Electrical pendulum clock modifications

Hello Steve and Douglas Reid, I read the above thread with great interest as I am in the middle of trying something similar with my Electro Magnetic pendulum clock. I have built the SP1 and am on my third Sp2 weight driven clock. I have the SP1 in my office at work and have assembled an SP2 for me at home and one for my son in law at his business and am finishing another SP2 up right now for my 2nd daughter. They all run with an error of less than 1 minute in a week. But I have never been able to get the EMPC to run accurately. No matter what combination of pawls I tried it would always eventually advance by two teeth per swing or not advance at all. I put it aside for a while as I needed to get some other things done. (Like the SP2 I promised my daughter almost a fear ago lol). In the last few weeks I've been messing with the EMPC a bit more. I am running it from two 18650 cells through a LM2596 buck converter. I was hoping a more stable voltage would lead to a more consistent pendulum swing. At the moment I'm not concerned with battery efficiency. I am just trying to get the clock to run consistently. I had the LM2596 laying around so that is what I'm using. I hadn't even thought of a boost converter and I am not sure of the pros and cons of either one. The results so far have been mixed. I will say it is much easier to tune the clock by adjusting the input voltage which has a direct effect on the pendulum swing amplitude. A 10mv change in voltage has a noticeable effect on the pendulum swing and corresponding pawl travel distance. It leaves me wondering how this clock could ever achieve reasonable accuracy running directly on AA batteries. The clock is now running on about 2.5Vdc. The active pawl is near the top of the pendulum. Previously I was running with less voltage and the active pawl was lower. While doing so I became aware of some mechanical issues with my clock. It was running and keeping good time but then stopped. Two of the gears had slid toward each other and came into contact which locked up the clock. I installed some tiny washers where needed to space the gears. I may have installed one to many washers on one of the gears as afterwards the clock would not run at the same voltage. It seems I added some friction. I didn't feel like taking it apart again so I just increased the voltage and raised the active pawl a few notches. That is why I say I have had mixed results. One day soon I'll remove the extra washer but for now I just wanted to prove it could run consistenly. Yesterday I designed a base to hold the 18650 cells and the LM2596. The clock with its original base will just sit on top of this base and be wired to it. If this all proves successful I will probably redesign a replacement for the original base to accommodate the regulator and batteries.

Douglas Reid, I am very interested in any ideas you may have concerning the electronics for this clock. I have mentioned it here before and you may already be aware of them but just in case here are a couple links showing how Dick Bipes has accomplished a similar idea using a micro-processor to regulate his electro-magnetic clock. I think his use of a cammed pendulum driving the active pawl is key. I think that arrangement alone might help this clock quite a bit.

Clock Link:

https://www.instructables.com/A-wood-gear-clock-with-a-unique-drive-mechanism/

Electronics Link:

https://cdn.instructables.com/ORIG/FGB/5WH6/I4IXCJU9/FGB5WH6I4IXCJU9.pdf

James

Hi Steve,

My ideas for an improved electromagnetic kicking circuit have been evolving.

The initial objectives are to provide more accurate control of amplitude and timing of the EM pulses, together with reduced power consumption compared to use of the Chinese modules.

With the benefit of being microcontroller-based, an elaborated version could also measure and correct timing by changing the amplitude of the pendulum swings

- in theory, the long-term accuracy could then be as good as that of the quartz crystal in the circuit.

Apart from design and construction of the electronics, this involves changes to the clock base such that the custom coil and PCB together with different batteries can fit within the base.

I am also thinking that a version of the clock made from a hybrid of wood and 3D printed components would be good.

If the base and supporting framework were to be made in wood (i.e., all parts seen in orangy-brown plastic) were to be made in wood, the clock would look much better!

It is completely reasonable and good to 3D print all the gears, but the base and vertical supports, being in plastic, detracts from the overall appearance.

Using the step file of the base which you kindly provided, I have produced a variant that could be easily routed out of wood, refer graphic.

If desired, there could be a slope down towards the front - but that would complicate construction.

Other variants are also possible such as a simple square/ rectangular or a hexagonal or octagonal base.

Arguably, the simpler the base, the better from the aesthetic perspective of the design.

A simple rectangle of polished wood may be the best?!

My thinking is that the top surface of the base should be flat such that it is not obvious that the pendulum is traversing over a kicking device!

I should expect that the EM signal is able to pass through a thin thickness of wood.

In any event, the top surface can best be flat whether or not it may be found that there needs to be a hole.

The clock framework is best also made in wood.

Only simple 2D curves are involved which can be easily routed by hand using a template.

There is now the challenge of how to attach the framework to the wooden base?!

Rather than having to cut rectangular holes (apparently also having the complication of sloping sides!) at an angle in the base, it would seem better to use simple angled circular holes.

I should appreciate if you could provide the step files of the support framework parts - I should then be able to modify the ends where the verticals attach to the base and also cut holes into the base itself.

Much better would be if you could make available the overall Fusion 360 model - the inter-relationships and accurate dimensioning between the parts could then be checked.

Furthermore, it will then be possible to visualise how the overall clock looks with the various alternative shapes of base.

If you prefer, we could communicate directly. What is your email address?

Best Regards,

Douglas

Subject: Coup Perdu

Hi Steve,

Congratulations on finishing your new design!

Your YouTube video discusses the pros and cons of the Coup Perdu and Graham deadbeat alternative mechanisms and I understand that either could be used for this clock (despite it being called "Coup Perdu") which involves usage of the small motor for rewinding. Maybe the alternative variant is best referred to as the "motorised Graham" or something ...?!

Do the plans for the Coup Perdu include for both alternatives?

Based on your comments and, in particular, that the running is quieter, I should favour construction using the Graham deadbeat mechanism.

I already bought plans for the EM pendulum clock and am wondering whether or not to (also?) construct that earlier design of clock.

Does the EM pendulum clock use the same Graham deadbeat mechanism as you are talking about in relation to the Coup Perdu clock design?

Or has a better variant of the Graham deadbeat mechanism been developed since your EM pendulum clock design?

There has been some mention of quirkiness regarding the EM pendulum clock and I noticed in a video that it sometimes seemed to miss a beat.

Does the quirkiness arise from some aspect inherent in the (earlier design of?) Graham deadbeat mechanism (or whatever the mechanism used in the EM clock design is properly called for)/ 3D printed gear construction

or is the issue relating to the electromagnetic kicking aspect?

I appreciate that for any 3D printed clock there is a need for careful adjustments and some fine tuning.

My question is really which should be considered as being the inherently most robust design.

Usage of the small motor for rewinding is an interesting idea.

However, I consider that the electromagnetic kicking approach used in the earlier design also has potential as a means to power such 3D printed clocks.

As I have posted in the EM clock thread, I have some ideas for an improved EM kicking circuit which has the potential to provide more accurate control of amplitude and timing of the EM pulses, together with reduced power consumption.

I shall add to that posting today with some new ideas.

Douglas

P.S. Probably an extra forum category is required for the Coup Perdu?

Summarising my thoughts following this analysis ...

It would be interesting to investigate whether the rewind time could be speeded-up to, say, just 2 seconds.

If so, the battery life could probably be extended by a factor of 3x.

[The current drawn by the motor may vary under different voltages and loading conditions - so this statement would need to be checked in practice!]

From the mechanical perspective, such a speed-up could be achieved either by using the version of the motor with gear ratio 380:1 (or another variant with ratio such as 286:1)

or by using the existing 1006:1 geared motor with a change in the clock's rewind gearing ratio to achieve the 3x rewind speed-up.

Using 4x AA could negate the need for a voltage converter as long as the rewind process happens for voltages between 6.4 and 4 volts.

The speed of the motor will likely change as the voltage changes but that is not important since the actual rewind time is unimportant.

However, the question arises as to whether the motor torque would remain sufficient to perform the rewind task as the voltage drops.

The torque becomes more of an issue as the gear ratio is reduced (with the objective of achieving the desired speed-up).

Use of a voltage converter could be beneficial if it enables the motor to properly perform across the range of battery voltage.

Also, whether the motor would be happy with a starting voltage of 6.4 volts?

I did some calculations to obtain an estimate of the battery life for the Coup Perdu clock.

Please check my figures (it is early in the morning and I have not properly woken up yet - so there could be a mistake?!)

From specification on eBay page for JL-12FN20-1006: 5v at 0.054 A

That may be incorrect - a similar table which appears to originate from the manufacturer shows 0.077 A, refer N20GearMotor.pdf

5v at 0.077 A --> 0.385W when running.

From video:

"Rewind operates every 3 or 4 mins."

Looking at the video it very roughly seems to take 6 secs for a rewind.

So rewinding takes 6 secs and occurs every 180 secs (assuming the worst case 3 mins figure).

Duty cycle: 6/ 180 = 0.033 (or 3.3%)

---

How many watt-hours from an AA battery?

https://sustainable-nano.com/2016/04/29/aa-batteries-mercedes/#:~:text=A%20typical%20AA%20battery%20contains,have%20to%20change%20the%20batteries.

A typical AA battery contains about 3.9 watt-hours, ...

https://sciencing.com/energizer-watthour-battery-specs-7425932.html

Data sheets for Energizer's technical information can be searched on their website.

For their AA battery, it looks to have a capacity of 2800 milliamps hours, or 4.2 watt-hours.

---

Hours * 0.385 * 0.033 = 3.9 (using that figure for the watt-hours)

Therefore,

hours = 307

If step-up conversion efficiency is 80%, the hours figure reduces to 0.8 * 307 = 245.6 hours

That corresponds to 10 days for a single AA battery.

For 3x AA, 30 days, which is not unreasonable for usage, although use of rechargeable batteries makes sense.

Using 3x or 4x AA could possibly be convenient if it means that a step-up converter is not required.

Starting voltage would be 3x 1.6 = 4.8v, towards end of battery life dropping to 1v per AA - would the motor still operate the rewind mechanism with only 3v? (Can you please check that?)

For 4x AA, starting voltage would be 4x 1.6 = 6.4v (hopefully, the "5v" specified motor would survive that?!), towards end of battery life dropping to 1v - would the motor still run with only 4v?

It would be good if you could test the motor with 6.4v, 4v and 3v!

(Under the load of doing the rewinds.)

The version of the motor on the eBay page has torque 102.4 N.m with gear ratio 1006:1

If the clock rewind could operate with a reduced torque (and assuming other variants in the list of motors could be sourced), the power consumption could be reduced.

For example, the -380 version of the motor has torque 75.6 N.m with gear ratio 380:1 whilst drawing 0.068 A.

So it operates 3x as fast, assuming the torque is sufficient, whilst consuming around the same power.

So a single AA battery could then last for 3 months?

Hi Steve,

Thanks for your prompt response.

Yes, a constant amplitude swing would improve the timing accuracy - so that is another benefit.

I shall consider in more detail the subject of using a boost converter.

The conversion efficiency of such are typically in the range of 80% to 90% across a wide range of input voltage.

Probably 90%+ of the battery life can be used whilst producing a constant voltage with associated constant current drive to the coil.

Since the current can be chosen to meet the actual requirement, the battery can be expected to last much longer.

How long does a battery typically last with the Chinese modules?

The current consumption could be further reduced by optimising the magnet/ coil system.

Obviously a powerful magnet moving close to the coil would require a lower current pulse and/ or of a shorter duration compared to a smaller magnet spaced further away.

I shall think about how many/ the type of battery that could best be used.

The cost of batteries, whether disposable or rechargeable type together with the battery holder and convenience to replace or recharge batteries should also be considered.

Could the base benefit from being weighted for stability?

Maybe a steel plate at the bottom?

I haven't studied the instructions in detail - maybe this subject (adding weights within the base) is discussed?

The new design looks very interesting!

Since it is a "much better design" and will be available shortly there seems little point in my constructing the existing model.

I should always be thinking that I should have built the better design!

I am interested in both table- and wall-mounted versions.

Is there any possibility of a special deal for me on the new design, given that I shall not be using the existing?!

I shall focus my current efforts on constructing the Silent desk clock, together with thinking about electronics.

Regarding the Coup Perdu Clock, please let me have details of the rewind motor and possible sources of supply.

It may be time-consuming to obtain this special part from China or wherever so it is best that I order it as soon as possible.

With knowledge of the motor specification, I can also reapply my ideas for efficient use of batteries for powering that motor!

[Voltage, current consumption, how frequently motor is switched on and the duration of each rewinding process?]

Regarding the Silent desk clock, I shall be thinking about the motor drive electronics ... I already have some ideas for a more elegant solution.

It would be useful to also have the step file for the base of that clock.

I should then be able to modify it to accept a custom PCB and also better understand how the motor (of particular size) will be attached to the base.

Douglas

Hi Douglas,

You make some very good points. The Chinese modules might be drawing excess energy when the battery is full to allow continuous operation near discharge. The useful life of the battery could be limited to a lower limit of 1.1V. Here is a random alkaline battery discharge curve. Assume approx 40% life between 1.6V and 1.3V, 50% life down to 1.1V, and the remaining 10% discarded.

The target use model of the pendulum drive circuit is just to swing a dummy pendulum back and forth. It is acceptable in this case to allow the pendulum amplitude to slowly degrade as the battery voltage declines over time.

My EM clock is a different use model where a constant amplitude swing is more desirable. A constant supply voltage seems like a good idea. Would a boost converter have a low enough power to be able to extend the battery life significantly? What about just using a lithium-ion battery? They have a very flat discharge curve. They run at 3V which works well in this clock. They have a long life and they are rechargeable.

Here is the file for the base if you want to experiment. Rename it to .step. I design in inches, so it should be scaled up 25.4X in size. Please share any results you may find. I am always eager to learn new ideas.

I have no plans to make this clock wall mounted. However, I am nearing a release of a much better design that will have a wall mount option. It has a much more reliable escapement. The motorized rewind seems very tolerant of the alkaline battery discharge curve. My first prototype has been running over a month using batteries that were too weak to run a TV remote. It should be released in the next few weeks. The early version can be seen here https://youtu.be/i7kuVZNyTLY

Steve