but the corner of the room was not quite a right angle and I had not thought to check before I started.   On this clock the torsional pendulum spring runs very close to the clock frame and the clock will not run if the spring touches the frame. If the wall the clock is hung on is vertically this will not be a problem.   Remembering my house in Derby, there is no guarantee that the wall will be vertical and I need to include provision to adjust the clock hanger to accommodate an imperfect wall.  

Below are a series of photos showing the attachment of the clock to the wall.

To compensated for a non-vertical wall, the distance of the top of the clock from the wall can be adjusted by changing how far the collar is threaded on to the brass rod.

Brass insert at top of the clock frame.  A threaded hole in the side of the insert accommodates the locking set screw that seats in to the groove on the collar.

I can’t put off cutting and polishing the remaining pinion rods much longer, so that is my plan for tomorrow…  but I could cut the clock hands.

The day started well when I started pulling out the electronic components required to control the electric motor that winds the clock “remontoire” and found that I had an assembled set of components that I had built as a spare for my last clock.   I decided to use this and wire it to the new motor that I bought for this clock.

I discussed the remontoire and choice of electric motor in an earlier post (https://cedarclocks.com/the-new-clock/ ) that included a video of my last clock in which you can see the remontoire operate at about the 60 second point in the video.  The remontoire motor is controlled by a tilt sensor, when the angle of the remontoire arm rotated below a given angle a small steel ball in the tilt sensor rolls down the sensor tube on to a set of contacts reducing the resistance across the sensor to almost zero.   When this happens the circuit is designed to activate the motor that winds the remontoire wheel up the winding wheel until the angle of the arm causes the steel ball to roll back down the sensor tube resulting in an open circuit that stops the motor. 

In theory, one could wire the tilt sensor between the motor and battery and the system should work.   However, the motor draws a larger current that the tilt sensor is designed to withstand for long term use so I decided to use a MOSFET switch.   The MOSFET switch is activated by the tilt sensor, however, only a very small current passes through the sensor and the larger current required by the motor passed through the switch.

I developed this circuit on my last clock, Electra, and, as I only have a basic background in electronics, I was very happy when it worked first time.  However, after a small number of operations it stopped working and it became clear that the MOSFET switch had failed.  I tried again with a second switch with the same result, it worked a few timed and then the MOSFET failed.   I assumed I had a bad batch of switches and bought a few new switches from a different source.  The first of these switches failed in the same way as the previous ones.  I consulted Google and soon found that this a well know issue when using a MOSFET switch to control an inductive load like a motor and I needed to add a “flyback” diode to the circuit.   The reason for this is explained in the Wikipedia article https://en.wikipedia.org/wiki/Flyback_diode.   Once I added a flyback diode the circuit worked well.  How did we do stuff like this before Google?

Below are a series of photos of the electronic set up and the remontoire arm.  The tilt sensor is the gold colored component on the right of the circuit board.  Finally, I took a video of the remontoire motor operating when I tilt it by hand.

Having successfully assembled the remontoire arm I decided to start working on the brass parts that attach the clock to the wall.  The first part requires cutting 24 turns per inch thread on the end of a 3/8” diameter brass rod.  I decided to cut this thread on the lathe rather than use a die.  I have very little experience thread cutting on my lathe so I knew it was going to be a learning experience.  My first attempt failed when the cutting tool jumped and ruined the thread.  I tried again making very small cuts but it became clear that there was an issue with the lathe as the cutting tool kept jumping leaving a very poor cut.  I investigated and found that the lathe saddle was loose on the lathe bed.  These Chinese made mini-lathes are excellent value for money, but to adjust anything requires completely disassembling the lathe.   Two hours later I had striped down the lathe, adjusted the saddle slide, cleaned everything and re-assembled it.   I cut a couple of simple part and the lathe was cutting very smoothly leaving a much better surface finish that I have achieved recently so I suspect that the saddle has been coming loose for some time.  Tomorrow, I will re-try cutting the thread.

The 3/4” diameter brass rod that I need for another part of the clock hanger arrived today so I plan to cut that tomorrow with my newly adjusted lathe.

A light day of clock making today as it was my turn to shop for the week’s food.   I never thought that I would be selecting vegetable at the supermarket looking like a bank robber in my homemade mask.

After returning and putting everything away I was not in the mood for sanding the 312 wooden gear teeth that make up the clock.  I went down the “to-do” list and decided to tackle the most complicated metal component in the clock – the “torsion pendulum spring fork” that transmit the motion of the torsional pendulum spring to the escapement pallet.

  Below is a photo of the “torsion pendulum spring fork” on my previous Electra clock. 

Note that you can only see the brass forks as the remainder, the body, is hidden under the spherical wooden cover.  

The “torsion pendulum spring fork” clamps to the torsional spring using a set screw so that its position on the spring can be adjusted.  

On a previous clock I make a small steel jig to hold the body of the part and allow accurate drilling of the three holes, one for the clamping set screw, and two for the forks that are made from 1/8” diameter brass tube.

My current thinking is to turn my attention to the electronics of the Remontoire motor controller tomorrow, but that could change.  There are still 312 wooden teeth to sand and 20 more pinion rods to cut.

The escapement mechanism is the most dimensionally sensitive part of the clock.   The pendulum controls the rocking motion of the escapement pallet on its arbor.  In turn, the pallet allows the escapement wheel to rotate one tooth at a time.   To work correctly the following dimensions have to be correct:

• Distance between the pallet and escape wheel arbors.
• The shape of the pallets.
• The radius and width of the scape wheel teeth.

Nothing is perfectly accurate and I deliberately cut the escapement wheel teeth to be longer than required and then trim them to the length needed to for the escapement mechanism to run.   The photo below shows how the initial, over length, escapement teeth results in the pallet arms not letting the escapement teeth pass.

I use a dial indicator jig and the Zenbot CNC router to check and trim the escape wheel teeth.   The first step is to use the dial indicator to find the shortest tooth and then adjust the router cutter to just touch this tooth.  

I then turn the escape wheel by hand through the cutter to trim them to the same length.   I then checked to see if the escapement still binds. It did not and I returned to the CNC router to trim 0.005” off the tooth, re-checked if the escapement worked and repeated multiple times until the escape wheel just passed through the pallet arms.

Some of the teeth still rub on the pallet arms as they pass through, so I suspect that I will need to trim more off, but I did not want to trim too much off as this results in a clock that has a loud tick-tock.

Next, I made three 1/2” diameter brass collars.  These will hold the hour hand, minute hand and secure the end of the great arbor.

Finally, I cut the arbors and the brass bushings for the wheels give 12:1 gear ratio between the hour and minute hands.  

I received an email from Hobby Metals telling me that the 3/4” diameter brass rod that I ordered has shipped.  Should arrive on Thursday. 😊.  I am not sure what to tackle tomorrow.  Although the CNC router gives a good surface finish, I still need to sand every gear tooth.  This is a mind-numbing task ideal for doing when listening to a Cincinnati Reds baseball game.  With no baseball, I guess I will do it while listening to Corona virus news.  I also have to cut the clock hands.  I will decide tomorrow.

Started the day by rough sanding the joints between the frame and cross members but decided to wait to until I have assembled the clock to determine how to blend these intersections.  

As I am getting impatient to see what the complete clock looks like, I decided to drill the centers of the wheels and cut the arbors (shafts).    Drilling the centers of the wheels is the most critical process in making the clock.   If the centers are too close together the wheels will be tight and the clock will not run.  If the centers are too far apart the gear teeth with rub across each other, rather than rolling over each other, as the wheels turn resulting too much friction.

I use the Zenbot CNC Router to drill to achieve the accuracy required to drill the holes for the wheel centers.

With the wheel centers drilled, I had to cut the arbors from drill steel, start to cut the brass tubes that fit into the wheel and turn on the arbors, and begin to assemble the Going Train of the clock for the first time.   I deliberately cut all the parts to be longer than needed so that I can trim them to the correct length as I align all the wheels and pinions on the arbors.   Over the next week I will make extensive use of my Dremel tool with a cut-off or grinding wheel to cut and adjust the brass tubes that position the wheels.

Very satisfying to see the clock take shape for the first time.

I am not sure what to tackle tomorrow.  I am tempted to trim the escape wheel teeth to length so that the escapement mechanism works and I get closer to running the clock for the first time, but there are still a number of small brass parts to turn on the lathe and the last 20 pinion rods to cut.   I am still waiting for the 3/4” brass rod that I need to mount the clock on the wall to arrive.   Although there was nothing on their website about being closed, I am a little concerned that Hobby Metals may not be operating.

Today, I drilled holes through the frame and in to the cross members and glued in walnut dowels to strengthen the joints.

As it is a gray day in Cincinnati, I decided take on a task that I enjoy while waiting for the frame dowels to dry.   The clock requires a number of collars, knurled adjustment screws and locking nuts all made from 3/8” diameter brass rod.   These relatively simple brass parts are turned on the lathe, drilled and tapped as needed.   I enjoy using the lathe to make these parts and get a lot of satisfaction from producing these small parts.

Things on the list for tomorrow include sanding the intersections between the frame and cross members, making more small brass parts and perhaps drilling the centers for the wheel arbors (shafts) in the frame.

This morning the glue on the great wheel was dry and I sanded the excess glue from the joints.   All the joints look good and I used the “radius tool” to check roundness.  Everything looks good 😊.

On my previous clocks I have used a conventional router (rather than my CNC router) to put a 1/8” radius on the edges of the frame.  This gives the frame an “organic” look.   I need to decide whether to round the edges of the frame of this clock or leave a sharper edge giving a more industrial look.   I have been back and forth on this over the last few days, but today I must decide as the frame edges needs to be rounded before adding the cross members.   I decided not to round the edges on the frame but will probably sand off the sharp edges later to give them a very small radius.  However, the cross members looked too square and I decided to use the router to produce a 1/8” radius on the edges.

Once the glue has dried, I will drill holes for dowels to strengthen the joints between the frame and the cross members.

Final job for the day was to cut additional pinion rods.   47 cut, 20 to go.

On my smaller clocks I do not cut the great wheel teeth until the wheel is assembled and then cut the center (arbor) hole and the teeth on the rim of the wheel using the CNC router to ensure that they are as accurate as the Zenbot can achieve.   Because the diameter of the great wheel for this clock is 21.3” which is larger than the capability of my Zenbot (16” X 24”) so I developed a process where I cut the teeth on each of the 5 segments that make up the rim and then assemble and glue the parts over a full size plot of the wheel.   Below are photos of this process.

While the glue dries on the great wheel, I planned to machine the parts that hold the clock frame to the wall from 3/4” diameter brass rod.  However, when I measured my brass rod it was 5/8” diameter not 3/4” diameter that I thought. ☹.   I have ordered 3/4” diameter rod from Hobby Metals and hope it arrives soon.

As I could not machine the frame hangars, I decided to cut pinion rods from 1/8” diameter brass rod.   Cutting and polishing the 67 pinion rods that the clock needs is a rather repetitive and tedious task.   The best process that I have found is to cut them to length using the scroll saw using a simple jig to set the length.

The plan for tomorrow is to add the cross members to the clock frame.

I hope you are finding my blog interests.  I would be interested in your thoughts, feedback, or questions.

Late yesterday I glued the two halves of the top of the frame together and the components that make remontoire arm and left them to dry overnight.

Below are photos of the remontoire arm once the parts are glued together.   The motor, battery, tilt sensor and electronic circuit board will be located in the cutouts.

A challenge on all my clocks is how to hold the top of the torsional pendulum spring.   As the speed of the clock is determined by the length of this spring, it is essential to rigidly hold both ends of the spring is essential for good time keeping.  I also wanted to be able to remove and replace the spring if necessary.   The design I came up with is to create a thin slot in a brass rod that holds the top of the spring.   A screw in the side of the rod clamps the spring in place.   The photo below shows the upper spring holder on my previous clock.

In this clock the upper spring holder is made from 3/8” diameter brass rod located in the “v-crotch” of the top of the frame.   Because accurately drilling a hole to accept the spring holder was going to be a challenge, I designed a jig to hold the top of the frame while I drilled the hole.  

Below is a series of photos showing the drill process. 

Now that the hole for the torsional spring had been drilled in the top of the frame, I could glue the frame legs components together.    I place a mirror on my work bench to ensure that I have a flat surface on which to construct critical part of the clock.   (A mirror must be flat so that it does not distort the reflected image.)  A full-size plot of the component is taped to the mirror so that the parts can be aligned for gluing and a sheet of parchment paper laid on top to stop excess glue sticking to the plot.  I use clamps, tape and weights to hold the part in place while the glue dries. 

While waiting for the frame arms to dry, I decided to swap from woodwork to metal work and create the upper spring holder discussed above.  The challenge is to create a slot for the spring that measures 0.125” long and 0.007” wide.   I start by drilling a 0.125” hole in the center of the 3/8” diameter brass rod using a mini-lathe.   I then grind 1/8” brass rod to form two semi-circular pieces that will be soldered in to the 3/8” rod to form the slot.  I hold these inserts in place to solder using a thin strip of aluminum from a soda can.  Because lead solder does not bond to aluminum, the aluminum strip can be removed after the solder has set to create the slot.   The final step is to cut the brass rod to the correct length and clean up the ends on the lathe.

Torsional Spring Holder after Soldering and Cleaning

The plan for tomorrow is to assemble the great wheel.