A Novel Free Pendulum Clock

I've been working on this clock for the last few weeks, and finally it is working. Please note that this is a test-bed for some ideas I have had churning in the back of the brain for many years, and is not (in this form at least!) intended for hanging on a wall somewhere, although at some point I will probably do the necessary tarting up to make it (or a Mk 2) domestically acceptable! There's a rather terse description in the YouTube entry, but I am writing it up in gory detail as we speak.

https://youtu.be/Sx-rwWu_gkM

Edited By Neil Wyatt on 12/03/2021 11:10:21

How many free swings between impulse do you get/allow Tony?

regards Martin

Hi Martin

The position of the left-most sensor determines the amplitude of swing - when the pendulum no longer activates that sensor, the microcontroller generates an impulse. There is no minimum or maximum number of free swings - it seems to have settled to around 2.25 minutes between impulses, but that can potentially change as conditions change.

Regards,

Tony

2.25 minute sounds pretty good. Sounds like a fairly high Q pendulum which is waht you need if you want to attain a good degree of detatchment. Should perform very well.

regards Martin

Yes, I am pleased with the way it is performing so far. The bob weighs around 4.5 kilos and the only things that erode the swing are suspension spring hysteresis losses and air resistance, so you would expect the Q to be high.

That's very similar to mine, but I have only 2 sensors, one at mid travel, one at extreme. I aim to impulse it every swing unless the extreme sensor has been triggered in which case it misses one out. I think this keeps the amplitude even more constant. If I knew how to post videos I would!

I've been adjusting it after swapping to carbon fibre pendulum rod, now better than 1 minute per week, but it's a slow job, timing it over a week, then adjust and so on

What are you using as the display? Re-purposed quartz clock?

The third sensor allows me to control when the impulse starts and finishes, based on the position of the pendulum. It was a design objective to make the pendulum as free as possible, and therefore maximise the number of swings between impulses, so I'm deliberately not aiming to impulse it on each swing, which is rather different from your approach by the sound of it. The rod is Invar. It has only been running since 10pm last night so I haven't sorted the rating yet - it has lost 20 seconds in 16 hours, but that is without any attempt to adjust the rate.

The motion work/dial is a Quartz clock movement with the electronics ripped out, so just the Lavet stepper and the gearing remains. The "ticks" are taken from the activations of the middle sensor.

Nice. Does the micro play a role in the stepping of the movement? On my two clocks it inhibits the step once every N swings so making the clock lose 1s (the pendulums deliberately run slightly fast). By adjusting N in software I can rate the clock to +/- 0.5 s. A lot easier than adjusting the bob. One of them has an additional tweak in that it skips a second every N seconds except every NxM seconds which gives a finer adjustment. The other clock (a Synchronome derivative) doesn't have that yet (it will do) so needs a slight tweak via a small weight on top of the bob.

Now you can join the precision clock measurement club - welcome!

Interesting approach! Yes, the stepping of the clock movement is entirely under the control of the micro, so it would be possible to give it a similar electronic rating mechanism. The I/O capability of the micro is a bit limited though, so short of embedding the values of N and M in the software, adjusting the rating would be difficult. Having said that, it will take live software updates via USB (or Bluetooth for that matter) so definitely a possibility.

Yes, mine are defined in the SW. On one though I can adjust the N value through a serial link - next clock will use Bluetooth I hope!

The BBC Micro:bit supports BT - you can programme it via a phone or a tablet with their app, which is pretty neat.

I've recently got an Arduino Nano BLE that I think can do that. The Arduino needs a hardware kluge to allow me to connect a serial port without resetting the processor which is a pain.

I'm dead impressed with the Micro:bit so far - mind you, what I am asking it to do isn't very taxing, but the ease of programming and setup is excellent.

Difference in approach, I reckoned that doing a little every swing was the way to go, whereas you give it a (relatively!) almighty clout every 150 swings. I very much doubt it makes any difference in even the short term, certainly not in the long term.

Edited By duncan webster on 11/03/2021 18:34:28

Nice to see this working Tony, which sensors did you use in the end?

I've embedded the video in your post.

Look out for the next MEW

Neil

A small impulse every swing or a larger one every so often....a keen philosophical debate! The argument for the latter is that the variation of impulse as a proportion of the whole is less so the variation in pendulum amplitude is lower. There are very good clocks using either approach.

My synchronome based one impulses every minute (30 whole cycles); my later magnetically sensed and impulsed clock has amplitude control and impulses every 3rd or 4th swing on average (I think...). Clock B's pendulum is never "free", the Shortt clocks were in effect. You pays yer money and takes yer choice...

My take on this for what it's worth, and all I'm doing here is thinking aloud, is that all escapements interfere with the natural period of a pendulum and that interference should be kept to the smallest possible. A lot of focus has been placed on the amplitude of the swing but for small oscillations circular error is small and provided that the variation is periodic and constant can be seen as constant over a long period but that assumes that amplitude is unaffected but pressure, humidity and temperature.

If we accept the above for a moment, the escapement error is that portion of the swing during which the pendulum is driven and for constant force impulses comes down to ratio of the swing under power to the rest of the swing. For this reason impulsing every swing is worse than impulsing after a number of swings. For impulsing by swing counting (fixed swings) you would expect a variation in amplitude due to atmospheric conditions with the corresponding vaiation in period due to the (small) circular error. Impulsing on (amplitude) demand would regulate this. The question is then does the variation in impulsing cause a bigger disturbance to to period than the variation in circular error. This can only be assessed on individual clocks as each impulsing system will perform differently.

As an aside I would like to see some figures for the angular precision of the optical triggering set ups on some of the clock being investigated at the moment. The devices generally have a long rise time so a degree of jitter would be expected. The Trinity clock instrimentation uses a shim on the pendulum passing through an IR detector which then gives 4 edges make/break either side of the centre position of the pendulum. Calculations are then done to fix the timing of the centre position.

I realise I may be in danger of 'teaching grandmothers to suck eggs' here as I'm sure there are more expirienced time fiends here but maybe this will give at lease some start to a conversation, hope so, it is quite interesting.

regards Martin

Thanks Neil. I went with Hall effect devices in the end - Allegro ones. I also tried an interesting 2 channel device but it is surface mount and only about 1mm square, so not the easiest thing to handle!

I agree - often Vs infrequent impulsing is an interesting debate. I haven't studied the subject enough to have a fixed opinion, but I suspect the fact that the Shortt clocks used infrequent impulsing may be telling us something. The reason I chose to use infrequent impulsing in this clock was that there was a certain appeal to me to build a clock that emulated the swing arc control of the Hipp mechanism, but where the sensing and impulsing is non mechanical, and where there is no mechanical linkage between the pendulum and the motion work. After all, if it was accuracy I was after I would have stuck with a £5 quartz clock.movement.