Silencing the Bill Smith Gearless Gravity Arm Clock

You will find details of my activity building the Bill Smith Hip Toggle Gearless Gravity Arm clock elsewhere on my site.   This was the first clock I ever made and it taught me a lot about techniques all of which were well documented by Bill in his write up and in his many other books and videos.

Let me state now that the clock design as intended by Bill works and works well.  It has one distinct disadvantage that every minute or so it gives a very loud ‘clunk’ as the pendulum amplitude diminishes, the hip toggle triggers and the solenoid resets.    Running it in the workshop was fine as I became immune to the noise but sadly the clock will never progress into the house given my wife’s sensitivity to noise.

This has been a frustration to me as the clock looks splendid and its motion work action is a fascination to behold.  It deserves to be on display in a more public arena than the workshop.

All of which lead to some head scratching and a compromise re-design.   If I accept that the clock is an electro-mechanical device then my conscience allows me to consider other electro-mechanical solutions that are significantly less noisy.   This is the fundamental premise to my re-design.

There are many clock designs that use magnetism attraction and repulsion as the driving force and my thoughts turned to this as a potential solution.

I 3D printed a magnet holder to fit on the pendulum rod.   This holds two magnets.   There is a large one facing the direction of swing and a small one perpendicular to the swing towards the backboard.  I mounted a Hall Effect Sensor (HES) on a prototype board onto the back board at the mid swing position.   The gist of my idea was to have the pendulum swinging back and forth across the HES with the HES being triggered by the small magnet.   I would count the number of swings detected by the HES and after a defined number of swings I would energize a solenoid to repel the large magnet.

There was a little bit of electronics involved.   I had a 4060 binary counter counting the swings and used the divide by 32 output to trigger a 555 in monostable mode.   This would create a delay period from the mid point to the end of swing before the solenoid was energized.   A second 555 would then define how long the solenoid repelling pulse would last.   I also added LED indicators to all key timing points so I can easy diagnose what was going on.   I also allowed selection of the 16,32 and 64 divisions from the 4060 until I established the optimum choice.

The pendulum period is 4800 beats per minute so one swing lasts for 750ms.  The first 555 must therefore provide a delay of 375ms before energizing the solenoid.   The second 555 delay would be a ‘suck it and see’ period to be determined.

The concept was lashed up and worked OK …. except that the pendulum amplitude just grew and grew until the large pendulum magnet attached itself to the solenoid core …. not a good idea .   What was needed was a maximum amplitude detector to act as feedback to inhibit the solenoid pulse action.   

A second HES was mounted at a position that represented the maximum swing position and the output from this, when triggered, would feed back to the 4060 RESET pin to stop the count until the amplitude diminished sufficiently.   This worked and the result was very repeatable.   There was one proviso that the pendulum must be started by triggering the over swing HES and releasing.   Without this the 4060 could be one count out of step and would energize the solenoid as the pendulum was swinging back from its furthest point.  This would cause a repelling and slowing of the swing.

Overview image of the new pendulum sustaining mechanism on the Bill Smith Gearless Gravity Arm clock
Overview image of the new pendulum sustaining mechanism. Top left is the timing board and lower right is the Hall Effect Sensor board. The solenoid is the original Bill Smith design. The white block on the pendulum is the magnets holder. You can see the two Hall Effect Sensors hanging below the blue prototyping board.
close up view of the sensor board
A close up view of the Hall Effect Sensor board where the two sensors can be more clearly seen along with diagnostic LEDs and interface converters to the timing board.  The pendulum magnet holder can just be seen coming into shot.
Oscilloscope display of the new timing sequence for the Bill Smith Gearless clock
Oscilloscope display showing the yellow regular negative going pulse from the centre swing HES detector, the blue delay pulse to allow the pendulum to travel to its extreme and commence its swing back and the purple pulse triggered from the back edge of the delay pulse which defines the solenoid ‘On’ time.

The prototype has proved the concept but I now need to engineer a clean solution.   First choice is perhaps an Arduino Mini but it could also be PIC based.   

More to follow.

Similar or related subjects : –

A Spanner in the Works – or simply a Stick

From previous posts you will be aware of my involvement maintaining the local church clock.   Over the past months my colleague and I have been nibbling away at various little problemettes with the movement and things are now looking quite good.   For the past two weeks it has run sweetly and maintained +/-1 second over that period.

Then last night it stopped.

This morning we wandered round to see what the problem might be.  The first thing we do on arrival is look at the front dial to see at what time it had stopped.   This time it had stopped at around 10.35 last night. We climbed the tower and inspected the movement.   

There did not seem anything obviously wrong so we decided to swing the pendulum and get it working again. We had arrived at just before 10am and our inspection took us over the hour and the front the dial was showing 10.35.  Because we were now ‘within the hour’ it was acceptable to wind the hands back to the correct time which was now just after 10am.   

I pulled out the motion work locking pin and began to move the hands (which were now independent of the movement) in a backwards direction to set the time. Except the hands would not move backwards.   There was resistance.  Something bad had happened to the motion work.   

We checked the mechanism to both the front and rear dial but there was nothing obviously wrong but the hands refused to go backwards under light pressure and I did not want to force anything at this stage.

We went outside again and this time checked the front and now also the rear dial and this is what we saw : –

church clock with stick jamming the mechanism

Our feathered friends had built a nest on the belfry window ledge and a stick had fallen from the nest and jammed itself in the dial.   The odds of this happening must be pretty thin.

A careful waggle of the hands back and forth broke the stick free and we then reset the time and hopefully all will now be well.

The interesting observation was that the stick was only brushing the hand in the forward direction but in reverse it was pushing against it.    The forward resistance was still sufficient to reflect back through the motion work into the main mechanism to stop the escapement and therefore the clock.

Another bit of knowledge gained.

Similar or related subjects : –

Droning On and another Toy

Another drift to the dark side I am afraid.

I bought a drone.

Yes I know ….. Lots of reasons why I justified it to myself and I went through the prior post decision making process and got a ‘yes’ as a result.

Seriously I have been considering one for sometime.   In France the house sits looking out over a valley and we often see deer and foxes (but not yet a wild boar) and I have always hankered for being able to get up close to them.  France aside my work on turret clocks often needs a close view of the clock face without the hassle of ladders and scaffolding.   So two good reasons (in my book anyway).

I had been watching the market trying to decide when to jump.  The two big players for the semi professional market are DJI and Parrot.   Both these run out expensive.   Then in January I got a mailshot from Banggood about a new device to be launched by Fimi called the X8 SE.   This seemed to be only available from Banggood and was on back order status.    I missed out on the first delivery but finally my toy arrived last week.   

It is amazing.  And at a fantastic price.

Not ever owning a drone before I cannot compare with anything but it is so easy to drive and has so many automated flying routines.   Battery life is around 30 minutes and range is stated as 4km and it can skim along at 18m/s speed (yes that is 65km/hr ….).    It does 4k video and 12M stills.

So that is my entertainment sorted for summer, should summer finally arrive.

fimi x8 SE drone and controller
Fimi Drone and associated controller.

It is worth watching Dustin Dunhill on YouTube if you want a serious review of the device.  He does tech reviews and there are 3 or 4 Fimi videos on his platform.

Similar or related subjects : –

    Automated Wheel Cutting GCode for the Tormach with the Sherline CNC Rotary Table

    I have been putting this off as I thought it would be hassle and in fact it was very simple.

    Here is the code which is the first time I have ever used a sub-routine.

    The top section is my standard set up routine for the Tormach.

    The middle section has some out of the way locations to try the idea so don’t get too fixated by these.   The important bit is the M98 call for the sub routine, the sub routine name (1001) and the number of repeats (the L5 for five repeats).

    The last section is the sub routine indicated by the O1001.   The M64 command is specific to the Tormach USB Expansion board and it makes relay P0 in the box close its contact which in turn activates the Sherline CNC Rotary table to move one step.   The cutter than moves across to cut the tooth and then returns whereupon the relay drops out (M65), waits and then closes once more to increment the table.   Once five repeats have happened the M99 closes the sub routine and the program jumps back to the Z10 line in the middle section before stopping the spindle, homing and ending.

    The joy of this method is that it is a simple edit of one line (the M98  instruction) to change the number of tooth cutting increments. I like it a lot.

    Some fine tuning is still needed on the back and forth distances needed to clear the cutter through the wheel blank.

    The normal test of the cut depth routine will still be needed before this could be run but once this is done it should be a sit and watch job.   Hopefully.

    I have yet to run a wheel in anger so I will let you know how it goes. 

    A compiled write up of all the related blog entries on this subject can be found on the downloads page.

    Similar or related subjects : –

    Cycloidal Cutters and Finding the Cutting Centre

    It has been a thoughtful morning on the Tormach wheel cutting setup. 

    In order to cut clock wheels the first step is that I need to be able to set the cycloidal cutter centre line accurately on the centre line of the blank brass diameter.  See the picture and description below.

    Exaggerated mock up of the cutting setup showing a brass blank mounted on a super glue arbor and a cutter mounted in the Tormach Slitting Saw arbor. The centre line of the cycloidal cutter teeth sits on the centre line of the brass blank and cuts on the rear edge as seen above. After each cut the CNC rotary table increments the blank by one tooth ready for the next cut. (In practice the super glue arbor would need to be much larger in diameter in order to be more in keeping with the diameter of the brass blank and so ensure maximum support while the cutting was done).

    From previous posts you will know I have got the chuck securely and centrally mounted on the CNC rotary table and this assembly is in turn rigidly fixed on the tooling table. The position of the centre line of the chuck is now fixed relative to the tooling plate on the bed.  The chuck and rotary table mounting bracket is sufficiently Woody over engineered to hopefully be repeatable.   Likewise the distance from the spindle to the chuck can be repeatably zeroed using the Haimer and its associated tool table entry (#90).

    Expanding this a little, if I put my favourite piece of 11mm diameter silver steel in the chuck and bring the Haimer down to contact it, rock the Haimer back and forth in Y to get the steel diameter peak, I can get a Z zero reading to the top of the steel.   By creating a new entry in the Tormach tool table (#91) which is the Haimer length plus 5.5mm (the radius of the silver steel) I can use this virtual length stored as a new tool #91 to allow me to set the Haimer on the silver steel while actually giving me Z0 on the centre line of the chuck. So far so good.

    As you might have read from an earlier post, the idea of using the Tormach Slitting Saw arbor to hold my cycloidal cutters would in theory create a repeatable tool length to the centre line of the cycloidal cutter teeth.   Having this as a tool table set up in the Tormach would simplify setting the cutter centre to the centre line of the chuck and therefore the  centre line of the wheel blank being cut. This is where the thinking drifted somewhat.   

    I created a new tool table entry (#77) that was the length of the saw arbor to the shoulder that the cycloidal cutter fastens against.   I thought I could then follow the same routine as detailed above and add to this length the half thickness of the cutter and create a new tool table entry to match.  This would once again create a length which would give the centre line of the cycloidal cutter.

    That was fine until I measured my tray of cycloidal cutters to see what the thickness of the cutters were …… sadly consistent they are not.   There seems to be no standard by manufacturer or diameter.   I have cutters with thicknesses from 3mm through to 7mm.   I could create a new tool table entry for each thickness but this is a recipe for a mistake when selecting the correct tool table entry for the cutter being used.

    The simple solution I think is to use slitting saw arbor tool table length (#77) as the initial setting length to Z0 and then do a G0 Z-x.xx where x.xx is the half thickness of the cutter being used.   Once Z has dropped to this reading the Z axis can be re-zeroed to run the wheel in question with the cutter in question now sitting on its centre line on the centre line of the chuck.

    Simple diagram showing the concept of using the Tormach tool table facility to allow easy setting of the centre line of a wheel blank and cycloidal cutter centre lines

    I hope that all makes sense …. I could of course just eyeball it and not try to be so fussy but when you have the tools to make things easier you might as well use them.  I also need to look after my precious piece of 11mm diameter silver steel.

    Similar or related subjects : –