Building Bernard Tekippe’s Precision Regulator

[Chris Raynerd 2Participant @chrisraynerd2]

Hi Michael

Definitely not off topic. That is not how I was going to use the shellac – and now I am!

Thank you very much.

Chris

 

[Chris Raynerd 2Participant @chrisraynerd2]

Evening all

I am just sat here trying to lay out the escapement and I want to do it based on some instructions provided by Bernie Tekippe below:

So I have attempted to follow these instructions, drawn my escape wheel, marked 70 deg and then drawn a tangent back up to the horizontal. If I’m reading it correctly then the height back from the centre of the escape wheel to this point, which is 52mm, should be the distance to the centre of the suspension spring. I should have considered this before building everything as my current distance is 50mm, but not a problem, I can shave 2mm of the top spring support which will raise the suspension spring. So hopefully this is an easy fix.

The bigger question is still understanding the layout and location of the actual pallets! It’s that last sentence. If I’m reading it correctly, I’ve taken the distance from the “suspension spring pivot point” to the mark on the escape wheel where it intersects the 70 deg line. I’ve then used this to form a new arc. I presume where this new arc dissects the rods of my pendulum support arms is the location of the pallets, but of course this is an arc and whilst I know dead beat pallets should be an arc, Bernie does do this. Therefore is the correct location for the slot on the pendulum supports a tangent to this line?

Any help would be much appreciated. My big question is if I’m slotting the pendulum supports, where and at what angle based on my drawings and the instructions. Thanks again everyone.

[John HaineParticipant @johnhaine32865]

Where does 70 degrees come from? The text says 75 degrees. Also the actual bend point is unlikely to be at the centre of the spring but higher up. If you have the mass of the bob and rods and the spring dimensions we have a program that can predict the bend point

[Chris Raynerd 2Participant @chrisraynerd2]

@john haine – yes the 70 deg was an oversight
– 75 deg you are correct.

Thank you for the offer to calculate the bend point. The mass of the rod and bobs is 1.2kg and the suspension spring dims are 0.05mm thick, 15mm wide and 3.5mm tall.

Presuming I drew 75 deg rather than 70, do you have any instructions as to the correct location of the slot for the pallets on the frame?

[Chris Raynerd 2Participant @chrisraynerd2]

@john haine – yes the 70 deg was an oversight
– 75 deg you are correct.

Thank you for the offer to calculate the bend point. The mass of the rod and bobs is 1.2kg and the suspension spring dims are 0.05mm thick, 15mm wide and 3.5mm tall. They are the dimensions of the exposed spring.

Presuming I drew 75 deg rather than 70, do you have any instructions as to the correct location of the slot for the pallets on the frame?

[John HaineParticipant @johnhaine32865]

To do the calculation we need the radius of gyration of the pendulum so the dimensions of the rods and the mass of the bob itself.  I can’t help with the pallets, I seem to recall that the builder over on the NAWCC MB built a location jig for them?  From his photos it looks like he made quite a large flat on each rod with a bridge over the top that clamps down the pallet.  Is that not shown on his 3D model?

[Chris Raynerd 2Participant @chrisraynerd2]

Evening John

Steve did an absolutely incredible job using micrometers in his jig! It was quite phenomenal but seems quite complex compared to the layout instructions Bernie provided. Steve, if I remember correctly, opted for a later design in which the pallets were not horizontally level in respect to the escape wheel. I’ve sent him an email and will see if he replies.

The simple wooden jig I posted, apparently worked fine to locate the pallets in a copy of the Tekippe clock built by another clockmaker. However rather than create a large pocket to adjust the pallets, Colin used his simple jig to cut an accurately located slot in the pendulum pillars and the only thing that needs adjusting is the depth of the pallets – the jig and therefore the slot takes care of their location and angle. The pallets are held in place by a simple clamp style arrangement that you described and I’m happy with this arrangement.

The mass of the bob and pendulum is 1.2kg, the rods are 6mm diameter and 51mm apart to centres.

Perhaps I just need to look again over the next few days to digest them!

Thanks as always for your help and advice.

Chris

[steve7750Participant @steve7750]

Hi Chris/ other thread followers,

Ok, let me answer a few questions and explain the background for some decisions.  When I first built the pendulum back in ’23 I didn’t have that single sheet document of Bernie’s, so I had to scale everything from the photograph below.  I knew the Invar rods were 1/4″, so worked from there.  The slots in the side arms have to be wider than the pallets to allow for the adjustment.  The odds of cutting it right the first time are very low.

The jig I built with the XY micrometers was not really for the initial setup, but were for adjusting the pallets in as controlled a manner as possible.  Not having Bernie’s golden screwdriver or intuition, there needed to be a way to capture the existing position of the pallet tip, and move it a known quantity.  The secret sauce of his design was the subtle tuning of recoil, and this gave a way to start knowing what changes were made, for better or worse.

The two brass rods protruding through the centre hole are each on a separate XY stage.  Each rod has a small vee machined down the side, which the pallet tip fits into. The rods have a generous hole through them so they can float a mm or so to snug against the tip.  The brass segment plates let you record the pallet angle, with the small holes at 0.5 degree increments for reference. The tips of said plates also rest in the same vee as the pallets. So, the pendulum frame is locked in place with the Al plates at the bottom, and the pallet positions “recorded”.  The pallet clamp plate can then be loosened, the X,Y or angle changed and then it’s re-tightened.  The pendulum goes back on the clock for further testing.

I’ve not been following this thread as well as I should have, but there were a few other questions earlier.

To make sure the holes on my plates lined up as well as possible, I had a jig clamped on the mill table with some bosses that the plate was pressed against before clamping them down.  This ensured that each plate was using the same datum.

Also there was a question about making the Titanium escape wheel.  I used the profile cutting method on a Sherline Accupro.  The blank was turned first and then dropped into a holder which doubled as an oil bath.

 

There’s probably other stuff I’ve forgotten to answer, but I’ll try to pay more attention to this thread.  I’m very pleased to see your progress Chris!

 

 

 

 

[Michael GilliganParticipant @michaelgilligan61133]

A very interesting and useful contribution, Steve

Thank You

MichaelG.

[John HaineParticipant @johnhaine32865]

Steve, what cutter dia, DOC and feedrate did you use please when making the scape wheel?

[BazyleParticipant @bazyle]

Are the slots in the side arms in order to bring the pallet ends close to centreline to avoid twist from the impulse? There seems to be room between the side arm and escape wheel to fit some sort of holder with screw adjustments to put the pallets on centre and avoid the slotting.

[John HaineParticipant @johnhaine32865]

I think I recall a photo showing that the pallet “blades” are offset from the “stalks” to bring them on centre.

[steve7750Participant @steve7750]

John,

The cutter was a 1mm and I was able to run it in a single pass at 2mm/min feed, 2500 rpm.

[steve7750Participant @steve7750]

Gents,

Yes the pallets are “flag” shaped.  The first is from Bernie’s #7, and then our interpretation before grinding the impulse face.

Indeed there is space for some form of adjustment mount, but I imagine the slot was used by him to maintain his principle of design simplicity.

 

[Chris Raynerd 2Participant @chrisraynerd2]

Thank you very much Steve, excellent info as always!

I have the clock all depthed now and it seems to be running nice and free.

I have spent a long time drawing out the escapement and with the help of the CAD model that David Robertson kindly did for me on the NAWCC thread, I have increased the height of the suspension spring to 61mm by shaving a few mm off the top suspension spring support hanger and also using a brass riser block under the pendulum hanger (long brass flat from which the pendulum hangs). This had given me the required 61mm and I also think my plans I have drawn, show me where to slot the pendulum rods for the pallets. Can I ask what angle you started with for first tests or where this info is shown? I’ve looked through the NAWCC article and can’t see any suggested initial angles!

Thanks in advance

Chris

[steve7750Participant @steve7750]

You’re welcome.

Glad your train is done.  Before the Pendulum goes on, you should be able to hang a small weight on the barrel and have the escape wheel spin slowly.  Any tight spot will show up immediately.   Took me a long time to free everything up, but your machining is probably better!

As for the pallet angle, I just scaled off the photo I showed earlier, but add here again. I just put a protractor on it.  I forget now what it actually was.

[steve7750]

On steve7750 Said:

 

On steve7750 Said:

John,

The cutter was a 1mm and I was able to run it in a single pass at 2mm/min feed, 2500 rpm.

Thanks Steve.  Very slow feed rate then, is this needed for titanium generally of is it because of the low spindle speed?  I like the way of fixturing the blank in an oil bath, did you have any trouble with chip build-up?  I have profiled 10-leaf pallets into the end of a piece of FC steel bar, then “releasing” them by parting off a slice.  IIRC I applied small amounts of cutting oil and blew the chips away with my usual blower.  Made 4 pinions, only broke one cutter!  Feedrate was 50mm/minute, 1mm slot drill, I think at 5000 rpm.

[SillyOldDufferModerator @sillyoldduffer]

I’m looking forward to the next stage of this fascinating project, not least because the clock’s claimed performance is about 5x better than Woodward thought possible.

The proof of the pudding is in the eating.  Next stage is adjusting the escapement to get high performance, which appears to be the secret.   Performance isn’t simply achieved by following the design and tweaking the rate knob.   Instead, the pallet shapes and positions are carefully adjusted to balance escapement and circular errors.  With the bob at dead centre, the intent is to apply just enough positive recoil to the pendulum.  Whether or not a setting applies too much or too little is revealed with a Microset or similar.  Only possible today with electronics, which quickly show timing changes.

Relatively quickly I suspect, not instant!  I think at least one complete revolution of the escape wheel is needed per adjustment, probably many more as optimum is approached.  Bernard Tipkippe doesn’t say how many attempts he needed to get spot on.

For all it’s virtues, have to say I’m suspicious that the Simple Regulator is as extraordinary as claimed. The problem, if there is one, might be the way performance is measured.  What I’ve seen extrapolates from Microset measurements, which could mislead.  More work needed.   For example, my clock is measured with two different methods.  Short term to about 0.1 microsecond with a Microset/PiCPET type approach.  Long term – several months – with NTP, which, without doing anything special on a Linux computer is never more than 100mS off atomic time, usually much better.   (Not difficult to DIY an NTP Stratum 1 service that’s never more than a few microseconds out.)

Anyway, measuring accurately inside a long run shows clocks wander, often for reasons unknown.

Though electronic clocks leave them in the dust, pendula are extremely interesting.  Practical pendulum development stopped leaving lots of room for more research when quartz clocks appeared in the 1930s.  Bernard Tikippe concludes with: The design of the Simple Regulator is presented here so it can be further developed.  I’m hooked, keep up the good work!

Dave

[steve7750Participant @steve7750]

Hi John,

The slow feed rate was due to the tiny cutter, longish reach and lowish speed.  The odds were stacked against me.

The chip slurry did have to be flushed away periodically with fresh oil.

I did pick up a small brushless Faulhaber motor that mounts to the Sherline laser cutter attachment.  It should be able to be controlled by the Masso G3, and hits 30K rpm.  The air drive high speed chucks just make too much noise (and reminds me too much of the dentist).  I’ve not had a chance to try it out yet, but it’s a promising solution.

[steve7750Participant @steve7750]

Glad to hear you’re hooked, Dave.   My goal was always to entice others to try their hand at this, and I’m sure Bernie would have been delighted with the direction things are heading.

Between ours, Chris’ and Colin’s we should be able to gather some decent data on the tuning procedure.  There’s nothing in print (that we’ve seen at least) describing the exact method, only the basic principle of “just the right amount of recoil”

[John HaineParticipant @johnhaine32865]

As I understand the Tekippe principle, it balances escapement deviation (ED), which increases the rate as amplitude increases; against circular deviation (CD) which decreases the rate.  At some amplitude these cancel out so the rate vs. amplitude curve has a maximum.  Operating the clock at this amplitude gives it zero first-order sensitivity to amplitude changes.  Tekippe’s objective was to make the clock insensitive to amplitude changes induced by torque variations.

Maybe this is the key to a tuning procedure.  He adjusted the pallets and hence the recoil to get the right amount of ED but this requires the clock to be stopped, the pallet clamps loosened, slightly adjusted, and re-clamped.  Getting a precise adjustment may need the pendulum assembly to be removed.  Then the clock is restarted, allowed to come to equilibrium, measured, then adjusted again if needed.

Contrast that with Clock B, which was adjusted through a “hill test”.  The escapement drive torque could be adjusted through the remontoire in steps, though I don’t think it was straightforward and still needed the clock to be stopped.  For each torque setting the rate and amplitude were noted, and the rate plotted against torque to get the hill.  In the case of Clock B it is actually operated on the downward slope ot the hill to get barometric compensation.

So why not fix the pallet angle for the Tekippe regulator and vary the torque by putting additional weights on the drive?  This could be done without stopping the clock.  For each new weight, allow the amplitude to stabilise, measure the rate, plot the new point on the curve, then once well above the hill summit, remove weights to get back to the one needed to be at the maximum of the rate/amplitude curve.

[Chris Raynerd 2Participant @chrisraynerd2]

Hi all,

There’s been a lot of excellent technical discussion lately – particularly from John Haine, as well as Steve, Gerry, Michael and Dave, whose insights have been incredibly helpful. I’m not adding anything groundbreaking to that side of the thread here, but I did want to share a quick update on the build.

After a lot of careful work and more than a bit of fiddling, the clock is now finally ticking! I’ve uploaded Part 7 of my build series on YouTube which covers the depthing process, shows the full train assembled between the plates, and includes the first test of the deadbeat escapement. There are a few little adjustments I had to make, including modifying the bottom posts and tweaking the top pillar near the escape wheel, and I talk briefly about those in the video too.

Still lots to do, but hitting this milestone feels great. As always, very open to feedback and and constructive criticism. I end the video realising I’ve only got 11.5 teeth span on the pallets but that was last night. I’ve spent today correcting that and the clock seems to be running… or at least it has been for the last hour. I show that in my next update. Thanks for watching.

Chris

[Michael GilliganParticipant @michaelgilligan61133]

Ticking is Good !

Well done, Sir

MichaelG.

[gerry maddenParticipant @gerrymadden53711]

Chris, it is really excellent to see these machines come together, and function too! I’m impressed by the speed with which you get things done, unlike me who just seems to spend forever thinking about them.

For your next project, one point that you may like to consider is the fixing of the bearing rings. We hadn’t previously talked about this aspect, but from your excellent video, you appear to have the outer rings retained in their housings and inners a loose fit with the shafts. Ideally this should be the other way around.

With a rotating shaft and a constant direction load you can imagine from the snip below what’s going to happen.

There will be a tendency for the shaft to simply roll around inside the inner ring as its rotated. In a clock, in the short term the effect will just be energy loss. In the longer-term oxidised steel dust will be released and get into places where you don’t want it to be. Therefore in your next design, try to fix these rotating rings on the shaft. A dab of loctite (…perhaps even Shellac…) will be useful here but whilst it sets, do hold the ring against an abutment to ensure the ring stays properly square with the shaft.

The outer rings can be left loose in their housings. They will not tend to rotate as the load direction is fixed with respect to them. But this is useful as you would find it very difficult to assemble the clock with both inners and outers fixed in their seatings :).

I very much look forward to your next video.

Gerry

 

 

[Author]

Posts