Measuring Clocks

[duncan webster 1Participant @duncanwebster1]

I’m not even attempting to match a Schorrt clock, getting as good as a Synchronome would be good. My Clock1 (3/4 sec pendulum and opto switches) hasn’t been set right since clocks changed in March, and is currently 1 minute fast. I’m not planning any improvements! Clock2 (1/2 second pendulum) relies on magnetic sensors but for reasons I don’t understand has low Q so I’m not expecting miracles.

[blowlamp]

On blowlamp Said:

I’ve lost track of the aim of the project.

Is it to make an accurate pendulum clock, or is it to find the best way to electronically impulse a pendulum clock?

 

Martin.

Both.

The aim is to make a pendulum clock more accurate than a Shortt-Synchronome.  Achieving that level of performance requires close attention to many factors, one of which is impulsing.  Electronic impulsing offers several advantages over mechanical escapements, hence the extended discussion.

There are many interactions in this project, and I too am having trouble tracking them.   So far this topic has led to me taking 11 pages of notes…

The actual topic is about measuring.  But its led us into what is being measured and why.

A web forum isn’t good for disciplined engineering development.  Topics tend to drift, often usefully, but drift can generate lots of noise.

At work, much more controlled.  We would write papers, and take comments formally, tracked with a database.  Then, after filtering, a cleaned up draft would presented at a moderated meeting.  More than one, sometimes, depending on the audience.  An iterative process supported by walk-throughs, inspections, steering committees and lots of lovely documentation!  Hard work, and I doubt it would be popular here!

Dave

 

 

 

[blowlampParticipant @blowlamp]

I think if I were mixing a pendulum clock with electronics I would largely stay with Shortt’s principles.

I’d use a free pendulum impulsed by a gravity arm, but use a quartz clock in replacement of the mechanical slave clock. This would be used to deploy and return the gravity arm at the correct moment in the pendulum’s swing. Furthermore, the drop of the arm after impulse would signal pendulum position and would be used to reset the quartz clock to ‘pendulum time’.

 

Martin.

[duncan webster 1Participant @duncanwebster1]

I don’t see the point of the quartz clock, using electronic sensors the master pendulum can release and reset its own gravity arm. Apart from the force exerted by photons bouncing off a vane a slotted opto doesn’t disturb the pendulum at all.

[blowlampParticipant @blowlamp]

As I envision it, the pendulum would be impulsed by control of the quartz clock say every 30 seconds to maybe a minute or two and would in addition provide the time display.

 

Martin

[duncan webster 1]

On duncan webster 1 Said:

I don’t see the point of the quartz clock,…

I’m happy with Martin’s idea.  The Shortt-Synchronome uses it’s slave clock to count 30 beats before releasing the gravity arm.  Mechanically counting beats disturbs the Slave pendulum slightly, doesn’t matter because the master pendulum is isolated from the slave – almost!

Quicker, easier and cheaper to make a quartz slave than a second pendulum clock, and they’re small too.   Needn’t be accurate, a ceramic resonator Arduino would do.

Replacing the gravity arm is key to my plan though.  Shortt’s implementation minimises contact with the pendulum, it does not eliminate it; indeed Woodward’s description of the Shortt gravity arm uses the word ‘jolt’.   It’s very very good, but a system that impulses the pendulum without any physical contact should improve on it.

The potential exists because mechanical systems are slow and clunky compared with electronic. That said the Shortt gravity arm’s errors are low enough to allow the clock to detect tides, so I’m chasing a tiny improvement.

I’ve only recently realised is that Shortt accuracy requires a gravity arm AND an ultra high-Q pendulum.   Sob, I have grave concerns about my pendulum, it’s sticky.  I think it’s twisted…

Not getting much done today – off-colour,

🙁

Dave

 

 

[duncan webster 1Participant @duncanwebster1]

I still don’t see the point of the quartz. Dropping and lifting the gravity arm needs to be at the right part of the master pendulum swing, so just fit a slotted opto at centre, a wide vane on the bottom of the pendulum and every 30th swing drop the arm when opto blocked, lift when it opens. Has to be done when pendulum going in the right direction of course. By all means use a motor out of a quartz clock, just not the crystal and associated circuitry.

[duncan webster 1]

On duncan webster 1 Said:

I still don’t see the point of the quartz. Dropping and lifting the gravity arm needs to be at the right part of the master pendulum swing, so just fit a slotted opto at centre, a wide vane on the bottom of the pendulum and every 30th swing drop the arm when opto blocked, lift when it opens. Has to be done when pendulum going in the right direction of course. By all means use a motor out of a quartz clock, just not the crystal and associated circuitry.

Ah, I see what you mean.

Dave

[duncan webster 1]

On duncan webster 1 Said:

I still don’t see the point of the quartz. Dropping and lifting the gravity arm needs to be at the right part of the master pendulum swing, so just fit a slotted opto at centre, a wide vane on the bottom of the pendulum and every 30th swing drop the arm when opto blocked, lift when it opens. Has to be done when pendulum going in the right direction of course. By all means use a motor out of a quartz clock, just not the crystal and associated circuitry.

Is this getting dangerously close to John Haine’s steppernome?

[John HaineParticipant @johnhaine32865]

Arduinome(TM) please!

The reason we know that the Shortt free pendulum could detect tides is that Pierre Boucheron discovered the 3 pendulums at NBS were still under vacuum though no longer running.  He persuaded NBS to let him set one of them going maintained by a timing circuit that happened to use the “local” time standard but not in a way that affected the pendulum’s period.  There was a quartz window in the tank through which he shone a laser to detect the motion, and of course a local source of precise UTC against which to measure the period!  That’s what you could emulate using a quartz clock.  The results he got were for the free pendulum alone, I don’t think we could say that the “Shortt Synchronome” could detect tides.

In principle the Synchronome could be insensitive to just where the gravity arm contacts the pallet except that the pallet shape in the actual clocks isn’t really correct.  Shortt himself worked out what the shape should be and that’s what I have tried to emulate in my clock, where the roller falls on to a dead roll just before the impulse ramp, the ramp is shaped according to Shortt to give a raised-cosine force profile, and the pallet face is another dead roll after the ramp.

I don’t think the motor from a Quartz clock would be any use – 2 pulses per rev and almost no torque.  I use a NEMA08 stepper.

[duncan webster 1Participant @duncanwebster1]

I meant use the motor from quartz to drive the hands

[John Haine]

On John Haine Said:

Arduinome(TM) please!

The reason we know that the Shortt free pendulum could detect tides is that Pierre Boucheron discovered the 3 pendulums at NBS were still under vacuum though no longer running.  …  The results he got were for the free pendulum alone, I don’t think we could say that the “Shortt Synchronome” could detect tides.

…

Woo hoo! Though I knew about Boucheron I missed the point that the Shortt-Synchonone wasn’t running as a clock when he tested it, only the pendulum.  Means the gravity escapement is still in the dock, accused of loading the pendulum. A magnetic escapement might be worthwhile because it gets closer to the ideal ‘free pendulum’ than a gravity arm.  Worth a try.

Unless it performs badly I will stick with the electromagnetic impulsing for another reason: much easier to make than a gravity arm,  especially one of Shortt’s advanced design.

The electromagnet also allows impulse power to be changed, either tuning in to find the optimum value, or adjusting amplitude on the fly.  Important to me because my pendulum swings in air whilst I debug the clock, but is intended to run in a vacuum.   The impulse power is different:

• more energy needed to swing a pendulum in air, because of friction, turbulence, and draughts.
• much less energy needed to swing in a vacuum.  How much less is unknown, because I’m not sure how low my ancient vacuum pump will go.  Managed about 500mB on a somewhat leaky test rig, so I anticipate somewhere between 400mB and 800mB inside my clock.  800mB is safely below the lowest natural air-pressure recorded;. 870mB in the eye of a super-typhoon.  Shortt’s pendulum swung in a hard vacuum, and I expect he could calculate the power needed.  In contast, I’ll have to pull the best vacuum I can, and then find the optimum impulse power needed experimentally.  Much easier to do that electronically than by tweaking a mechanical arm.

Vacuum depth makes a big difference to pendulum performance.  Being unable to pull a hard vacuum on my clock is the most likely reason it will fail to get close to Shortt performance.

I’m getting close to a vacuum trial.  Pump available, the sealing wax recommended by Robert has arrived, as has the valve. Waiting for the tyre thread tap to arrive next week.   Need to seal the pipe top, and reinforce the plastic fitting – have to turn a ⌀124mm Aluminium disc.

Thanks again,

Dave

 

[John HaineParticipant @johnhaine32865]

To be clear, an external timing circuit controlled the gravity arm rather than a Synchronome, so tidal detection was with the gravity impulsing. And it doesn’t have a hard vacuum, in fact I believe they adjusted the pressure for fine adjustment of rate through the buoyancy.

[John Haine]

On John Haine Said:

To be clear, an external timing circuit controlled the gravity arm rather than a Synchronome, so tidal detection was with the gravity impulsing. And it doesn’t have a hard vacuum, in fact I believe they adjusted the pressure for fine adjustment of rate through the buoyancy.

Ta, I got the hard vacuum ‘fact’ from Woodward.   Good to know I might not need it.  I’ve been swotting up on Mercury pumps!

Dave

[SillyOldDufferModerator @sillyoldduffer]

Back to measuring.  I am implementing a function to find the pendulum’s Q.  Works thus.

On command:

Stop impulsing, so the pendulum loses energy,
measure amplitude.
theta ∝amplitude
Q ≈ pi / ln(previousTheta/currentTheta)

Not tested yet.

Problem is testing Q by decaying the pendulum stops the clock!  Acceptable during testing, but I’d prefer something that didn’t!

In the Mk1 Clock I used an electrical Q calculation based on the frequency spread of the pendulum. (Q indicated by how far an oscillator deviates from centre frequency – a bell-curve. Sharp is high Q, broad is low Q)  Used the period logged to the PC.

John Haine said no!  Is there another way of deriving Q from the information normally logged to the PC for analysis.  The amplitude and period of every beat is available.

My head hurts!

Dave

 

 

 

 

 

[SillyOldDuffer]

On SillyOldDuffer Said:

Back to measuring.  I am implementing a function to find the pendulum’s Q.  Works thus.

On command:

Stop impulsing, so the pendulum loses energy,
measure amplitude.
theta ∝amplitude
Q ≈ pi / ln(previousTheta/currentTheta)

Not tested yet.

Problem is testing Q by decaying the pendulum stops the clock!  Acceptable during testing, but I’d prefer something that didn’t!

In the Mk1 Clock I used an electrical Q calculation based on the frequency spread of the pendulum. (Q indicated by how far an oscillator deviates from centre frequency – a bell-curve. Sharp is high Q, broad is low Q)  Used the period logged to the PC.

John Haine said no!  Is there another way of deriving Q from the information normally logged to the PC for analysis.  The amplitude and period of every beat is available.

My head hurts!

Dave

 

 

 

 

 

Are you sure that’s right? According to Wikepedia (never known to be wrong!), Q is initial energy / loss of energy in one radian,  which might be initial amplitude×pi/(loss of amplitude in one swing)

I suspect if you have a high Q pendulum, trying to measure it in one swing will give erratic results as the loss in one swing will be very small.

[John HaineParticipant @johnhaine32865]

There’s a plethora of definitions of Q which all give the same number – each definition implies a different method of measurement.

Almost by definition, if the pendulum is impulsed every cycle so its amplitude is constant you can’t use a decrement method.

But I guess if you had an accurate number for the impulse energy and know the steady-state amplitude you could calculate it.  In fact for the Robertson clock I know the energy input since the net drop and mass of the gravity arm weight is known, so easy to calculate Q – came out at ~10,000 if I recall.

What I did with the Arduinome retrospectively on logged data was to compute the amplitude of each cycle between impulses – 30 of them.  Then by looking at the decay you can compute the decrement rate and hence the Q.  It would be equally possible to do this in real time if you are using the “pulse every N” or probably also “pulse when amplitude reaches threshold” strategy.  We had a long discussion in a thread a while back on Q definitions if you recall.  One definition is “(average energy / loss of energy in one cycle) x K” but I can never remember what K is, may be pi or 1/pi…

[duncan webster 1]

On duncan webster 1 Said:

On SillyOldDuffer Said:

Back to measuring. …

Dave

 

 

 

 

 

Are you sure that’s right? According to Wikepedia (never known to be wrong!), Q is initial energy / loss of energy in one radian,  which might be initial amplitude×pi/(loss of amplitude in one swing)

I suspect if you have a high Q pendulum, trying to measure it in one swing will give erratic results as the loss in one swing will be very small.

Not entirely sure!  The code is in development, compiles ok, otherwise untested. I’ll read Wikipedia again and have another go.

Mea culpa, I hit send in a rush when my daughter arrived earlier, and the description above is incomplete.  Doesn’t say the code calculates multiple beats, which is vital.  As the pendulum decays, each step compares current swing (theta) with previous swing.  Q pops out at the end, after as many beats as the pendulum takes to decay.

Keen to measure Q because I reckon the pendulum is faulty.  Need to prove it’s bust before I fix it!  If possible though, I’d like to do it off-line and without turning the impulse off.

Dave

 

[Michael GilliganParticipant @michaelgilligan61133]

Ref. … The earlier discussion:

https://www.model-engineer.co.uk/forums/topic/pendulum-q-value-and-measurement-methods/

MichaelG.

[John Haine]

On John Haine Said:

…  We had a long discussion in a thread a while back on Q definitions if you recall.  One definition is “(average energy / loss of energy in one cycle) x K” but I can never remember what K is, may be pi or 1/pi…

All noted thanks.  Part of the problem is coming back to this after a long break.  First disaster was finding the box of bits needed was empty.  Second is the notes are missing too.  Obviously together in a safe place, and I can’t find it!  Third. last night, the CAD design is in Fusion and I’ve switched to Solid Edge.

What I’m doing with Q is based on faulty memory.  I’m making new notes, but it’s slow going.

Dave

[duncan webster 1Participant @duncanwebster1]

You learn something new on this forum every day! Turns out that for very small change between amplitude between successive swings, both expressions give the same answer. Mine takes a lot less processor power.

[blowlampParticipant @blowlamp]

What are we even measuring? 😉

[duncan webster 1Participant @duncanwebster1]

For this who like this sort of thing, this is very interesting. It explains derivation of Q and the exponential decay

https://xueqilin.me/engsci-2t4/phy180/labs/lab-2/lab-pendulum-1.pdf

[Michael GilliganParticipant @michaelgilligan61133]

On a quick skim through, that looks an excellent piece of work, Duncan

Thanks for sharing it

My only note of caution [barely awake] is the ‘ballpark’ in which the author is working … Q=310 … Does that help or hinder our comprehension, I wonder.

MichaelG.

[blowlamp]

On blowlamp Said:

What are we even measuring? 😉

Thoroughly enjoyed the video thanks!

Now I’m tempted to display clock-time in ora, semesters, moons, sennights, with thrds, quartes, fytes and syxtes!   Or for those who prefer them, ounces and inches.

By coincidence I was pondering less than second time yesterday.   At the moment the clock counts in seconds and microseconds.  Probably not enough precision if the clock ever work properly, so I might upgrade it to count in seconds and nanoseconds.   That led me to wonder if there is a minimum time.   Is time a smooth analogue flow, or, is it a series of tiny steps?  Physics suggests many physical properties are quanta, so time might be too.   I believe science has so far failed to answer the question.

What precision to count in is part of a bigger problem.  After I’ve got measuring Q to work, I have to tackle setting the clock. Thought it would be easy and it’s not.  In a nutshell, pendulum and GPS ticks are asynchronous, differing in both frequency and phase.  I’m struggling to ask the question, let alone answer it.

Martin’s light relief is most welcome!

Just turned on a electric fire to take the chill off and got a puff of smoke…

Dave

 

[Author]

Posts