Building Bernard Tekippe’s Precision Regulator

[Michael GilliganParticipant @michaelgilligan61133]

A difficult choice, Chris … The escapement sounds good to me, and the expedient option is to not start tinkering.

But the ‘vibration tester’ in me is wary about that big overhang

Will vibrations in that arbor get fed into the ‘scape wheel, and if so what might be the effect ?

Frankly, I don’t know

MichaelG.

[gerry maddenParticipant @gerrymadden53711]

Hi Chris, I didn’t want to say anything earlier as it’s so difficult to make accurate judgements from videos, but when I first watched your second-to-last video, I was thinking to myself at 18:10 that the scape-wheel arbor was not looking as free I thought it should be. After being given a nudge, the wheel seems to stop quite suddenly. That’s often a sign of excess bearing friction.

Three things could be causing this. One is misalignment of the two support positions, (as previously discussed). The other is your housing fits are too tight and/or out of round. Finally you might even have some bearing-to-bearing preload.  **

Unfortunately without a detailed engineering drawing showing positions of abutments, dimensions etc, it’s difficult to pinpoint your issue/s, so my feeling is one really has to sort this out by trial and error investigations.

I would probably start by looking at those housing fits which I’m really not happy about. If your housing holes are just drilled, they WILL have significant shape errors. If the rings are ‘press fits’ in these drilled holes, these shape-errors will easily be transferred through the rings into the raceway/ball-set where they will cause all manner of nasty effects. So you really need to aim for a loose-fits on these outer rings. One should really talk about machining to specific diameters, but since that’s not really practical in this environment, the next best thing could be to use fine emery paper to gradually ease the hole. You should aim to get to a point where the outer-ring will just slide through it.  A poorly shaped loose-fit hole is far kinder to the bearings than a tight poorly-shaped one.

If you are currently using the tight housing fits to locate/secure the shaft assemblies axially, you may now need to add a small abutment plate or two in the appropriate positions to keep everything in place. Try to arrange it so that just ONE of the two bearings on the shaft, locates it axially. This will ensure you don’t end up with the two bearings fighting each other.

With regard to your reduced-span support modification, from the point of view of loading, you need not have any concerns. The load capacity of these bearings is way in excess of the forces that a clock mechanism can impose. I would be more concerned about the alignment. Reducing the span between the bearings is likely to increase the misalignment angle which is critical from a friction point of view. However if you take the steps on the housings outlined above, misalignment issues will be mitigated.

Gerry

** just thought there may be a forth, grease. Were these ‘degreased’ bearing that you are using? Regardless, I would still focus on the points above as these aspects must correct if we are ever to get the optimum low-friction performance from these rolling element bearings.

 

 

 

 

[steve7750Participant @steve7750]

These are all excellent points, Gerry.  I’d mentioned to Chris on his other thread that the preference would be to keep the front bearing between the escape wheel and hand to spread the (admittedly tiny) load.  The train is really sensitive to the smallest resistance especially at the escape arbor.  It can be stalled by the merest hint of friction.

Our plate holes were all drilled undersize with end mills and then reamed, to try and keep everything where it was designed to be.

Ours has the hybrid ceramic bearings which had to be thoroughly degreased in acetone, or they were just too sticky.  Time will tell if there is a corrosion issue, but being stainless we should get a decent lifetime to test things out.

[Michael Gilligan]

On Michael Gilligan Said:

A difficult choice, Chris … The escapement sounds good to me, and the expedient option is to not start tinkering.

But the ‘vibration tester’ in me is wary about that big overhang

Will vibrations in that arbor get fed into the ‘scape wheel, and if so what might be the effect ?

Frankly, I don’t know

MichaelG.

Hi Michael

Vibrations I certainly have. The clock seems to be running well but the pendulum seems have a strange vibration or judder!

It is very strange.

On gerry madden Said:

Hi Chris, I didn’t want to say anything earlier as it’s so difficult to make accurate judgements from videos, but when I first watched your second-to-last video, I was thinking to myself at 18:10 that the scape-wheel arbor was not looking as free I thought it should be. After being given a nudge, the wheel seems to stop quite suddenly. That’s often a sign of excess bearing friction.

Three things could be causing this. One is misalignment of the two support positions, (as previously discussed). The other is your housing fits are too tight and/or out of round. Finally you might even have some bearing-to-bearing preload.  **

Unfortunately without a detailed engineering drawing showing positions of abutments, dimensions etc, it’s difficult to pinpoint your issue/s, so my feeling is one really has to sort this out by trial and error investigations.

I would probably start by looking at those housing fits which I’m really not happy about. If your housing holes are just drilled, they WILL have significant shape errors. If the rings are ‘press fits’ in these drilled holes, these shape-errors will easily be transferred through the rings into the raceway/ball-set where they will cause all manner of nasty effects. So you really need to aim for a loose-fits on these outer rings. One should really talk about machining to specific diameters, but since that’s not really practical in this environment, the next best thing could be to use fine emery paper to gradually ease the hole. You should aim to get to a point where the outer-ring will just slide through it.  A poorly shaped loose-fit hole is far kinder to the bearings than a tight poorly-shaped one.

If you are currently using the tight housing fits to locate/secure the shaft assemblies axially, you may now need to add a small abutment plate or two in the appropriate positions to keep everything in place. Try to arrange it so that just ONE of the two bearings on the shaft, locates it axially. This will ensure you don’t end up with the two bearings fighting each other.

With regard to your reduced-span support modification, from the point of view of loading, you need not have any concerns. The load capacity of these bearings is way in excess of the forces that a clock mechanism can impose. I would be more concerned about the alignment. Reducing the span between the bearings is likely to increase the misalignment angle which is critical from a friction point of view. However if you take the steps on the housings outlined above, misalignment issues will be mitigated.

Gerry

** just thought there may be a forth, grease. Were these ‘degreased’ bearing that you are using? Regardless, I would still focus on the points above as these aspects must correct if we are ever to get the optimum low-friction performance from these rolling element bearings.

 

 

 

 

Thanks for the great response Gerry. I have gone for a tighter bearings on the ID as it was discussed that the arbor looked too lose on the inner ring. The holes were drilled and reamed out and the bearings a great fit – just clear to hold them but with almost no pressure on the OD of the bearing.

If it stops again, the next plan is to remove all the bearings and clean them. I admit, I posted a while ago about degreasing bearings and then with all the issues I had ordering more, I’m worried one has slipped in that is still greased!

I just need to slow it down a bit now. No point going any further until I have the clock running.

Chris

[steve7750Participant @steve7750]

Hi Chris,

Bernie’s design of minimal cross section does make for a very whippy pendulum.  If it wasn’t doing this when you first started it (which can happen if the Bob is released too quickly), what you’re seeing is the result of a “misfire” when you weren’t there.  The escape wheel didn’t cleanly drop to the pallet, but hung up on the way.  The other incoming pallet kissed the tip of a tooth and sent a shudder down the whole assembly which will reverberate for ages.  The escape wheel freed up and the running continued.

Once you have a Microset on it, you’ll be able to catch these things, but these events certainly don’t do any favours to the suspension spring, escape wheel teeth or pallets.  You need to try that run down test I mentioned earlier, to make sure the train is totally free.

[Chris Raynerd 2Participant @chrisraynerd2]

Thanks Steve

Interesting points made. You have mentioned the run down test a couple of times and apologies for not replying. It is always one of the first things I do when I have depthed the train. It seems to be lovely and free! There may be a few spots where it varied in speed a little, but with such small weights applied, it seemed incredibly free!

 

[Michael GilliganParticipant @michaelgilligan61133]

I would say that the wooden back-board is very likely contributing to the disruption of the pendulum … a good solid wall would be a better mounting point.

MichaelG.

[steve7750Participant @steve7750]

I did find an old video of ours after a similar incident.

https://youtube.com/shorts/tza2mUYaYfY?feature=share

[Chris Raynerd 2Participant @chrisraynerd2]

Evening all

Well it’s been over a week with no updates so thought I’d show some progress. I managed to fix the issue with power which was the alignment of the bridge. We knew that anyway as the bearing in the front plate allowed it to run fine. I’ve now realigned the bridge and removed the front frame bearing and put it back on the bridge where it should be. Thankfully this has also cured the horrid juddering in the pendulum!

It seems to be running ok now and will run for days without stopping. Does the deadbeat look to be running ok?

Thanks

Chris

[Michael GilliganParticipant @michaelgilligan61133]

Looking good, Chris

My tired old eyes couldn’t see any recoil.

MichaelG.

[Chris Raynerd 2Participant @chrisraynerd2]

Thanks for the reply Michael and I’m pleased you have said that. I posted elsewhere and someone mentioned they could possibly see some recoil on the exit pallet but I can’t!

Chris

[SillyOldDufferModerator @sillyoldduffer]

Enjoyed the video, thanks!  Looking good.   My interest in Tekippe is the claimed accuracy, so I’d prioritise measuring that over cosmetics like dials etc.

My understanding of Tepkippe’s design is it claims high-performance from a straightforward build but accuracy is achieved by tweaking in response to high-end measurements taken with a Microset or similar.  Not just by building to the plan! What others have said on the web about adjustment is thin stuff, so I’m hoping you’ll fill in the details!

Any idea how accurate it is so far? Few seconds per day, 1 second per week, or better than a digital wrist watch, a few seconds per month?  (The goal is to keep time better than the wrist watch.)

If you haven’t got a copy already, I recommend RAWLINGS A,L The Science of Clocks & Watches 3rd Edition BHI 1993.   The 3rd edition was enhanced by Dr Philip Woodward, who explains the need to measure the clock’s stability rather than rate.  Stability is the variation of rate over time, not just the simple rate over a period.

Excellent work!

Dave

 

 

 

 

[John HaineParticipant @johnhaine32865]

Perhaps some close-up video of just the seconds hand and / or scape wheel would be useful to see if there is any recoil?  Step through frame by frame to see what’s happening in slow-mo.  Good if your camera can run faster.  We did some video of the pendulum of one of the RAS Regulator replicas at 100 fps.  I can also recommend some open-source software called “Tracker” for video analysis.

The principle behind Tekippe’s design is to have a controlled amount of recoil to introduce some extra escapement deviation to speed the clock up as the amplitude increases. This is balanced against circular deviation which slows it down.  If you plot rate against amplitude you get a “hill” curve and the idea is to adjust things so the working amplitude is at the summit of the hill.  Then small changes in torque from the train which change the amplitude always reduce the rate so the system is first-order insensitive to torque variations due for example to wheel irregularities.  Tekippe’s method of adjustment was to tweak the pallet angle and hence the recoil but this is rather tedious – I think a better method would be to set a small fixed recoil and adjust the drive weight to get to the top of the hill.  Reminiscent of adding weights to trim the pendulum.

I believe that he achieved something like 1s over 75 days so within a whisker of Clock B, but since the clock is not pressure compensated its longer term performance would be significantly worse.  A barometric compensator could be added of course, or the “barycentric” method used in Clock B adopted perhaps.

[Author]

Posts