Building Bernard Tekippe’s Precision Regulator

[Andrew CrowParticipant @andrewcrow91475]

<p style=”text-align: left;”>Nice work Chris, you do make it look easy, but I’m sure it isn’t.</p>
Looking forward to the next video.

Andy.

[Chris Raynerd 2Participant @chrisraynerd2]

Thanks Andrew, much appreciated.

There has been some discussion on bearings in this thread and whilst I’ve been asking the question, I’ve not got involved as I don’t know enough to comment. Whilst the traditionalists are strongly against them and seem to argue that a brass bushing or jewel is the best way in clocks, I can’t help but go off the evidence. There have been a number of links posted in this thread and then this simple test looks to be undeniable – even out performing jewels. I was all set to spend my money on jewels and whilst they certainly look the part, if they are out performed by cheaper bearings, I don’t think they can be ignored….

I’m going to go ahead just like in Bernie’s plans and use bearings!

Chris

 

[Michael GilliganParticipant @michaelgilligan61133]

Thanks for posting that, Chris

… it still leaves a lot of questions unanswered [or rather ‘variables not fully explored’] but it’s good to see some raw data from a simple experiment.

MichaelG.

[Chris Raynerd 2]

On Chris Raynerd 2 Said:

….

There has been some discussion on bearings in this thread and whilst … the traditionalists are strongly against them and seem to argue that a brass bushing or jewel is the best way in clocks, I can’t help but go off the evidence. …

… 

Going with the evidence is always best, but slow down!

The video documents an unusually good experiment.  Not because it delivers conclusive proof, but because it’s well described.  The detail lets us test the apparatus, method, results, and conclusions. Always necessary when seeking proof, because even small mistakes in a really good experiment can mislead.

There are several issues in the experiment, not unusual and it’s honest:

1. The discussion relates to the use of bearings in a clock’s counter, that is the escapement and gear-train that turns the hands.  This experiment tests bearing friction with a pendulum, not the same thing.
2. The method is dubious. A pendulum is a somewhat heavy swinging object, not typical of what happens in a gear train.  For example, as tested with a pendulum, the experimental bearings never make a complete rotation.  In a gear train the same bearings rotate many times.
3. Best results in the experiment are achieved with de-lubricated steel bearings.  Not a surprise, nor is it the answer! De-lubricating reduces friction because lube is sticky as well as slippery.  Has to be, otherwise it falls off!  Problem is that removing lubricant causes rapid wear, so the improvement gradually disappears as the bearings grind.   Not good in a clock.  Beware of ‘improvements’ that look good on the bench; they may not stand the test of time.  Note that jewelled pivots can run for many decades.
4. The experiment tests many different types of bearing, which is good.  But all these bearing types exist to satisfy particular use cases, probably not swinging pendula!   I hope no-one believes the experiment proves that steel bearings are best for everything from big-ends to chronometers!   More to choosing bearings than measuring friction as demonstrated in the video.  If there was one type of bearing that did everything, all the others would disappear.
5. Bearings are rarely used to support the pendulum in a high-precision clock.   Far easier to get good results by suspending the rod from a leaf spring.  Springs store and return energy to the bob, a simple way of ensuring reliable long-term low friction.   Lower than spring friction has been temporarily achieved by swinging the pendulum on a razor edge resting on agate, but this type is vulnerable to dirt and wear.  Over time the razor goes blunt, dirt gets in, and the agate grooves. The initial high-performance doesn’t last.  Hero to zero after a year or two!
6. Though the experiment keeps it simple by not measuring ‘Q’, it could have done without much extra effort.  As a comparison I suggest the experiment should have tested a spring suspension too, as the control.  Judging by the table (repeated below), the Q of the bush-type bearings are very low, and the best result feels poor too.  (Haven’t attempted the maths:  300 swings to decay 6° ain’t good though.  A home-built pendulum should aim for Q of 10,000 or higher.)   This comparison would have highlighted frictional orders of magnitude.

The results:

To be clear, I’m sure bearings will work in a clock.   The problem is they may not be better than the alternatives, particularly over a long time.  I’m all for challenging traditional methods and keeping up-to-date, but pivots running in jewels work extremely well in clocks.  They’re light and low friction.   Main disadvantage is they can’t take heavy loads or transmit much power.  Other bearings are better for power applications, but the cost is higher friction and inertia.

Gut feel, the choice is to do with size and weight, with a point at which advantage flips from pivots to bearings.  Daft to put conventional bearings into a wristwatch, equally unwise fit a tower clock with jewelled pivots.  Bernard Tekippe’s clock is somewhere in the middl; is the gear-train power hungry or light-weight?   Not knowing suggests further experiments:  how much power does it take to drive a bearing fitted Tekippe compared with a pivot and jewelled Tekippe?

Bearing inertia is certainly a problem.  In addition to losing energy due to friction, work is done moving mass.  As bearings are much heavier than pivots, this can’t be ignored.  Clockmakers go to considerable trouble to reduce the weight of moving parts, all the big gears being thin and much pierced. Inertia matters.

By all means build with bearings, just don’t do so on the basis of this experiment.  Though ten-out-of-ten interesting and informative, it’s not conclusive.   Bearings in a Tekippe might work well or they might not.  Measurement not opinion.  Please report back – might include a confession that the clock worked well for a year before becoming sticky and then seizing up.   Many beautiful clocks don’t work particularly well!

Consider:

• How much power is needed to run the clock?
• How accurately does it keep time?
• How reliable is it?

All three get progressively more difficult to improve.    My first clock was Meccano throughout.   No pivots or bearings. Standard brass involute gears,  pendulum swinging on an axle, escapement a sprocket wheel. Mechanically clunky and the power needed was enormous.   Accuracy about 10 seconds over 15 minutes, which is how long it ran for reliably.  High maintenance and extremely fussy.     Since then I’ve learned that although working clocks can be extremely crude, cuckoo, they don’t work well and there are multiple ways they might be improved.  Unfortunately many improvements turn out to be neutral or make things worse.  As much of the experimentation needed was done in the past, it saves time to study the old books.   Maybe the old boys missed a trick, but in the 60 years mechanical clocks could have included modern bearings, few did.    My guess is bearings weren’t advantageous.

Dave

 

[gerry maddenParticipant @gerrymadden53711]

This might help you define one boundary, Dave

https://waitingtrain.blogspot.com/2008/08/

Have a look at the text on the knurled disc on the pendulum support 🙂

It’s there for the avoidance of Brinelling and such.

Very interesting Michael. I’ve not seen such a mechanism before on a clock and often wondered why not. But it seems they do!

Probably very helpful for minimising the risk of false-brinelling though is to use oil, not grease. When I get around to my version of the popular tower clock, I plan to have an oil cap (with sprung lid) on the housing so that I can occasionally put in a drop or two of modern low viscosity engine oil.

Oil will also generally create less friction than grease too, which is of course welcome in a pendulum support.

Gerry

[John HaineParticipant @johnhaine32865]

Sheer conservatism.  Erwin Sattler use ball bearings in their clocks. Please also read Rex Swenson’s publications I referenced above.

Rex recommends washing grease out in several changes of white spirit then blowing dry. Do not be tempted to use an ultrasonic cleaner, it spalls the bearings.

[Michael GilliganParticipant @michaelgilligan61133]

Gerry

Although the ‘Waiting Train’ uses a hefty bearing for its pendulum support … it’s worth noting that this pendulum is more of an energy store than a timekeeper.

Think of it as being rather like a motor/flywheel

MichaelG.

[gerry maddenParticipant @gerrymadden53711]

On the general discussion of rolling element bearings versus plain type, unquestionably rolling type are technically superior in the short term. They can be in the long term too, provided they are properly lubricated, kept clean and aren’t allowed to corrode. Of course plain type also have their long-term issues too, if you don’t look after them in the same way.

There are only really two things against rolling type. Cost and complexity. A small ball bearing is always going to cost several times the cost of a plain bush. There is additional cost in building them into the structure and this is where the complexity come in.

Most clock makers like to work to ‘thous’. You have to think microns if you want to make rolling bearings function properly. There are lots of pitfalls. Corner radii on shaft abutments for example. This has to be smaller than the tiny one on the bearing itself. If it isn’t when you press the bearing up against the shoulder it will sit cockeyed. Most assemblers won’t even notice this but the bearing will, and show it by increasing friction. Shaft fits are critical. A few microns the wrong way will spoil the ship. Take the shields out to help remove the grease and you are very likely to contaminate the things with dust particles. These particles might be small to you but they are like rocks in a car gearbox to a tiny rolling bearing. You won’t feel or hear the notchiness, but it will be there. Shaft assembly axial play, housing shape errors, the list could go on for hours.

The bottom line is simply that if you want to glean the advantages of using rolling bearings in a clock, you need to put in 20x the normal level of attention to detail, and plan the design, machine the parts, measure them with precision and assembly them with great care.

Gerry

 

[John Haine]

On John Haine Said:

[…] Erwin Sattler use ball bearings in their clocks. […]

Mmm … as and where considered appropriate:

https://erwinsattler.com/en/portfolio/excelsia/

Eight ball and fifteen jewel in that one ^^^

Surely that keeps the debate open !?!

MichaelG.

.

Edit: __ credibility check ?

https://erwinsattlerclocks.com/products/pulley-with-ball-bearings-copy

[gerry maddenParticipant @gerrymadden53711]

Nice products, but no, I don’t think this keeps the (technical) debate is open. 🙂

Jewels are technically preferable to metal bushes. They have generally slightly lower coefficients of friction and wear less. However, no jewelled bearing can match rolling bearings for friction and rolling bearings won’t wear either.

So I suspect the reason Sattler have used jewels in places is for their well-known marketing appeal and, to avoid some cost and complexity.

With regard to escapements, there is some folklore suggesting that the oscillating nature of the shaft is bad for rolling element bearings. The background theory is basically correct, rolling bearings in industrial applications do not appreciate repetitive movements, and the smaller the amplitude worse it can be for them. Having said that, light loads, oil lubrication, indoor levels of humidity, lack of external vibration are all things in a clocks favour. I will certainly use them here in my small tower clock when I get around to it. 🙂

Gerry

 

[Michael GilliganParticipant @michaelgilligan61133]

Horses for Courses, Gerry

Sattler uses/offers ball bearings and jewels at appropriate locations.

https://erwinsattlerclocks.com/products/rubies-for-the-escapement-lever-bearings

Your small tower clock is more likely to demonstrate the advantages of the balls.

MichaelG.

[SillyOldDuffer]

On SillyOldDuffer Said

Bearing inertia is certainly a problem.  In addition to losing energy due to friction, work is done moving mass.  As bearings are much heavier than pivots, this can’t be ignored.  Clockmakers go to considerable trouble to reduce the weight of moving parts, all the big gears being thin and much pierced. Inertia matters.

 

I wouldn’t be so certain.  The moment of inertia is proportional to the mass times the square of the radius.  Taking as an example the escape wheel arbor on my regulator, which is likely to be the most critical, It is mounted on two 2 mm ball bearings with OD of 5 mm.  the escape wheel rim (ignoring the teeth) is about 30 mm diameter so r squared is about 36 now the weight of the rim is certainly more than the combined weight of the bearings so the moment of inertia of the bearings is not really significant when compared with that of the wheel and all the other moving parts.

One thing that might be causing a problem in hindsight is that I used ball bearings for the pallet arbor and ball bearings might have a shorter life with the reciprocating motion rather than full rotation.  After ten years of running the timekeeping is becoming more variable and needs investigating.  Of course it could just be the strange weather we are having as the clock sits in sunshine for part of the day>

Russell

[bernard towersParticipant @bernardtowers37738]

you could be right about the pallet arbor, I have heard it discussed on a open day at upton hall that dirt can build up in the space not used by the balls causing the bearing to come to an abrupt halt!

[John HaineParticipant @johnhaine32865]

There’s no pallet arbor on the Tekippe clock, they are mounted on the pendulum which has a spring suspension.

[John Haine]

On John Haine Said:

There’s no pallet arbor on the Tekippe clock, they are mounted on the pendulum which has a spring suspension.

Yes, of course John.  Apologies for going off topic.

Does anyone know of a source for drawings (or CAD flies) for the Tekeppe clock?

Russell

[Russell Eberhardt]

On Russell Eberhardt Said:

[…] Does anyone know of a source for drawings (or CAD flies) for the Tekeppe clock?

It looks like John probably does:

https://groups.io/g/Horological-Science-Newsletter/topic/clock_b_vs_tekippe_s_clock/112278009

MichaelG.

[Chris Raynerd 2Participant @chrisraynerd2]

The 3D CAD drawings have been produced by Steve on the NAWCC forum. I’m just out but I will send you his detail via pm when I get back after I check that is ok with him.

I have also collated a lot of info on my site and will update this with additional info this evening:

Tekippe Regulator Clock Build

 

[John Haine]

On John Haine Said:

 

Rex recommends washing grease out in several changes of white spirit then blowing dry. Do not be tempted to use an ultrasonic cleaner, it spalls the bearings.

We’ve probably done the topic to death but it’s actually very very difficult to remove all grease from pre-grease bearing without mechanical assistance, particularly if they are fitted with metal cages. Just soaking in solvents will soften the grease (temporarily) but some agitation is necessary to remove it, especially where pressed metal cages are fitted. Grease trapped between the cage and the balls is not easy to dislodge. When possible therefore use bearings which are fitted with snap-in plastic cages as these can be snapped out for a proper clean with a fine brush. It goes without saying that the cleaning fluids and implements must be super clean. If you hear rattling when you spin it dry, you weren’t doing it right!

Gerry

 

[Russell Eberhardt]

On Russell Eberhardt Said:

On John Haine Said:

There’s no pallet arbor on the Tekippe clock, they are mounted on the pendulum which has a spring suspension.

Yes, of course John.  Apologies for going off topic.

Does anyone know of a source for drawings (or CAD flies) for the Tekeppe clock?

Russell

I’ve just send you an email with his contact details. I’m sure he will be happy to share the plans with you.

please also check out my website for the sketches taken directly from one of Bernie’s clocks as well as the article written on this regulator.

Tekippe Regulator Clock Build

I’ll add more details and files soon.

Chris

[Author]

Posts