Escapement error vs driving weight

[peterbalch]

I'm trying to understand pendulums. I seem to be descending into a real rabbit hole. I'm worrying about the driving weight vs the pendulum period. I suppose I'm talking about "escapement error".

(For reference, it's a 150 year-old "black forest shield clock", wooden frame, brass bushed, pendulum bob = 7g, pendulum rod = 7g, trapeze suspension, "strip pallet" escapement, 550g weight. Runs well but keeps poor time.)

I think the "escapement error" effect depends on Q. So, first question: what is the Q of my clock?

I made a video of a free-swinging pendulum and analysed the individual frames. (To get a "free-swinging pendulum" I detached the weight.) Q varied as the pendulum ran out of energy but was in the range 180 to 300. Let's say

Q = 200

Is it reasonable to measure Q with a free-swinging pendulum? Q is

2*pi*(energy at start of a swing) / (energy lost during the swing)

Some of the energy lost is due to the friction of the escapement - e.g. the pallets rubbing on the escape wheel - so if the escapement isn't running, it's not a fair test. One could calculate how much energy the weight puts into the pendulum during each swing. But to do that, you'd need to know the efficiency of the geartrain. (I did the maths and the gear train is extremely inefficient.)


In theory, does the mass of the driving weight affect the period of the pendulum? Yes. But by how much?

Consider a foliot clock. Q = 0. The foliot escapement works using the inertia of the foliot arms vs the torque from the escapement wheel - the clock will run faster as the driving weight increases and slower as the foliot inertia increases. The period of a pendulum clock is a combination of the classical pendulum we learned in high-school and the inertia of the pendulum acting like a foliot.

(Imagine the clock in space - only the inertia would matter. Or imagine removing a pendulum leaving only the crutch: the crutch is driven hard and slows the escape wheel due to its inertia. In both cases, it's working like a foliot. I think the frequency of a foliot is proportional to the square root of the driving weight.)

With a pendulum clock, Q is between 200 (for my clock) and 100,000 (for a Shortt-Synchronome). If Q is small, then the torque becomes more important. What is the equation for the period vs driving weight?

My weight weighs 550g. I added 10% to it and the pendulum period decreased by 1%. A ten-to-one ratio.

Is that a bigger than usual effect? Is it big because of the low Q? Or the light pendulum bob? Or the heavy pendulum rod? Or the strip escapement? (It's definitely not circular error.)

Thanks

Peter

 

[Uhralt]

I can't help with your math questions, but here are some thoughts: A lot of energy is lost in the trapeze suspension. In a Black Forest clock a pendulum pushed with the driving weight removed will swing only for a very short time. If one replaces the trapeze with a suspension spring (not that I suggest doing this) it will swing much longer. There are some additional factors involved that have an influence on time keeping, like humidity. The wooden plates of the clock and the frame swell and shrink with changes in humidity. The affects the meshing of the gears and how much power is lost to friction. Temperature also has a large effect. The iron rod of the pendulum will change it length dependent on temperature. The heavy rod in comparison to the weight of the bob makes the position of the real center of gravity rather obscure with hard to predict changes when the ambient temperature changes.

Uhralt

 

[leeinv66]

I am moving this over to clock repair.

 

[Dick Feldman]

If you are trying to understand the physics of pendulums, the information source I am familiar with is The Modern Clock by Ward Goodrich. Don't be confused as that book was written a hundred plus years ago.
Second caveat it to know that all of the calculations are based on theoretical pendulums.
Theoretical pendulums have only one dimension, that being length. They are devoid of things like friction. Once they are put into motion, they will swing forever. How's that for a goal for a clock?
Unfortunately, theoretical pendulums are out of reach of us who live every day with gravity, wind and friction.
I believe you are mixing apples and oranges by adding weight and trying to come up with some sort of meaningful number.
A practical approach would be to run the clock off the energy from a spring powered scale. When the clock quits, read the weight measured on the scale and add a reasonable amount (I have heard 25%) to that to get an proper appropriate weight value. By appropriate weight value it is meant the weight will run the clock but not add "too much" pressure on the clock pivots and cause premature wear.
If you are insistent in calculating the effect of drive weight on the clock movement escapement, be sure the friction on every pivot is zero.
That is how I feel,
Dick

 

[John MacArthur]

This kind of stuff has been pondered and argued for something like 300 years with no real definitive solution. As Dick says, too many variables.
Johnny

 

[tom427cid]

Some observations, as has been stated re friction et al, it seems that the weight does have an effect on swing/amplitude of the pendulum. But only to the extent of overswing. After that I think it is just wasted energy and could cause premature wear. I have experimented with determining how little weight one can reliably use and maintain accuracy with a seconds beat pendulum. A heavier bob seems to provide the the most stable rate. At present working with a 6lb bob and 20oz of weight

 

[peterbalch]

Thank you for your answers.

> Uhralt: A lot of energy is lost in the trapeze suspension.

Interesting. What I've read elsewhere is that air-resistance is the major effect but that's usually being said by people building high-quality pendulums. Friction acts differently from aerodynamic drag (I've no idea whether that's important or not).

> In a Black Forest clock a pendulum pushed with the driving weight removed will swing only for a very short time.

Yes but longer that I expected. It was still swinging after a minute.

> influence on time keeping, like humidity ... temperature.

I'm not at that stage of refinement yet. The pendulum's adjustment is by turning the whole bob (a disk) round by +/- one whole turn - i.e. 0.5mm or 0.15%. As you say, the wooden frame is susceptible to changes in temperature and humidity. The bob is very light and is not enclosed so is easily affected by draughts. The chain weighs around 10% of the weight so the torque varies by 20%.

Those are all huge effects. I think the thermal expansion of the rod would make a difference of less than 1 sec per day. I'm at the opposite end of the horological spectrum from the people who are putting invar rods in a vacuum.


> Dick: The Modern Clock by Ward Goodrich.

I've read it (well, OK, I skimmed through it and checked the index!) but could find nothing on how the weight affected the period.

The standard answer is that the weight doesn't affect the period. Of course, in reality it does. Hence there are fusees, remontoires, etc. even for pendulum clocks.

> I believe you are mixing apples and oranges by adding weight and trying to come up with some sort of meaningful number.

You could well be right. But I know that, practically, I added 10% to the weight and decreased the period by 1%. Is that normal? Someone in the NAWCC will have done it.

I'm pretty sure it's going to depend on Q, the mass of the pendulum rod and the escapement mechanism (recoil vs deadbeat?).

There are theoretical treatments that include friction, drag and a crude model of an escapement. Like this

Friction and Dynamics of Verge and Foliot: How the Invention of the Pendulum Made Clocks Much More Accurate

The tower clocks designed and built in Europe starting from the end of the 13th century employed the “verge and foliot escapement” mechanism. This mechanism provided a relatively low accuracy of time measurement. The introduction of the pendulum into the clock mechanism by Christiaan Huygens in...

www.mdpi.com

(It's worth reading his analysis of foliot and verge - he reckons foliots only work because of friction, not in spite of it.)

> a spring powered scale ... read the weight measured ... add ... 25%

Yes, I've come across that figure. But I'm not asking "what is the minimum weight" or "what is the ideal weight". The clock already has a weight and runs reliably. I think it's not the original weight; I think my father made it 60 years ago. But the clock always ran and he wound it and adjusted it every night. It still runs and still needs adjusting. I just want it to be as trouble free as a quartz clock I can buy for pennies.

> be sure the friction on every pivot is zero.

Zero? Hah! I wish! One calculation that's easy to make is to ask what is the weight and how far does it move in a day? Combine that with the Q value and you can calculate how efficient the gear-train is. I reckon my gear-train loses 85% of the energy before it gets to the pendulum. Presumably that's due to friction in the bearings, meshing teeth and rubbing pallets.


> Peter R Lee: I am moving this over to clock repair.

OK. But I'm not wanting to repair it, I'm wanting to electronically regulate it.

I'm constrained because I want to make no changes to the clock other than (e.g.) add a magnet to the pendulum. I want the mechanism to be inconspicuous, perhaps just under the clock (it's a wag-on-the-wall that sits on a bracket).

Putting the magnet and a coil just under the clock is the least effective place - sensible people put the magnet/coil at the end of the pendulum. I want it to run on AA cells for a year.

I have a big problem with the large variability of the pendulum period: successive swings can vary by 1% or 2%. Which may be normal when Q = 200. There must be lots of NAWCC people who've put a timegrapher on a wag-on-the-wall.

I can get it to synchronise with a pulse sent blindly from some electronics but it takes a huge current - milliAmps - whereas I want under 200uA. If I sense the swing and apply the pulse at the "right" time, I can change the period by 1% (on average) with an average current of 500uA but then the sensor takes milliAmps. It's a difficult problem mainly because it's such a poor-quality clock.

I've already built an auto-winder and one side-effect of the design is that it can vary the average force on the chain.

It occurred to me that I don't have to sense the pendulum. The auto-winder can count how much it's turned and so know whether the clock is gaining or losing. The auto-winder can then adjust the force on the chain appropriately.

It's a weird idea. I don't think anyone's done it before. If it works, it would be great: a combined auto-winder and regulator that can be fitted to any wag-on-the-wall clock.

But it depends on the weight changing the pendulum period. Does it in general? By how much?

Peter

 

[John MacArthur]

Rawlings "Science of Clocks and Watches" Get the newer printing, post '94, I believe
Johnny

 

[Willie X]

"Invar rod in a vacuum" ... wow!

This thread reminds me of a very old thread, when I used to talk about the flatulent bacteria? Ha Willie X

 

[DeweyC]

Is that a bigger than usual effect? Is it big because of the low Q? Or the light pendulum bob? Or the heavy pendulum rod? Or the strip escapement? (It's definitely not circular error.)

Thanks

Peter

Peter,

I am a bit unclear about your pursuit.

"Q" is a measure of an oscillator's efficiency. The higher the Q, the better the tendency of the oscillator to maintain its resonant frequency. In electronics higher Q results in a more selective tuning.

In mechanical oscillators, this translates into higher Q (the band of frequencies at which the oscillator will resonate) allowing less energy needed to maintain the desired frequency; of course you want the oscillator to resonate at the desired frequency and only that frequency. But the more selective (higher Q), the greater the cost in dollars.

So yes, a low Q pendulum would be expected to be sensitive to changes in power. The same is true in watches, watches with a higher moment of inertia for the balance keep tighter rates than watches using a "lighter" balance because they are less sensitive to power difference at 270 and 180 degrees of amplitude.

Discussions of this are beyond Goodrich and Hamilton. I know scientist/clock collectors have studied this but I have no idea where they published. It is one of those lanes for pusuing absolute precision in mechanical timekeeping and since Q was derived for electronics, it post dates almost all of the theoretical writers in the clock watch industry.

 

[DeweyC]

Oops. Meant Rawlings.

 

[peterbalch]

Dewey,

> I am a bit unclear about your pursuit.

Yes, I'm sorry about that. What I'm trying to do and the research I've done so far is a long way from what most people here are doing. It would take several pages to describe what I've done so far.

I'm trying to use electronics to "improve" an old black forest shield clock. The changes to the clock must be minimal and reversible. There are two "improvements": an auto-winder and a regulator. (And, yes, I realise that a lot of people would see the idiosyncrasies of a shield clock as part of its charm.)

After lots of development, I've made an auto-winder that I think is an improvement on previous designs. So now I'm thinking about a regulator.

I think that maybe I can make a regulator that's built into the auto-winder. It does the regulation by changing the force in the driving chain. That's a really big 'maybe'. For it to work, it would depend on how the force on the chain affects the period of the pendulum. There are discussions in this forum but they mostly seem to be comments by people who don't actually know. They can't even agree of whether more force makes a clock run faster or slower. That's not meant to be a criticism - it's not a subject that affects most clockmakers.

There are two questions: what's the theory and what happens in practice?

You could say to me: "just build something and see if it works". Building it would take a lot of effort and it would be nice to know in advance if it's unlikely to ever work. And I'd like the solution to be applicable to all shield clocks or all wag-on-the-wall clocks or all long-case clocks or whatever. That way, anything I find out will be useful to other people.

> The higher the Q, the better the tendency of the oscillator to maintain its resonant frequency.

I do electronic design and digital signal processing so I'm happy with Q in that context. It was a it of a surprise when I found clockmakers using the term. I've built a timegrapher and will maybe do some statistics to see whether the variance in period is what is predicted by the Q of my clock. This person introduces another term "P" which is to do with the variance of the period. And this person introduces entropy! So maybe there's not a simple relationship between Q and variance.

> So yes, a low Q pendulum would be expected to be sensitive to changes in power.

Aha! You're the first person I've heard say that. Do you have a reference or suggestions where I can search or discussions on NAWCC? Any thoughts on what the relationship is?

My thoughts go like this: my shield clock has low Q. It keeps poor time because it has low Q. It is sensitive to the changes in power because it has low Q. Therefore I can use changes in power to regulate it. You have a better clock that has high Q. It keeps better time because it has high Q. It is less sensitive to the changes in power because it has high Q. Therefore if I use changes in power to regulate it, the effect will be smaller but I only need a small effect because it already keeps good time.

In other words, you can use changes in power to regulate any clock with any Q. That's nonsense of course. It's not going to work with a grasshopper escapement.

Peter

 

[R. Croswell]

Dewey,

> I am a bit unclear about your pursuit.

Yes, I'm sorry about that. What I'm trying to do and the research I've done so far is a long way from what most people here are doing. It would take several pages to describe what I've done so far.

I'm trying to use electronics to "improve" an old black forest shield clock. The changes to the clock must be minimal and reversible. There are two "improvements": an auto-winder and a regulator. (And, yes, I realise that a lot of people would see the idiosyncrasies of a shield clock as part of its charm.)

After lots of development, I've made an auto-winder that I think is an improvement on previous designs. So now I'm thinking about a regulator.

I think that maybe I can make a regulator that's built into the auto-winder. It does the regulation by changing the force in the driving chain. That's a really big 'maybe'. For it to work, it would depend on how the force on the chain affects the period of the pendulum. There are discussions in this forum but they mostly seem to be comments by people who don't actually know. They can't even agree of whether more force makes a clock run faster or slower. That's not meant to be a criticism - it's not a subject that affects most clockmakers.

There are two questions: what's the theory and what happens in practice?

You could say to me: "just build something and see if it works". Building it would take a lot of effort and it would be nice to know in advance if it's unlikely to ever work. And I'd like the solution to be applicable to all shield clocks or all wag-on-the-wall clocks or all long-case clocks or whatever. That way, anything I find out will be useful to other people.

> The higher the Q, the better the tendency of the oscillator to maintain its resonant frequency.

I do electronic design and digital signal processing so I'm happy with Q in that context. It was a it of a surprise when I found clockmakers using the term. I've built a timegrapher and will maybe do some statistics to see whether the variance in period is what is predicted by the Q of my clock. This person introduces another term "P" which is to do with the variance of the period. And this person introduces entropy! So maybe there's not a simple relationship between Q and variance.

> So yes, a low Q pendulum would be expected to be sensitive to changes in power.

Aha! You're the first person I've heard say that. Do you have a reference or suggestions where I can search or discussions on NAWCC? Any thoughts on what the relationship is?

My thoughts go like this: my shield clock has low Q. It keeps poor time because it has low Q. It is sensitive to the changes in power because it has low Q. Therefore I can use changes in power to regulate it. You have a better clock that has high Q. It keeps better time because it has high Q. It is less sensitive to the changes in power because it has high Q. Therefore if I use changes in power to regulate it, the effect will be smaller but I only need a small effect because it already keeps good time.

In other words, you can use changes in power to regulate any clock with any Q. That's nonsense of course. It's not going to work with a grasshopper escapement.

Peter

Peter, I find this interesting, but from a practical standpoint, the weight is what it is so unless you plan to be changing the weights why does it matter how much the pendulum rate changes? There is one variable that I didn't see mentioned. If we are talking about a recoil escapement, increasing the weight is going to increase the overswing and increase the duration of the impulse. If we are talking about a deadbeat escapement, the entire impulse face of the verge is used regardless of the weight, so while the force of the impulse may increase the duration of the impulse (in degrees) remains a constant. I don't believe that you can derive a mathematical formula to predict the rate change in the period of the pendulum based solely on the change in the driving weight without considering variables in the going train and type of escapement which will be different, and unknown, for individual clocks. Yes, I think the rabbit hole is coming into view.

RC

 

[shutterbug]

Maybe the bottom line is that mechanical clocks never were, and never will be perfect time keepers.

 

[peterbalch]

Yes I do plan to change the weight! Sort of.

If you look at my auto-winder, you'll see that the weight moves slightly side-to-side. In other words, the tension due to the weight shifts from the left-hand chain (usually the taut chain) to the right-hand chain (usually the loose chain). I haven't worked out the amount of change yet. So with a little cleverness inside the winder I can change the "weight". Will the clock mind? No. The weight is 550g and the chain is 50g so originally as the weight descended, the tension increased by 20%. The clock is happy with a varying weight. How will the clock react? I think it will change it's speed.

 

[DeweyC]

Dewey,

> I am a bit unclear about your pursuit.

Yes, I'm sorry about that. What I'm trying to do and the research I've done so far is a long way from what most people here are doing. It would take several pages to describe what I've done so far.

I'm trying to use electronics to "improve" an old black forest shield clock. The changes to the clock must be minimal and reversible. There are two "improvements": an auto-winder and a regulator. (And, yes, I realise that a lot of people would see the idiosyncrasies of a shield clock as part of its charm.)

After lots of development, I've made an auto-winder that I think is an improvement on previous designs. So now I'm thinking about a regulator.

I think that maybe I can make a regulator that's built into the auto-winder. It does the regulation by changing the force in the driving chain. That's a really big 'maybe'. For it to work, it would depend on how the force on the chain affects the period of the pendulum. There are discussions in this forum but they mostly seem to be comments by people who don't actually know. They can't even agree of whether more force makes a clock run faster or slower. That's not meant to be a criticism - it's not a subject that affects most clockmakers.

There are two questions: what's the theory and what happens in practice?

You could say to me: "just build something and see if it works". Building it would take a lot of effort and it would be nice to know in advance if it's unlikely to ever work. And I'd like the solution to be applicable to all shield clocks or all wag-on-the-wall clocks or all long-case clocks or whatever. That way, anything I find out will be useful to other people.

> The higher the Q, the better the tendency of the oscillator to maintain its resonant frequency.

I do electronic design and digital signal processing so I'm happy with Q in that context. It was a it of a surprise when I found clockmakers using the term. I've built a timegrapher and will maybe do some statistics to see whether the variance in period is what is predicted by the Q of my clock. This person introduces another term "P" which is to do with the variance of the period. And this person introduces entropy! So maybe there's not a simple relationship between Q and variance.

> So yes, a low Q pendulum would be expected to be sensitive to changes in power.

Aha! You're the first person I've heard say that. Do you have a reference or suggestions where I can search or discussions on NAWCC? Any thoughts on what the relationship is?

My thoughts go like this: my shield clock has low Q. It keeps poor time because it has low Q. It is sensitive to the changes in power because it has low Q. Therefore I can use changes in power to regulate it. You have a better clock that has high Q. It keeps better time because it has high Q. It is less sensitive to the changes in power because it has high Q. Therefore if I use changes in power to regulate it, the effect will be smaller but I only need a small effect because it already keeps good time.

In other words, you can use changes in power to regulate any clock with any Q. That's nonsense of course. It's not going to work with a grasshopper escapement.

Peter

Peter,

Like I said, I have no idea where this stuff gets published. Thirty years ago (1995) there was a Clocks listserve out of Syracuse populated by scientists, engineers, conservators and such. This is where we explored this the most.

I generally agree with your thoughts. However, the simpler way is to design a high Q oscillator and a means to avoid variations in driving force. Weights, the Model 22 with its very long and light mainspring (will run over 48 hours; but was wound daily to keep it in the same section of the power curve), chronometers with high q and fusees, etc. Modern wrist watches use autowind, not as a convenience to the user, but to ensure constant power to the escapement which is a major factor in their remarkable timekeeping.

I learned about Q in college physics back in 1971. No idea when the concept was developed.

The only person I can recall who may have some information on this would be Bryan Mumford, of Microset fame.

 

[DeweyC]

Maybe the bottom line is that mechanical clocks never were, and never will be perfect time keepers.

I agree in general; but perfect compared to what? Solar time? Sidereal? USNO (rubidium fountains)?

The search for increased mechanical clock precision went on scientifically from at least 1840 to 1950; and was usurped by electronics and nuclear physics. It drove astronomical regulators (Shortt for example) and chronometers.

Timepieces are simply counters. A 1900 pocketwatch that achieves a count error of 5 out of 86400 is remarkable to me. An astronomical regulator that increased precision by 100 to 1000 times is tighter than the observer error of a someone establishing time using a transit. At that time, I am pretty sure that would have been considered perfect.

 

[Rod Schaffter]

...But I know that, practically, I added 10% to the weight and decreased the period by 1%. Is that normal? Someone in the NAWCC will have done it.

Keep in mind that unless you are adding the additional mass at the exact center of mass of the pendulum, you are changing the effective length of the pendulum...

 

[peterbalch]

Thanks but I'm adding the extra weight to the driving weight - not the pendulum.

I'm calling it "the driving weight" because I don't know what the correct term is - everyone seems to just call it "the weight". I mean the thing that hangs on a chain and makes the clock go. Is there another term for it? Maybe it's a "going weight" as opposed to a "striking weight".

I added 10% to the driving weight and the pendulum period decreased by 1%. A ten-to-one ratio. Is that normal for this kind of clock? Why did it happen?

I asked an engineer friend who said that as a child his family were given a cuckoo clock. The first thing he did was to hang the cuckoo's weight onto the driving chain - doubling the driving weight. The clock went "significantly faster". (He hasn't yet explained why it did in spite of decades as a professor.) My belief is that as you increase the torque on the "main wheel" of a weight-driven or spring-driven clock, (with a pendulum or a balance wheel or a foliot) then the clock runs faster.

There is an NAWCC thread in which some people say a pendulum clock runs faster. But some say, no, it runs slower due to circular error - I think they're wrong but what do I know? Circular error is tiny; 1% is huge.

It seems a pretty fundamental property of a clock - people have made fusees for 500 years to "solve" the "problem". I'm sure somebody somewhere knows

(This isn't just idle curiosity. I want to built a regulator that changes the tension in the driving chain in order to electronically regulate the clock.)

Peter

 

[JayKosta]

An important question is 'what stops / limits the L/R swing distance?'
Is the 'stop' caused by loss of momentum due to gravity alone?
Is the 'stop' due to some mechanical interference? And how does that interference affect the speed of the pendulum compared to gravity?
How would 'drive weight changes affect the mechanical interference, or the pure gravity?

 

[shutterbug]

I think it's mostly gravity. A spring suspension clearly imparts some resistance, but a silk string suspension would impart very little and the action is very similar. The impulse itself also uses power, so that's a factor ... and a recoil type probably uses the most because it reverses the whole train. But the impulse clearly delivers more power than it uses.

 

[peterbalch]

> 'what stops / limits the L/R swing distance?'

I've just started reading papers on the physics of escapements. I think the answer is:

If the escapement has a recoil - like a verge or recoil anchor escapement or (some? all?) grasshoppers - then the escapement applies a torque during the overswing. That will help stop the pendulum. The more the overswing, the longer the torque is applied thus limiting the swing.

If the escapement has no recoil like a deadbeat then the pendulum is stopped just by gravity.

> How would 'drive weight changes affect the mechanical interference, or the pure gravity?

It's complicated. Oh boy, is it complicated! I'll try to answer in a few days if I make any progress with the papers.

 

[John MacArthur]

Friction, friction, friction....don't forget friction.
Johnny

 

[peterbalch]

>> 'what stops / limits the L/R swing distance?'
>Friction, friction, friction

Personally, I don't think so. If the swing is 4 degrees and Q=600 (a mediocre clock) then after one cycle, the swing will be 3.98 degrees.


How important is friction (i.e. one surface sliding on another)? This site says that the efficiency of an escapement can never exceed 50%. He doesn't say where he got that number from. (I think he's only talking about escapements where a tooth rubs on a pallet - not, e.g., a gravity escapement.)

So if the efficiency is 50%, where does the other 50% of the energy go? A tooth rubbing on a pallet making friction? I'm sure I've seen websites saying the main losses of a pendulum are air-resistance - which is treated as viscous drag.

One measures the Q of a pendulum by removing the driving weight and watching it swing. For instance, if the Q is 600 then 1% of the energy of the swing is lost to air resistance. If the pendulum has 1 joule of potential energy at the start of the swing then it has lost 0.01 joules after one cycle. But if you've removed the driving weight, the Q you've measured doesn't include friction.

That 0.01 joules is replenished by the escapement. If the escapement is at best 50% efficient then 0.02 joules must have been given to the to the escapement. The losses due to friction in the escapement are equal to the losses due to air resistance of a free pendulum.

So when someone says "the Q of this clock's pendulum is 4000" they're saying the free pendulum Q is 4000. Once they start driving it with the escapement the Q will be 2000.

Or maybe I'm just talking nonsense.

I measure the Q of my clock as 200. Given the angle and mass of the pendulum I can calculate the losses during the swing. That tells me how much energy the pendulum loses in a day. I know the mass of the weight and it's drop which tells me how much energy is supplied in a day. And the answer is ...

The whole mechanism is 15% efficient. 85% is lost to friction in the gears and bearings and escapement. Is that typical?

Peter

 

[peterbalch]

I think I'm starting to understand why and how the driving weight affects the escapement error: why my clock runs faster when I increase the weight. Whether what I think matches reality is a different matter!

The NAWCC paper by Dr. George Feinstein is tough going but in it he says "any forces applied to the pendulum and acting toward the position of equilibrium reduce the period of the pendulum, whereas forces acting away from the position of equilibrium increase the period".

His paper seems to be entirely about applying an "impulse" at different times during the pendulums cycle and what effect that has. An impulse is an instantaneous increase in the pendulum's velocity - like flicking it with your fingernail. I think he applies his ("positive") impulse in the direction of movement of the (theoretical) pendulum - he's always speeding it up.

Presumably a "negative" impulse does the opposite. Feinstein doesn't discuss how the size of the impulse affects the period - only when it is applied.

A real escapement isn't an impulse. It applies a torque over a certain section of the pendulum's cycle. That torque may vary during the cycle but let's assume it's constant and is always "on" or "off".

Consider a verge escapement. The torque is always "on" but changes direction as one pallet then the other is pushed by the crown wheel. The direction of the torque varies like this:

The dotted lines show when the escapement advances by one tooth. The red and blue arrows show the direction of the torque. The torque in the red and blue parts between the dotted lines are equal and opposite so cancel and do not speed-up or slow-down the clock. In the parts outside the dotted lines (the overswing), the torque is towards the centre and so decreases the period of the pendulum. Increasing the torque will (presumably) speed-up the clock.

This paper discusses when torques are applied by a verge or a dead-beat escapement. They say a dead-beat escapement is like this

Note that they think the torque is not applied symmetrically about the rest position of the pendulum. (I don't know why they think so but it's clearly there in Fig 3.) The same assertion is made in this paper (Fig 5). On average, more torque is applied while the pendulum is moving away from the centre so increasing the torque will (presumably) slow down the clock.

My clock has a strip-pallet recoil escapement. According to Fig 3 of the same paper the torque is applied very much like a verge. I suppose that's true of all recoil escapements.

So the result is that if you increase the driving-weight and if the clock has a recoil escapement then it will speed-up. If it has a deadbeat escapement it will slow-down.

The amount of recoil speed-up depends on how big the overswing is. The amount of deadbeat slow-down depends on how asymmetric the deadbeat pallets are.

Is that what happens in practice with a real clock?

I don't have lots of clocks to test but I've written a simulation. It behaves as I've described. With a simulated recoil escapement, increasing the torque provided by the escapement speeds-up the clock. The amount of gain depends on the size of the overswing and how much extra torque is applied. A typical result is that 10% extra on the driving-weight makes the clock gain by 1%. Which is what I've measured in real-life.

With a simulated symmetrical deadbeat escapement, increasing the torque makes a tiny difference to the speed of the clock. That tiny difference is due to the swing angle increasing which leads to increased circular error (a loss of a few dozen parts per million).

With a simulated asymmetrical deadbeat escapement like the one above, increasing the torque makes a small difference to the speed of the clock. The size of that small difference depends on how big the asymmetry is but a typical value is that the clock loses by 0.02%.

Of course, I'm not an expert and may be talking complete nonsense.

Peter

 

[R. Croswell]

While we are all over-thinking this, consider one more factor. Consider the actual motion of the escape wheel and the rest of the going train. It is a "stop and go" motion. The escape wheel (and the other moving parts) all have mass and therefore inertia. When we increase driving force the escape wheel will accelerate to a higher velocity between stop and go and impact the next pallet more quickly and with for force. Said another way, the elapsed time from release at the entrance pallet to impact at the exit pallet will be reduced as the driving weight is increased because the escape wheel moves more quickly. Especially on a recoil escapement, and to a lesser extent a half-deadbeat, It would seem that this would result in the driving escapement attempting to force the pendulum to move faster than its theoretical natural period.

RC

 

[JayKosta]

I'm interested in this because my GF clock (weight on chain) runs fast when the weight is fully raised. After the weight has dropped in a day or so, the rate reduces and becomes consistent.
This is my old Jauch 77 with marginal drive power due to wear. I have disabled the chime, strike, and moon mech and the time-only function seems good - for now anyway.

I wonder if the weight distribution due to chain length is making a difference.

I might try attaching the loose end of the chain to the top of the weight to see if that has an effect - anyone have experience doing that?

 

[shutterbug]

With a weight powered clock, there is very little difference in power between fully wound and nearly on the floor. That's the beauty of weights compared to springs.

 

[peterbalch]

there is very little difference in power between fully wound and nearly on the floor.

With my clock, the weight is 550g and the chain is 50g - about 10-to-1.

So the "weight" seen by the main wheel increases by 20% during the day.

20% is enough to add 2% to the speed. It's a recoil escapement so increased weight increases the speed.

 

[shutterbug]

I think it would increase 10° during the week, not every day ... yes?
Looking back at your first post, I suspect that the trapeze suspension might cause a lot of friction.

 

[R. Croswell]

With my clock, the weight is 550g and the chain is 50g - about 10-to-1.

So the "weight" seen by the main wheel increases by 20% during the day.

20% is enough to add 2% to the speed. It's a recoil escapement so increased weight increases the speed.

My first reaction is that 550 g doesn't seem like much weight, so why not increase this? That would make the driving force greater relative to counter forces of the chain weight, internal friction etc.

RC

 

[peterbalch]

Sorry, I should have been more clear. I was providing a counter example to the statement "there is very little difference in power between fully wound and nearly on the floor".

It might be claimed that some clocks have a heavier weight but in that case they'll have a heavier chain too. (The safe working load of a chain is proportional to its weight per metre. If you double the driving-weight you'll need to double the weight of the chain.) Does anyone know what is a "normal" ratio?

Admittedly, my clock may have a particularly light weight and a heavy chain but it doesn't look that way to my eye. (And the clock works 100% of the time with a 550g weight.)

I'm interested in this because my GF clock (weight on chain) runs fast when the weight is fully raised. After the weight has dropped in a day or so, the rate reduces and becomes consistent.

That doesn't sound like the effect I was investigating.

It sounds more like some sort of threshold being reached. As the weight falls the torque on the escape wheel will increase and hence the swing angle. If the swing starts off below some threshold, could the escapement be going wrong - a pallet not catching a tooth properly? When the swing increases, the pallets make a more positive engagement?

As a test, what if you added 10% to the weight when it's at its highest? If that fixes the problem then what?

Don't Jauch anchors usually have adjustable pallets? Or is adjusting the pallets an admission of "my clock is so worn it's time to replace the movement"?

 

[JayKosta]

...
That doesn't sound like the effect I was investigating.

It sounds more like some sort of threshold being reached. As the weight falls the torque on the escape wheel will increase and hence the swing angle. If the swing starts off below some threshold, could the escapement be going wrong - a pallet not catching a tooth properly? When the swing increases, the pallets make a more positive engagement?

As a test, what if you added 10% to the weight when it's at its highest? If that fixes the problem then what?

Don't Jauch anchors usually have adjustable pallets? Or is adjusting the pallets an admission of "my clock is so worn it's time to replace the movement"?

----------------------------------------------
I don't have knowledge about 'adjusting pallets' and have only done simple 'while assembled' clean & oil' the pivots, and getting an even and steady 'beat'.
Yesterday I attached the loose end of the drive weight chain to the top of the drive weight - so there is some chain weight added to the weight.
With the weight fully raised, the running rate seems a little more consistent in the 1st 24 hours, but will have to see how/if the rate changes over several days - as the amount of chain weight shifts between the loose-chain side and the weight side.

I'm reluctant to add more 'dead weight' to the weight itself for fear of increasing wear. YES 'proper repair' or replacement would be the 'professional answer', but at this point I am content to DIY tinker.

 

[R. Croswell]

It might be claimed that some clocks have a heavier weight but in that case they'll have a heavier chain too. (The safe working load of a chain is proportional to its weight per metre. If you double the driving-weight you'll need to double the weight of the chain.) ........... Admittedly, my clock may have a particularly light weight and a heavy chain but it doesn't look that way to my eye. (And the clock works 100% of the time with a 550g weight.)

In the real world one never loads the weight chain anywhere near the failure point or even the maximum working load. There are only several sizes of chain and dozens of different size weights. You could probably double the weight and still not come close to busting the chain. The chain size has to properly fit the sprocket, so you can't use a lighter chain, so the only way to change the drive weight to drive chain weight ratio to a more favorable number is to change the weight. The first thing I would check is to be sure that 550g is the correct weight specified by the maker. I can't promise you anything, but if it were my clock, I would not be afraid to increase the weight by 50% temporarily to observe the results.

You said that you "have only done simple 'while assembled' clean & oil' the pivots, .... but at this point I am content to DIY tinker". In all probability your clock isn't performing up to its full potential and would benefit from a proper cleaning, pivot polishing, and perhaps several bushings. After a proper service it will require less power to run and should be less sensitive to small changes in power. You also said, "I'm reluctant to add more 'dead weight' to the weight itself for fear of increasing wear", but you apparently don't have a problem with continuing to run a clock that hasn't had a proper cleaning? Very likely running dirty and/or unpolished pivots will cause more wear than increasing the drive weight. It's probably time to take the next step to learn how to do it right, you won't regret it.

RC

 

[peterbalch]

The first thing I would check is to be sure that 550g is the correct weight specified by the maker.

It's a Black Forest shield clock. My guess is that it's 170 years old. The only text on it is a label for the shop where it was sold: "A Kleiser, watch and clock maker, Stonegate, York". A Kleiser's shop was in Stonegate in 1851 and was bought by his brothers J and M in 1861 and changed its name. He was born in Scollach, Baden: a region that must have produced millions of cheap clocks over a century. Maybe he has importing from there via family connections.

Any thoughts as to what is a typical weight? It's a one-day clock and the weight drops maybe 1.4 m per day. I see that modern one-day Black Forest clocks have weights under 300g but their movements are of far better quality. (Of course, I can't really complain about the quality after 170 years!)

I can't promise you anything, but if it were my clock, I would not be afraid to increase the weight by 50% temporarily to observe the results.

It runs 100% of the time. It has never stopped. What is the advantage of increasing the weight?

Peter

 

[R. Croswell]

It's a Black Forest shield clock. My guess is that it's 170 years old. The only text on it is a label for the shop where it was sold: "A Kleiser, watch and clock maker, Stonegate, York". A Kleiser's shop was in Stonegate in 1851 and was bought by his brothers J and M in 1861 and changed its name. He was born in Scollach, Baden: a region that must have produced millions of cheap clocks over a century. Maybe he has importing from there via family connections.

Any thoughts as to what is a typical weight? It's a one-day clock and the weight drops maybe 1.4 m per day. I see that modern one-day Black Forest clocks have weights under 300g but their movements are of far better quality. (Of course, I can't really complain about the quality after 170 years!)




It runs 100% of the time. It has never stopped. What is the advantage of increasing the weight?

Peter

You were concerned about the change in pendulum rate caused by the changing driving force as more chain is added to the weight side and taken from the "pullup" side. Yiu said the ratio of weight to chain is 10 to 1. Adding to the main weight would reduce this ratio and allow you to see just how much difference it would make. The clock does not just run or not run; it needs to run with a certain amount of overswing (pendulum amplitude) in order to be stable and reliable. If you can find the makers weight specification that is usually the amount of weight to use. You can come pretty close by removing the weight and attaching a fish scale to the weight hook and anchoring the other end of the fish scale. Wind the clock until the fish scale indicates at least as much as the weight you have been using and allow it to run until it stops. The weight indicated is the minimum weight at which the clock will not continue to run. There is no complete agreement on how much additional weight should be added for normal use but 20% to 50% more should be reasonable. At least this should give you some indication whether the weight you have now is "in the ballpark". This o course assumes that the clock is clean and otherwise in good condition.

RC

 

[peterbalch]

OK, you convinced me I should do the test. I don't have a fishing scale that sensitive so I just hung different things on the chain. The clock would work with a 100g weight while ticking very faintly.

I don't believe it would run all day at 100g but it did run for 10 minutes. So in theory 100g + 50% = 150g.

That was with the chain made into a loop so it had no effect on the torque at the main wheel. The chain is 50g so I'd need a 200g weight (and the effect on the torque would of a weight weighing between 150g and 250g).

So the 550g weight is about twice what it "needs" to be. That's useful - thank you for persuafing me to do it.

Does anyone have a Black Forest clock from that period? What's the weight?

... a certain amount of overswing ... in order to be stable and reliable

How does stability depend on overswing? I put a simple "timegrapher" on the pendulum. The period varies by up to 2% from one swing to the next. That seems bad to me. Is that what you meant by stability?

Peter

 

[JayKosta]

... I put a simple "timegrapher" on the pendulum. The period varies by up to 2% from one swing to the next. ...

------------------------------------------
Is that 2% variance between a 'full cycle' complete L/R swing, or is it variance of a L swing compared to a R swing?
Does the variation repeat itself in a regular cycle - e.g. a full turn of a particular wheel / gear?
Is the average rate stable over a measured period - e.g. 1 hour? compared to the 1 hour earlier or later?

 

[peterbalch]

Is that 2% variance between a 'full cycle' complete L/R swing, or is it variance of a L swing compared to a R swing?

It's the variance between one 'full cycle' and the next. Here's a typical graph:

I built the circuit to test quickly what was the effect changing the driving weight. I can easily see it speed-up when I increase the weight. But it's just an Arduino Nano and a Nano uses a resonator so its long-term stability is poor.

Does the variation repeat itself in a regular cycle - e.g. a full turn of a particular wheel / gear?

Is the average rate stable over a measured period - e.g. 1 hour? compared to the 1 hour earlier or later?

Those are difficult questions. A "full turn" could mean 12 hours! I'll need a better "timegrapher".

The clock has been sitting in a box unused for years and I've just revived it with an auto-winder. My recollection is that sometimes it would run 2min per day fast and sometimes 2min per day slow. To me, that seems what I'd expect when I look at the "timegrapher".

Peter

 

[Simon Holt]

But it's just an Arduino Nano and a Nano uses a resonator so its long-term stability is poor.

Maybe a Raspberry Pi Pico would be better for monitoring. It's the only Raspberry Pi with a real-time clock - and it's still widely available during the chip shortage.

Simon

 

[gmorse]

Hi Peter,

The clock has been sitting in a box unused for years and I've just revived it with an auto-winder. My recollection is that sometimes it would run 2min per day fast and sometimes 2min per day slow.

Attempting to do detailed analysis on a movement that's been sat in a box for years and not cleaned or lubricated will be a largely fruitless exercise. Was it cleaned and lubricated before you started to dive into this?

Regards,

Graham

 

[Raymond101]

Hi came across this discussion really interesting. Hope not to but in too much.
This was something I worked on had a clock with a missing pendulum.
The main things in your graphs all show one thing missing. Any pendulum is gravity dependent so the angles of swing from pivot point to center should be equal. To get a perfect Q you would in theory need a mass greater than the weight of suspension rod , air + Rh (density ) . Taking into account location.
ie sea level . Northern or Southern hemispheres. Plus a pendulum will always want to take a slight rotational movement. So assuming the Bob is a perfect circle with the edge tapered to reduce air friction. The weight would have to have an increase to overcome all. But if too heavy would create a reverse effect.
The mathematical calculation would require exact details of the pendulum & location.
Add Micotesla UT & 64 to 160 .apx
As said above great erengineers have been working on this for 300 years +.

 

[peterbalch]

Hi Graham

Was it cleaned and lubricated before you started to dive into this?

Before it went in the box, I dis-assembled it and cleaned the gears (with water then turps). Back then, there wasn't a handy forum with people to give advice but the local library helped. I didn't attempt to wash the bearings. They're brass bushes in a wood chassis and I wasn't sure how to wash them without damaging the wood so I cleaned them with a dry toothbrush. Re-oiled the bearings with a very light mineral oil. Didn't oil the teeth (but I might have run a soft pencil over the escapement teeth). The strip pallet needed slight adjustment to make good contact with the escape wheel.

I cleaned the clock-face with soapy water. One of the side doors was missing so I replaced it with thin ply. (At one time I worked in a museum and was taught that all modern repairs should look like modern repairs. Nowadays I'd probably try and make the repair look original.) I left it at that; I didn't want to do too much to it - it's an antique.

It runs reliably. No one has yet answered the question: is that amount of variance between swings normal? How many minutes a day should it gain/lose? How will the gain/loss per day vary over a year?

It took me very many years to learn "if it ain't bust, don't fix it". I have no evidence that it's in any way "bust".


Hi Raymond

Any pendulum is gravity dependent so the angles of swing from pivot point to center should be equal.

Do you mean the amplitude should be constant from one swing to the next? That would be hard to measure. I assume the period is not constant _because_ the amplitude is not constant.

Would I expect it to be constant? No.

It's a driven pendulum. It's a non-linear system. So it's chaotic. A driven pendulum is a classic example of a chaotic system.

I have simulated a pendulum and escapement (with non of the extra factors you list) and it's chaotic. The period and amplitude are not quite constant.

To get a perfect Q you would in theory need a mass greater than the weight of suspension rod , air + Rh (density ) . ...

The pendulum is what it is. Is its behaviour normal?

Peter

 

[Raymond101]

Hi Peter,
The paper you showed from G Feinstein
Which I have read before the mathematics are a bit trickey .
But he does mention the amplitude & circular movement and gives the formula. As a pendulum has 4 dimensions and theory & practice. If you just look at it in 2D you will wind up going crazy. What most clock pros will but a ruler 0 at center stationary position. And adjust the amplitude to as nearest.
The amplitude must be equal from center line. ( tilt clock may solve small errors)
This clock your playing with is old the wood may be warped . Etc if it works with in a minute or 2 a day except it.
You can add pennies or small weights to help. If the pendulum is free hung it will
Always be inaccurate, due to outside disturbance ie wind , vibration, .....etc .
A pendulum will always have non straight back and forth. Even in space.
As far a adding electronics to this clock I personally think you will spoil a nice wooden clock that as said your father made .
On 1 of my clocks 160yrs old I added a small cone back weight to make circle motion less than a degree . 7 secs a day.
Being a perfectionist I except this as good.
A pendulum will never be truly accurate
While the earth spins with a wobble.

 

[R. Croswell]

Peter, I honestly think that perhaps you are over-thinking this issue (driving weight vs the pendulum period) and attempting to arrive at a simple equation to prove what has been known for years without considering all the variables or relevance to the real world. We are not talking about a precision regulator here. This is a clock with a wood frame, subject to the effects of moisture, and a simple strip pallet escapement. While all escapements have been shown to be affected to some extent by changes in driving force, the recoil strip pallet escapement is the one that is most affected.

I am a bit confused about just what you are trying to accomplish. Are you trying to determine if the variation in rate over the weight drop period is "normal" (it probably is for this clock), or what causes the variation, or how to "fix" or reduce the variation? Or are you concerned about the variation in amplitude and rate between each cycle of the pendulum and if and how that relates to the overall time keeping issues with this clock?

Graham said, "Attempting to do detailed analysis on a movement that's been sat in a box for years and not cleaned or lubricated will be a largely fruitless exercise", and I agree. You said, "Before it went in the box, I dis-assembled it and cleaned the gears (with water then turps).... I didn't attempt to wash the bearings. They're brass bushes in a wood chassis and I wasn't sure how to wash them without damaging the wood so I cleaned them with a dry toothbrush. Re-oiled the bearings with a very light mineral oil. Didn't oil the teeth (but I might have run a soft pencil over the escapement teeth) The strip pallet needed slight adjustment to make good contact with the escape wheel.". At the beginning of this thread you said, "I don't have knowledge about 'adjusting pallets' and have only done simple 'while assembled' clean & oil' the pivots, and getting an even and steady 'beat".

I'm confused. Did you or did you not disassemble this movement for cleaning? Regardless, cleaning bushings with a dry brush is ineffective and cleaning with water and turps is not something I would recommend. You say you have no knowledge about adjusting pallets, but adjusted them for "good contact" with the escape wheel. How do you define good contact? Please don't take this as being critical, but before you can do any meaningful assessment of the movements operation it must be clean and properly adjusted. This recoil strip pallet escapement should have equal lift angles at both pallets, equal drops off of both pallets, and sufficient lock for good pendulum amplitude, obvious overswing (recoil), and be in beat. Escape wheel teeth get a tiny bit of oil. Never heard of using a soft pencil. I would encourage you to study up on proper clock repair methods so that you can be sure your clocks are performing at their best.

It runs reliably. No one has yet answered the question: is that amount of variance between swings normal? How many minutes a day should it gain/lose? How will the gain/loss per day vary over a year?

The amount of variance depends on the type of clock, for this one, it is probably what is to be expected. You may be able to reduce it a bit if the movement were properly cleaned and lubricated and the escapement precisely adjusted. It sounds like you weight is proper based on the minimum to keep it running. The clock should be adjusted so that it gains or looses ZERO minutes over the run time provided by a full drop of the weight. It may run a bit faster or slower during the week (assuming an 8-day clock) but there should be net zero error over the week.

How does stability depend on overswing? I put a simple "timegrapher" on the pendulum. The period varies by up to 2% from one swing to the next. That seems bad to me. Is that what you meant by stability?

No, you are talking about two different things. If there is zero overswing the escape wheel drops a tooth exactly when the pendulum is motionless and about to reverse direction. Even the slightest loss of power, vibration, or air current will cause it to stop - it will be very unstable and hard to keep running. The overswing is a safety margin that ensures that the clock will keep running when faced with environmental changes or variations in power for whatever reason. There is no rule, but if the overswing is visually obvious it should be OK.

I would take the variations in period from one pendulum swing to the next with as they say, a grain of salt. There are a zillion things that can cause minute variations plus there is surely a +/- x precision error in your test method. It isn't unusual to fine the gears in mass produced clocks to be slightly out of round, and the escape wheel teeth, even when new had an acceptable manufacturers tolerance, but after 100 years of pounding by the pallets (and occasional abuse by persons unknown) it is likely that they are no longer perfectly spaced and uniform. Uneven pivot wear, or slightly bent pivots due to careless assembly / disassembly can cause variations if friction and power to the escapement.

One thing that has not been mentioned is the precision or units of resolution of the clock. The smallest unit of time for this clock is one minute. It cannot accurately report any smaller unit of time. The clock is designed to have a certain number of beats per minute to keep time. It really does not matter at all what the variations are between each pendulum swing as long as it has the correct total number of beats each minute. If this were a precision clock where 1/10 or 1/100 second accuracy was important if would have been made much differently and would not have pull up weights etc.

RC






The strip pallet needed slight adjustment to make good contact with the escape wheel.

 

[peterbalch]

Those are all excellent questions. Thank you.

"Before it went in the box, I dis-assembled it and cleaned the gears (with water then turps).... "

At the beginning of this thread you said, "I don't have knowledge about 'adjusting pallets' and have only done simple 'while assembled' clean & oil' the pivots, and getting an even and steady 'beat".

It was JayKosta who said "I don't have knowledge about 'adjusting pallets'".

It was me who "dis-assembled it and cleaned the gears".

cleaning with water and turps is not something I would recommend.

Apparently turps was normal at one time - I'd probably read an old book. Gasoline was popular too. I cleaned it with water first because the clock had sat in my parents' living room for 40 years and they were both heavy smokers. Everything had a fine layer of brown tar. Smoker's tar dissolves surprisingly well in soapy water so I washed and dried the gears in soapy water first. Then in turps.

I would take the variations in period from one pendulum swing to the next with as they say, a grain of salt. ...

It really does not matter at all what the variations are between each pendulum swing as long as it has the correct total number of beats each minute.

If I remember my first-year statistics course: if the standard deviation of a single cycle is 0.01sec, what is the standard deviation over a day? I think the answer is 3 seconds. So in that sense, you're right - it averages out.

However I was interested in building a regulator. 99% of the regulators people build drive the pendulum with an electromagnet. If the standard deviation is 10%, it's not possible to do so by simply sending a regular pulse. Or rather, it's not possible given the mechanical restrictions I has set myself without using a huge current which means it can't be battery operated.

I am a bit confused about just what you are trying to accomplish. Are you trying to determine if the variation in rate over the weight drop period is "normal" (it probably is for this clock), or what causes the variation, or how to "fix" or reduce the variation?

That's the big question (or, at least, here's a big answer).

Yes. I'm trying to determine if that variation is "normal". But more importantly is it "normal" for this kind of clock? And what other kinds of clocks it it "normal" for? That's why I came to the forum and asked the question.

I think that by reading the theory and writing some simulations I have convinced myself that is "normal" for clocks with a recoil escapement. But there's a big difference between theory and practice. This forum is full of people with decades of experience. I expected that it would be "well known" what difference the weight made and which clocks it applies to.

Sadly it doesn't seem to be. Probably because as someone said: 'the weight is what it is' or 'you just adjust the pendulum to compensate for changes'.

To me, a clock has just two functions: To keep time and to look good. I want the clock to be as accurate and hassle free as a quartz clock I can buy for pennies. That's what I'm trying to achieve.

I have already built a simple autowinder that should run for a year on AA cells. I believe it to be a brand new design of autowinder and I'm quite pleased with it. It's a wooden box with a chain wheel, electric motor, a battery and a tilt switch. This animation sums it up:

It's reliable and inconspicuous and requires no changes to the clock. I think every wag-on-the-wall or long-case clock should have one.

Then it occurred to me that the autowinder could also act as a regulator so the clock was as accurate as a quartz clock. Once again I think this is a brand new idea.

The winder box turns back and forth by about 20deg. As the winder operates, the weight moves from side to side. Sometimes it's under the taut chain and sometimes it's slightly nearer the loose chain. A 20deg rotation of the winder box changes the torque seen by the main wheel by up to 20%.

Therefore I could put some electronics inside the box that changed the range over which the weight moved. To make the clock go fast, the weight could go back and forth between 0deg and 10deg. To make the clock go slow, the weight could go back and forth between 10deg and 20deg.

The electronics could adjust the range so that, over a day, month or year, the clock kept good time - never being wrong by more that a minute. (The details of electronics don't matter here but my first thought was that the box would contain a processor, an accelerometer and a stepper motor. I now think there are better ways to do it.)

How does the autowinder+regulator know whether the clock is running fast or slow? Easy: it knows how far it has turned its chain wheel.

I think that would be pretty impressive: a magic box that sits at the bottom of the chain and both winds the clock and keeps it accurate.

If I built it and it worked then I would publish it. Other people would want to copy it and the first question they would ask is "will it work with my clock?". I want to be able to say "yes it will if your clock is like this". Whatever this is.

Hence I came here and asked the question: what kind of clocks it it "normal" for?

Peter

 

[Raymond101]

@ Peter,
This is the best rabbit hole ever .
After your last post I realized how you are thinking. But your previous post didn't actually say what exactly you had or hadn't made . Or I missed something.
The auto winder is cool .
But if you want to swing the pendulum with a N S N magnetic pulse. As you seem to want quartz quality. Then you can do away with the chain drive.
If the pendulum is swung electronicly the clock won't need to have power train ie from the 550 g weight.
Swinging a pendulum back a forth this movement will rotate the escapement wheel thus turning the train .
A quartz clock for example doesn't have a static spring to wind .
So if you just swing your pendulum it should just run fine . . In the correct direction to the escapement teeth.
.
What I think what you are doing is a good idea. But you only require one power source. Go for Swinging the pendulum and remove chain weight or a dumpy weight of just a few grams ie for appearance.
BTW Alice sent her regards.

 

[R. Croswell]

@ Peter,
This is the best rabbit hole ever .
After your last post I realized how you are thinking. But your previous post didn't actually say what exactly you had or hadn't made . Or I missed something.
The auto winder is cool .
But if you want to swing the pendulum with a N S N magnetic pulse. As you seem to want quartz quality. Then you can do away with the chain drive.
If the pendulum is swung electronicly the clock won't need to have power train ie from the 550 g weight.
Swinging a pendulum back a forth this movement will rotate the escapement wheel thus turning the train .
A quartz clock for example doesn't have a static spring to wind .
So if you just swing your pendulum it should just run fine . . In the correct direction to the escapement teeth.
.
What I think what you are doing is a good idea. But you only require one power source. Go for Swinging the pendulum and remove chain weight or a dumpy weight of just a few grams ie for appearance.
BTW Alice sent her regards.

Only one problem, just swinging the pendulum with zero driving force from the going train will not cause the escape wheel (or the going train and hands) to advance.

 

[Raymond101]

You will have to change the pallet. Look up how a mechanical quartz clock escapement is setup .
It does work. Your clock has a wooden wheels so there is a high friction ratio.
You use a magnetic pulse or induction coil depending on how much power is required to swing the pendulum..
Also placing copper plates at the end of the desired amplitude the magnet will slow down so the beat can be turned to micro seconds but your wood work would have far to greater gear mess resistance to work well . If at all .

 

[JayKosta]

...
Then it occurred to me that the autowinder could also act as a regulator so the clock was as accurate as a quartz clock. Once again I think this is a brand new idea.

The winder box turns back and forth by about 20deg. As the winder operates, the weight moves from side to side. Sometimes it's under the taut chain and sometimes it's slightly nearer the loose chain. A 20deg rotation of the winder box changes the torque seen by the main wheel by up to 20%.

Therefore I could put some electronics inside the box that changed the range over which the weight moved. To make the clock go fast, the weight could go back and forth between 0deg and 10deg. To make the clock go slow, the weight could go back and forth between 10deg and 20deg. ...

----------------------------------------------------
As I understand your idea, the regulator would never actually stop the pendulum, only attempt to make small rate changes to keep in sync with the processor time. This might be possible if the pendulum length has already been adjusted to be very accurate - so that only 'extremely small' rate changes are needed. Unless environmental changes such as temperature and humidity are kept constant, I doubt the regulator would be able to make the adjustments in a short enough 'adjustment period' so that the time shown on the clock would not be noticeably off.
I think that 'weight compensation' would not be able to keep-up with whatever factors are causing a rate change quickly enough for the clock to display correct time.