Repeatable mechanical “trigger” to start a free pendulum?

[S KParticipant @sk20060]

What sort of mechanical “trigger” mechanism would be able to start a pendulum cleanly and with high release position precision? Actuation and resetting can be manual.

Thanks!

[Michael GilliganParticipant @michaelgilligan61133]

I believe the classic experimental approach was to draw the pendulum into position with a cotton thread, and burn through that to release.

Variations on that theme would seem appropriate.

MichaelG.

[John HaineParticipant @johnhaine32865]

A short lever that can rotate in the pendulum plane, pivoted at one end with a pin sticking out at the other to act as a banking. To launch, quickly turn the lever so that the pin releases the pendulum.  Doug Drumheller used something like this operated with a small servo to release his test pendulums to do run-down tests in an environmental chamber. Described in his book “On the Pendulum”.

[Robert Atkinson 2Participant @robertatkinson2]

Ideally any mechanical release should be balanced so it does not impart any out of plane forces. This would indicate a scissor or pincer type release

[Peter Cook 6Participant @petercook6]

How precise do you want it? John H’s  suggestion is OK, but you would need to be careful that the release pin did not impart any velocity at all to the pendulum at the point of release.

To that end I would suggest a slow release using a lever whose final (release) curve had a radius that matched the distance from the release lever pivot to the point of release of the pendulum, and with a gathering curve that picked up the pendulum from rest and moved it to the release point. Something like this.

That shape rotated slowly anticlockwise would pick up the pendulum, take it to the release point and then hold it there while the release curve slid under the contact point and imparted no further movement to the pendulum until release.

If you manually reset everything each time so that the pendulum sat on the release curve in the release position, you could do away with the pick up curve and use a simple lever with its end radiused to the length of the release lever.

[John HaineParticipant @johnhaine32865]

Actually I’ve just re-read Drumheller and the servo (actually a small geared stepper motor) is used to lift the pendulum against an electromagnet which holds it while the lifting arm is retracted.  Then the magnet is switched off to release. Doug posts regularly on the HSN forum if you want to contact him –

https://groups.io/g/Horological-Science-Newsletter/topics

[S KParticipant @sk20060]

The thread idea is romantic, but a “no.”

The lever solution sounds simple enough. The lever arm and stop needs to be stiff and precisely perpendicular to the plane of the pendulum to limit out-of-plane forces, and the arm needs to move out of the way faster than the acceleration of gravity. The stop, being say a 1/4″ round shaft, should present only a small aerodynamic influence on the rod during run-down. It’s probably not difficult to automate, too.

The cam approach is a neat idea, especially since no separate stop is required. The sliding force involved bothers me a little, and I’d probably want an equivalent version that can work higher up the rod. I’ll think a little more about it.

I had been contemplating a lever to raise the rod towards an electromagnet, which captures it and then releases it electronically. But this means that the electromagnet would stay in very close proximity to the swinging rod early in the run-down (moving it out of the way defeats its advantage). I’d worry about aerodynamic influence plus residual magnetism accumulating and affecting the swing as it nears the electromagnet.

I guess the lever approach seems most practical. Are there any photos or drawings of an automated version out there?

[BazyleParticipant @bazyle]

I think anything magnetic is hugely variable. Rate of fall off of the magnetic field, residual magnetism, induced currents and fields in the pole piece and armature, all being temperature variable too.
Similarly metal on metal, at least for steels, in any other hold mechanism could become magnetic.
Anything implying a sliding contact brings in variable friction affects on timing and thrusts however minute from the mechanism bearings.
I see something like glass/diamond at the contact point. A mechanism that releases by itself physically moving in the direction the pendulum is going to swing, and moves faster than the gravity start for the pendulum so it gets out of the way. But it must be small enough that it provides no wind effects and must move initially in the same arc as the pendulum so there can be no side forces.

[John HaineParticipant @johnhaine32865]

Well, I’m not sure what you are trying to do, but Drumheller used the magnetic approach successfully for all his run down measurements.  Once the current is switched off, if the remanence of both the “stator” and “armature” are low there should be minimal or zero magnetic interaction.  However I think it would be easy to devise a scheme where the servo moved the pendulum to a precise position then let it go. His servo was I think one of those low cost geared steppers which can’t move quickly.

Doug’s system operated in a pressure chamber so he wanted to be able to do successive runs at different temperatures and pressures without demounting anything.

[S KParticipant @sk20060]

Just using a servo was my first thought, and in fact I have one rigged up now. But common servos, operate via relatively crude pulse width modulation, are noisy and not super repeatable. It’s more than fine if a very exact release point is not important, though.

I suppose I should join that other forum at some point, though I wouldn’t join a club that would have a bum like me as a member. 😉

 

[Michael GilliganParticipant @michaelgilligan61133]

Permit me a half-baked suggestion

Machine a hemispherical dimple in the bottom of the pendulum, and a matching hemisphere on the end of of a rod which is pulled away rapidly,  by whatever mechanism you choose … Assuming a small angle of swing this should give a almost interference-free release.

MichaelG.

[SillyOldDufferModerator @sillyoldduffer]

I adopted a different approach with my experimental clock, partly for practical reasons, and partly after measuring what happened when my pendulum was released.  I found a pendulum cannot be released precisely, and it’s better to let it swing until it stabilises at a known amplitude before measuring it.

The ideal free pendulum is isolated from everything that might interfere with it. Best sealed in a vacuum chamber, which disqualifies any release mechanism that has to be fiddled with.  For example though Michael’s thread is good because it eliminates wobbly operators, it fails here because there’s no way of installing the thread, or lighting a candle that won’t burn in a vacuum. Actually, anything requiring the operator to set a release mechanism inside the chamber is off the table.

My clock’s pendulum is designed to swing inside a vacuum chamber, creating a problem: how to start the bob swinging in the first place.  Hands off.

Another reason. Experiment suggests the act of starting a pendulum by any means is intrusive. Starting them isn’t “precise”, at least when measured in microseconds or faster.  Various causes: mechanical mechanisms are likely to be rough;  the bob will fly in an ellipse unless released exactly at a right angle to the bob and suspension; turbulence; and the rod may ‘twang’ as it takes the strain. Magnetic release should be smoother and less intrusive than mechanical, but how long release takes depends on the core and bob material, coil inductance, and the flyback diode.  Have measure how long it takes – anything from microseconds to several tens of milliseconds.

I concluded the first swing is always abnormal, not precise! When precision is required pendula require time to settle. Therefore I abandoned the idea that my clock would keep time from the first swing.  Instead, the pendulum is started with an excessive amplitude (about 6°), and allowed to swing free to 3.5°-ish, by which time, it is more precise.  Period is not trusted until the pendulum has settled.

Tried two starting strategies, both using a side-mounted electromagnet:

1. the bob is powerfully pulsed at resonant frequency, until amplitude rises above 5°.  Then not impulsed until the amplitude falls to about 3.5°, after which impulsed with just enough power to keep it at that level.   (3.5° to reduce circular error)
2. the bob is pulled hard on to the electromagnet by applying full power for a few seconds and then released.  Drop amplitude from the magnet’s pole is about 7°.  Then the bob allowed to swing free until amplitude falls to 3.5°, when normal ‘keep it going’ impulses are applied.

3.5° was found by experiment, not calculation.  My pendulum settles adequately at 3.5° from either type of start.  Other pendula will be different.

In practice, pulling the bob hard on the electromagnet starts reliably, but the drop is more violent and the pendulum takes longer to settle.  Resonant starting is less reliable. If the bob is already swinging the resonant pulses can be applied in anti-phase, with negative results.

An observation:  my bob is never completely completely still!  I guess it picks up enough energy from the environment to move ever so slightly.  Maybe vibration due to traffic outside, me clumping around, plus the house moving as it heats and cools.  That pendula make  good seismographs shows up distinctly when measured at high resolution.   Rigidity matters too.  Whilst debugging my clock is running on a dining table, which severely reducing the pendulum’s precision. Q is down by a third compared with running it on a concrete floor…

Starting and allowing the pendulum to settle to a known amplitude before measuring might suit SK?

Dave

[SillyOldDuffer]

On SillyOldDuffer Said:

…

 

An observation:  my bob is never completely completely still!

You have observed the “Bateman Effect”!  First described after Doug Bateman had observed the same on his high Q pendulum when the clock not running.  He also made measurements over quite a long period of time of the magnitude.  Subsequently Philip Woodward developed a theory of how this would affect the pendulum stability.  It would be great to have a linear sensor that could measure the motion to characterise it better.  The assumption is that it is primarily horizontal motions in the plane f the pendulum that cause an effect, and the principle effect is on the pendulum amplitude.

[S KParticipant @sk20060]

Yes, Dave, if you read my last post or two in the other thread, you will have an idea what I’m trying to do: calculate the “best” period from a run-down in the sense of the highest signal-to-noise ratio. It does not need a perfect launch, though less-disturbed launches will provide better data early on, which would be an advantage.

Being “free”, it could result in a higher S/N ratio than a continuously-impulsed one. But such a wacky technique will likely not be as accurate as a good standard approach.

 

[S KParticipant @sk20060]

It occurred to me that a simple modification of my current scheme can improve its position repeatability.

Currently, I have a good-quality standard servo operating as a cam to lift the pendulum by its rod near the bob. It’s fine, but servos are noisy (they constantly make tiny adjustments to their position) and not precisely repeatable.

If I added a stiff post as a stop, perpendicular to the plane of the pendulum, positioned right behind where the servo lifts the rod, and then made the lifting cam compliant to a degree, it could press and hold the pendulum rod against the stop without applying too much force. Then the release would be from the position against the stop, hence hopefully more uniform.

A release without any out-of-plane offset or twist, etc., is probably more important than a uniform starting position. At present, I’m only relying on good orthogonality and parallelism, and don’t have many better ideas for clean lifting and release. Maybe a ball bearing on the cam to reduce friction as it lifts the rod might help a little.

Thank you.

[Author]

Posts