30-Legged Gravity Escape Clock

Discussion in 'Clock Construction' started by tok-tokkie, Jan 15, 2018.

  1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.
  1. tok-tokkie

    tok-tokkie Registered User

    Nov 25, 2010
    Cape Town, South Africa
    Country Flag:
    Thirty-Legged Gravity Escape Clock.

    Here is a video of the clock.

    Some comments.
    It will get a stainless bezel.
    The holes for the swing scale can be seen. Yet to be made.
    At 2m43 I say pendulum clock instead of pendulum cock.
    At 4m23 you can see some solder wire on the escape wheel & also on the last train wheel a bit later. I had completely forgotten about them. I am not at all embarrassed about them. It exemplifies that this is a real amateur clock.
    I also like the amateurishness of the video. It is slick stuff that I don’t like to be associated with.

    1.1 Escape complete.JPG

    Picture shows the escape on the back of the clock. Clock is sitting on a 12V alarm battery. I started in about 1996 with precious little to guide me. I will not go into how I arrived at this layout but it is more like a tower clock. Basically the Huygens drive is on the hour arbor with 12:1 gearing to the minute arbor and then 60:1 gearing to the seconds arbor. The arbors are concentric. There are four triangular frames. I am a mechanical engineer & simply designed the mechanism that I thought would best accomplish the job – not knowing anything about horology. I had a Photostat of Grimthorpe’s book about Big Ben and chose to use his double-three-legged gravity escape. When I tried the clock it needed vast weight to make it run. That escape turns 60° each tick so there was originally a further 10:1 set of gears from the escape to the second’s arbor. I realised that 1 gram at the escape wheel would require 10x60x12= 7200 grams at the drive. So I decided to make a 30-legged version of the gravity escape and dispense with the 10:1 gearing. That is the origin of my gravity escape – needing to reduce 7200 grams to 720 grams.

    When I first got it to work I wanted to publicise it as I was astonished that it seemed to be a novel idea. By this stage I was on the internet & had found the Mini Lathe Horology group on Yahoo. So I took a video of it working, posted it on YouTube and wrote about it on HLH.
    Gravity Escape 30 legged
    That was January 2012.
    Right now when I went to get the link two other important gravity escapes are listed.
    Mudge had done one 1795 Thomas Mudge's Gravity Escapement, circa 1795
    Bloxham too in 1853 Bloxham's Gravity Escapement

    In the comment beneath the Bloxham version is a long extract from Grimthorp’s book (The diagram appears after 13 secs). He claims that the fault with the escape is its propensity to trip and thus break teeth. I have seen Bloxham’s clock for it is in the Science Museum in London – I was delighted when I discovered it there.

    In his book Grimthorpe says the problem with the Mudge escape is it too is likely to trip. The difference between Mudge’s version and mine is essentially where the left hand pallet lifting tab is. Mudge has it at the locking face whereas mine is far away below the center of the escape wheel. I considered placing it as per Mudge but the line of action points almost directly at the pallet pivot so there is a very short mechanical lever arm – it needs a very large force to raise the pallet with that geometry. Placed as I have done the forces are very modest. I ran a thread asking about the power (weight & drop) required by various clocks. The thread has lost some of its data tables and much of its formatting since the last upgrade of this site but here it is: http://mb.nawcc.org/showthread.php?108078-What-power-needed-to-drive-various-clocks&p=822282#post822282 In the second last post the data is still there. The power is micro Watts but the micro symbol does not show. My clock now runs on 14 micro Watts whereas Tinker Dwight’s dead beat grandfather clock runs on 67 micro Watts. (JHE your final posting in that thread of 1.2474 kg falling 46.055 mm in 12 hours is 13 micro Watts.)

    Despite what Grimthorpe says my clock simply never trips.

    Baron Grimthorpe (unaffectionatly know as Lord Grim) had three names during his life; christened Edmund Beckett Dennison but he dropped the Dennison which was something his father had added to his own name. When his father died he inherited the title so became Baron Grimthorpe. He went to Trinity College in Cambridge. On his death he bequeathed funds for a clock using his escape to replace the clock in the tower at Trinity. That clock has been the subject of serious research which is reported at their website. They refer to Grimthorpe as Dennison – probably because that was his name when he was a student there. The Trinity Clock Click the Escapement tab on the left & you will get a nice animation of the double three-legged escape in action.

    I draw your attention to this quote on that page:
    What Grimthorpe had done when he designed the Big Ben clock was make it be massively over powered so that it could deal with wind loads and ice loads on the hands on the four faces of the clock. He fitted a big air brake (air governor) to the escape shaft to limit the speed so the action of the clock was prompt but not violent. The big advantage was if the load increased because of ice or wind the fan (air brake) had less work to do but the speed and action remained the same so the clock worked fine. But the crucial other advantage of his design was that the actual force applied to the pendulum was not changed at all as well as the pendulum being entirely disconnected from the loads imposed by the hands. So there were three aspects to his design 1) there was a big reserve of power which was automatically controlled to match the load & 2) the pendulum was completely isolated from those varying loads & 3) the pendulum was always given the identical impulse.

    But it is that quote by Schoof saying how wasteful of power the clock is which is quantified as twelve times the required power that I want to discuss. That is what my design eliminates. There is no power dissipating air governor on my escape. As shown above my clock runs on 16 uW whereas a conventional dead beat clock needs 67uW (four times as much). But that is not a fair comparison because my clock uses ball bearings for all the arbors so I should compare it to dead beat clocks with ball bearings. JHE’s second clock has ball bearings & runs on 13uW with a dead beat escape (am I correct about the ball bearings because you stated that you planned on jewelling it but is that for the train arbors or the escape arbour?). JHE’s earlier pin wheel escape regulator with ball bearings runs on 38uW. Rex Swensen has made a very elegant dead beat Vienna style clock with 1 second pendulum and ball bearings his clock runs on just 5.72 uW (Rex Swensen's Web Site Vienna Regulator Page) . So my clock is pretty good and certainly no power hog.
  2. tok-tokkie

    tok-tokkie Registered User

    Nov 25, 2010
    Cape Town, South Africa
    Country Flag:
    1.2 Double Three-Legged.jpg
    That is Grimthorp’s double three-legged escape. Originally I made a small version of that but then changed to a 30-legged version, so it had two 30 toothed escape wheels and thirty lifting pins. I realised that I could dispense with one wheel but I needed new pallets to keep the clock in beat with just one escape wheel. I made the new pallets out of thin titanium plate so had two plates for each pallet, one each side of the escape wheel with the lock pin secured in each plate so it was very rigidly supported.

    1.3 Two sided pallets.JPG

    Notice the lifting tabs on the pallets and the 30 pieces of hypodermic needle as the lifting pins in the rim of the escape wheel. Notice too how close the upper lifting tab is to fouling the previous lifting pin. That was a limiting factor for the Bloxham design where he had the lifting pins close to the arbor.

    1.4 Four Legged.JPG
    Now my escape was a variation of Grimthorp’s Four-Legged escape. This escape has just a single escape wheel and the pallets are just one each side of the escape wheel. I learned in due course that the pallets need to be as light as possible so I changed to just a single pallet leg each side making my version more like Grimthorp’s.
    Later I realised that the lifting pins are superfluous because I could simply use the tips of the teeth to lift the pallet. I had also come to understand that you need to reduce the inertia of both the escape wheel and the pallets as much as possible. It is the inertia that makes things want to keep moving. In the case of the pallets there is just a small friction force which must arrest it when the lock is initiated; hence the need to minimise the pallet inertia. For the escape wheel the lower the inertia the easier it is for the drive to index the escape wheel each second; so getting rid of those thirty lifting pins right out at the rim of the escape wheel was a good modification.

    1.1 Escape complete.JPG

    Here it is with the pins removed, you can just see the little titanium lifting tabs that I have cyanoacrylated (superglued) onto the pallets. Notice also that I have replaced the earlier eccentric PTFE ‘feet’ that contact the pendulum with lighter and easier to adjust hypodermic needles. With these low inertia pallets my clock never trips.

    On the tower clocks with the gravity escape there is another crucial contribution that the ‘fly’ makes. It is attached to the escape arbor with a simple friction clutch. When the lock happens the fly continues to rotate against this friction clutch so the escape wheel does not bounce on the lock face of the pallet. This is crucial in preventing the escape from tripping. In my case I don’t need the massive overpowering of the escape because there are no wind or ice load to contend with. So, with the much reduced drive the escape does not accelerate so quickly. With the small 6° index each second as against 60° (double three-legged) or 45° (four-legged) my escape builds up much less speed so no aerodynamic governor is required. No friction inertia damper is required so there is no need of a fly or the damper in my escape.
  3. tok-tokkie

    tok-tokkie Registered User

    Nov 25, 2010
    Cape Town, South Africa
    Country Flag:
    3 Banking Pins.JPG

    These are the pallet stops – called banking pins by Grimthorpe. It is important that the escape wheel has a little free run before it collects the pallet and raises it. I adjust the pallet stop so the tip of the escape wheel tooth is 0.5mm from the lifting tab. When the other pallet is pushed aside by the pendulum the drive first overcomes the ‘stiction’ in the train & gets the escape wheel moving before it collects the pallet.

    4 3 Pallet Lifter & Lock Pin.JPG

    This shows the titanium lifting tab in the banked position before the escape wheel is unlocked. The tungsten carbide locking pin is lower down. It is semicircular. It is epoxied in an aluminium collar to the 1mm laser cut titanium pallet leg. The holes in the escape wheel are a legacy from when there were lifting pins to raise the pallets – that was before I switched to lifting tabs running against the tooth tips.

    5 9 Pallet lifter & Lock Pin.JPG

    This is the other pallet with the lifting pin below the escape wheel arbor. The hole at the knee of the pallet leg is also a legacy. Originally there were a pair of pallet legs each side so the locking pin was held on each side of the escape wheel. There was a spacer between the two legs at the knee. I later learned that the pallets must be as light as possible &, more importantly, have as little inertia about the pivot as possible. Hence I dispensed with the second pallet leg.

    6 Pallet Beat Adjuster Pushers.jpg

    These are the feet of the pallets. The pieces of hypodermic needle fit each side of the pendulum rod. The adjusting screws allow me to set where the pendulum collects the pallet. That is how the beat of the clock can be set. However the Airy analysis of clocks determined that the optimum arrangement is if the impulse to the pendulum is equally spaced each side of the zero position (pendulum vertical). These adjusters permit that to be done. I wrap cooking foil around the carbon fiber pendulum rod with thin transformer wire to a battery and led light. I can then draw the pendulum to each side with thin thread and vernier measure where the pendulum collects and drops the pallet and adjust for symmetry. The clock will be in beat as a consequence. The impulse angle is not the same for each pallet.

    7 Counter Poise.JPG

    This is the top of the clock with the two pallets on the right. They are pivoted on two ball bearings each. The arbors are 2mm with titanium collars Loctited each side of the pallet legs. The wider inner collar is a legacy from when there were two pallet legs each side. Between the two frames are the poise weights for the pallets. The white is UHMWPE (ultra high molecular weight polyethylene). They are a friction push fit onto the 2mm arbors of the pallets. Into them is a 2mm screw with a tungsten carbide milling cutter insert epoxied to it. Tungsten has a density 50% greater than lead. The distance of the poise weight is thus easily and precisely adjustable by screwing the weight in or out to adjust the poise of the pallet. In this way the impulse to the pendulum can be adjusted and the swing of the pendulum regulated. In the original video there were long guitar string whiskers with red jasper beads as the poise weights. I rather liked those whimsical poises but I discovered that the moment of inertia of the pallets must be a low as possible. Using tungsten weights reduces the radius of the poises and that really minimises the inertia. If the pallets have much inertia the friction force between the escape wheel tooth and the locking pin on the pallet is marginal and the pallet can slide off and the escape trips. This is what Grimthorpe said is the fault with this style of gravity escape. So in his version, as fitted to Big Ben, he fitted a ‘fly’. The fly is a two paddled flat fan on the escape arbor that limits the speed of the escape. But, much more significantly, the fly is held by sliding friction onto the arbor. When the lock occurs the escape wheel stops but the fly continues against the friction force for a short time but this force on the arbor prevents the escape wheel from ‘bouncing’ when it locks. So the escape wheel comes to a dead stop against the locking face on the pallet. It is this that prevents the escape from tripping.

    Another criticism of the four-legged gravity escape is it is very noisy. My escape is not at all noisy as you can hear on the new video.

    So my escape addresses the failings of Grimthorpe’s four-legged for it requires really modest power to drive, it is not noisy, it is simple and dispenses with the speed limiting air governor & friction damper, it is easily adjusted but it retains the marked advantage of being a true gravity escape imparting a consistent impulse to the pendulum which spans the zero position symmetrically.
  4. tok-tokkie

    tok-tokkie Registered User

    Nov 25, 2010
    Cape Town, South Africa
    Country Flag:
    The Dynamics of my Gravity Escape.
    8 Escape Basics.jpg
    Diagram of the escape in position #1 of the description that follows.

    9 Escape Action.jpg
    Diagram of the position of the pendulum at each stage of the description that follows.

    0. Pendulum vertical. It has yet to reach the right pallet.
    The left pallet is impulsing the pendulum to the right.

    1. Pendulum collects the right pallet.
    The left pallet contacts the pallet banking pin so separates from the pendulum. That is the left impulse completed.

    2. The right pallet has been pushed sideways so the escape wheel now unlocks.
    When the escape wheel unlocks it turns and raises the left pallet and then it (left pallet) locks the escape wheel. This is a short event.

    3. The pendulum comes to a stop at the end of its arc and then starts swinging towards the left.

    4. It passes through the unlocking position (2) but nothing occurs on the downswing.

    5. Pendulum is now back where it collected the right pallet. The right pallet was raised from position 2 up to position 4 so it extracted a little energy from the pendulum. On the downward leg the pallet now contributes that same quantity of energy back to the pendulum. During the ride of the pallet on the pendulum it has extracted and returned the same quantity of energy to the pendulum. It is now about to start the impulse arc.

    6. The pendulum is again vertical swinging to the left. The pallet has been pressing against the pendulum since position 5. That has been impulsing the pendulum.

    7. The right pallet now contacts the pallet banking pin and parts company from the pendulum. From position 6 to 7 the pallet has continued to press against the pendulum. This is a continuation of the impulse that the pallet has been giving to the pendulum. The escape has adjustable elements so that the arc 5-6 is the same as the arc 6-7 so the impulse is the same while the pendulum is falling towards position 6 (vertical) as it is when it is rising after position 6. That is the optimum condition first recognised by Airy.
    Just as the pendulum drops the right pallet at position 7 it collects the left pallet.

    8. The pendulum has pushed the left pallet so it now unlocks the escape wheel.
    The right pallet has been resting against the pallet banking pin. The banking pin is adjusted so that the lifting tab on the right pallet is just clear of the tip of the escape wheel tooth. The escape wheel thus has a short free run before it collects the right pallet – this makes it easier for the drive weight to start the unlocking and indexing event.

    9. The pendulum reaches its left apogee where it stops and then starts swinging towards the right.
    After the unlock at position 8 the escape wheel turned so the tooth against the lifting tab raised the right pallet. This is a quick action. It is the raising of the pallet that transfers energy from the drive weight to the pallet. That energy is stored as potential energy. That potential energy is transferred to the pendulum which receives it as kinetic energy during the impulse from positions 5 to 7. At position 7 the right pallet is at its lowest level. The drive weight lifts it to position 1 which is the same as position 5 (where the impuse will commence).

    10. The pendulum passes through the same position as 8 where the escape was unlocked but no event takes place on the downswing.

    11. The pendulum is now back to position 7 where it collected the left pallet. It is now that the impulse from the left pallet starts.

    12. Pendulum has now completed its entire cycle and is back as it was when it started. The impulse from the left pallet will continue until it reaches position 13 which is the same as position 1.

    10 Escape Graph .jpg
    This picture uses the same position numbers. Vertical axis is height of the CofG of the pallet. Horizontal axis is position of the pendulum. The lifting of the pallet happens quickly which is shown by that event being completed before the pendulum reaches the next position. That lift represents the flow of energy from the drive weight to the pallet. The impulse which is slow spans two pendulum positions. That is that transferred energy now being passed on to the pendulum.

    The merit of a gravity escape is the impulse that the pendulum receives comes from the pallet. It is gravity acting on the mass of the pallet that provides the impulse. The impulse is derived from the pallet when both the pallet and the pendulum are disconnected from the drive so it is inherently always the same. When the drive raises the pallet the pendulum is disconnected from the escape so any variation in the drive force (for example faults in the drive train) are not felt by the pendulum. The time it takes for the pallet to be raised and locked will vary according to the state of the drive train but that does not affect the pendulum. What the pendulum does experience is any variation in the locking friction force between the pallet lock pin and the face of the escape wheel tooth. But this is very different to an anchor or dead beat escape where the pendulum is constantly in contact with the escape and thus the drive train. In fact in a gravity escape the pendulum is only connected to the drive when the drive is stationary so the effects of any flaws in the drive train are much diminished in the case of a gravity escape. This detached nature of the pendulum from the drive in a gravity escape is a crucial advantage that it has. Only while the drive is stationary does the pendulum attach to the escape wheel when it unlocks the escape.

    From bitter experience I can also say that the pendulum does feel any variation in the pivots of the pallets. I used good ball bearings with stainless races and ceramic balls with all the grease rinsed out completely so the bearings were absolutely dry running. The bearings had metal shields both sides and the clock case has a sealing gasket that the door closes against. Yet dust could find its way into those bearings & cause a change in the pivot torque. The pendulum swing would decay and eventually it would stop. Simply by exercising the pallets by rotating them fully a few times while the clock remained in the case would flatten the spec of dust & the clock would run again fine. I now have jewels for the pallet arbors but have not fitted them as yet. I was aware that ball bearings are not recommended for small oscillating motions but I gave them a try.
  5. tok-tokkie

    tok-tokkie Registered User

    Nov 25, 2010
    Cape Town, South Africa
    Country Flag:
    Layout of the Escape

    I will describe how I drew the escape of my clock rather than give a generalised description. You can scale the dimensions for a different sized escape.

    11 Grimthorpe Layout.jpg
    The basic geometry of the escape. I have elected to have a R32 escape wheel.

    The pallet pivots are on R64. They are mounted in 6mm OD ball bearings so I have spaced them 6.7mm apart.

    The pendulum should be pivoted on the R64 circle according to Grimthorpe’s dictates. I had it like that but wanted the pallets to rotate more than the pendulum (it was tricky when the pallets only rotated the same as the pendulum). I moved the pendulum pivot up by 16mm but then moved it 32mm up. That is how I suggest you do it. When you draw it you should find that the pendulum rotating 1.3° gives the pallet 1.625° rotation. The raised pendulum pivot does mean that there is a slight sliding of the pusher foot on the pallet across the pendulum rod. It is very small – Rex Swensen has the calculation on his website (www given previously). I fitted ball bearings as the pushers but found no advantage but a big disadvantage resulting from the much increased inertia of the pallet assembly.

    Points B & S are where the lock pins must be in the escape wheel. The lock pin face should be radial to the escape wheel so the pendulum can slide the lock pin out radially so there is no recoil to the escape. However it is also essential that the escape turns exactly 6° each beat. The tangent to the R32 escape circle from pallet pivot A is at point B. The other tangent is from R to S. If A is on a tooth tip then S must be exactly half way between two teeth. The escape must be laid out like that if the clock is to be in beat.

    12 3 pallet locked.jpg

    1. This description starts by referring to drawing ‘3 O’Clock Pallet. Locked Position Geometry ‘ mainly and also ‘9 O’Clock pallet. Locked Position Geometry’ to a lesser extent. There are just four drawings and they all show the completed parts – they are not step-by-step drawings.

    2. Decide the escape wheel OD = 64. Radius = 32. Draw circle center O R1,5 for the seconds shaft. Draw circle center O R32 for the overall diameter of the escape wheel.

    3. Pendulum pivot (mid point of suspension spring) is thus 64 above center of Escape wheel. Draw circle R64 center O. The pendulum pivot at twice the escape radius is specified by Grimthorpe but I don’t believe it has to be there and there are advantages stemming from the higher position I use. I discuss that lower down.

    4. The escape teeth are 3,5mm tall so draw a circle R28,5mm and lay out the 30 locking faces of the escape wheel. The teeth are 12° apart. The top tooth should be drawn at 3 degrees to the right of the vertical line. That is one quarter of the angular pitch of the teeth. My drawing shows all the escape teeth as I hope it will make the drawing clearer. Give the teeth a land at the top of about 0.25mm – not a sharp point. Top land has R32.

    5. Construct the tangent to the R32 circle from the tip of the tooth at B. The pivot point A is where that tangent cuts the R64 circle. I was fortunate that it resulted in pivots A & R being about 6.7mm apart.

    6. I want the pendulum to swing through less than 3° in total.
    The impulse to be 0.65° each side of pendulum vertical. That is known as the Airy condition (impulse spans zero equally).
    The lock to be 0.5mm. (That is how I made it but subsequently I reduced it to 0.3 by grinding the tips of the teeth – on a cnc machining center, not by hand.)
    The overswing is the distance the pendulum travels after unlocking the Escape Wheel. This is to be 0.3°
    So the total swing is 0.65°+0.52°+0.3° = 1.47° per side for the total of 2.94°
    Those dimensions were selected as I wanted the impulse to be about 50% of the swing – it can be diminished by adjusting the adjustable pallet stops and adjustable pallet pushers. The lock of 0.5mm is generous when you consider that a Graham Dead Beat escape has line contact. The overswing has been set at about 20% of the arc each side as it is during this time that the escape wheel must index and the impulse energy transferred from the drive weight to the pallet (this is when the pendulum is travelling at its slowest so the time is much more than 20% of the 1 second beat).
    I have marked the centreline of the pendulum when it is vertical (0°) as line POQ.

    7. The lock of the escape wheel takes place where the tangent from the pallet pivot meets the outside of the escape wheel. That tangent is shown as a dashed centreline in the diagram, ABC.
    OP = 64mm. OB=32mm . Trigonometry shows that angle OPB=30°. However our pallet is not from P but it is from A. It turns out that our angle is 27°
    Point B is where the lock is taking place. It is where the Escape wheel tooth is resting on the locking pin of the pallet. It is at the tangent point of ABC to the outside of the Escape wheel.

    8. Draw a horizontal line from B to S. S is where the other pallet will lock. For the clock to tick exactly 6° each swing then point S must be exactly half way between two escape teeth. The angle between the 30 teeth of the escape wheel is 360/30 = 12° To get line SB to meet the required condition the tooth at the top of the Escape wheel must be drawn 3° to the right of the vertical line POQ (quarter of angular pitch). The tooth to the left will be at 9° to the vertical.
    As a sideline: Grimthorp’s single escape wheel design had four arms (the double escape has three) . In our case if the pallets were pivoted as Grimthorpe specified about P then B & S between the true tangent points would both be exactly at a tooth tip. Just the opposite of what is required.

    9. The pallet lock pin will be a 3mm tungsten carbide milling cutter ground to a semicircular shape. Draw the lock pin with the 0.5mm lock at point B. The flat of the lock pin is on a radial line from the center of the escape wheel. (There is a drawing lower down.)

    10. Draw a circle R3 centerd on the lock pin at B. The inner edge of the pallet is tangent to the R3 circle at the pivot at A and this circle. Draw that line representing the lower edge of the pallet.

    11. Draw a horizontal line OC from the escape center about 40mm long. The ‘knee’ will be on this line. Extend the inner edge of the pallet to meet this OC horizontal line at C. We want the pallet to taper uniformly from 6mm wide at the pivot to 4mm at the knee and 2mm at the ankle. Draw a 4mm diameter circle touching point C (not shown in the drawing). Now draw the outer edge of the pallet leg as a line tangent to the R3 circle at A and tangent to the 4 diam circle at C. Trim this line so that it has a bump at the lock pin – use the R3 circle that we drew there. So there will be 1,5mm of metal on both the inside & outside of the tungsten lock pin. We now have the upper part of the pallet leg drawn down to the knee at C

    12. The bottom tails of the pallet, CD, can be any length you wish. I wanted them to be the same as the top part. Point D is on the R64 circle. (Most gravity escapes have them quite a bit longer.)

    13. We are drawing the pallet in the locked position (the escape tooth is on the lock pin). From point #5 the pendulum is to be at 0.65° at this position. Grimthorpe demands that the pendulum be pivoted at twice the escape wheel radius = on the R64 circle at point P. I had the clock like that but found it advantageous to raise it another 32mm. Draw a line from the pendulum pivot point P at 0.65° towards the right 164mm long. Set off a line 3mm away from it towards the right. The first line is the centreline of the pendulum when lock occurs. The second line 3mm away from it represents the lifting surface of the pendulum. (Note: The pendulum rod is 6mm carbon fibre.) Draw a circle at D R1 tangent to the push face of the pendulum with the center on the R64 circle centered on the escape wheel. That 2mm diameter circle is the mounting hole for the adjustable pallet pusher.
    Draw lines from the end of the upper part of the pallet leg at C to be tangents to this R1 circle at D. Those define the lower part of the pallet leg.

    14. Draw a circle R3 at D. That is the boss for the R1 hole at D

    15. Now to mark where the lifting tab is resting on the tooth of the escape wheel. The lifting tooth is the fourth tooth to the right of the vertical center line of the escape wheel. Mark this position by drawing a line from the pivot of the pallet at A to the tip of that tooth.

    13 3 pallet locked detail.jpg

    Here is a detailed drawing. The thin magenta line marks the tip of the lifting tooth, it is drawn from the pallet pivot at A. You can see the 0.5mm lock. Note that that occurs at the tangent to the R32 Escape Wheel (dashed green line) and that the center of the locking pin is on the radial line of the escape wheel tooth (solid green line). Note too that the lock pin is not aligned to the pivot at A; it is aligned on a line 1mm parallel to the tangent. The boss around the lock pin is R3 with the lower edge of the pallet tangent to that circle (only the outer part of the circle is shown). The upper edge of the pallet has a bump formed by that R3 circle. So there is 1.5mm of pallet on both sides of the lock pin. Note that the escape wheel teeth have a flat of about 0.25mm.

    14 3 pallet banked.jpg

    16. Now to draw the pallet in the banked position (where the pendulum left it against the banking pin).
    Copy the drawing to a new layer. Rotate the escape wheel by 6°. We are now drawing the Banked level (when the pallet is resting on the adjustable stop at the bottom (not shown) while the pendulum is in contact with the 9 o’clock pallet.

    17. Rotate the pendulum about its pivot point 2*0.65 = 1.3° clockwise. This is the pendulum when it drops the pallet on the banking pin. It is where the impulse to the pendulum ends (from the 3 o’clock pallet). So the impulse is 0.65° each side of zero.

    18. Draw a big construction circle centered on A passing through the center of the R1 circle at the foot of the pallet at D. Draw the R1 circle at D just as in #13 but at the new position of the pendulum; this circle must be centered on the big construction circle (not shown in the drawing). Measure the angle of the two R1 circles at D. It was 1.625° on my drawing.
    Rotate the entire pallet 1.625° clockwise so it now joins up with pendulum in the new position. Also move the marker line of #15.

    19. The tip of the third tooth past the zero line is what lifts the #3 pallet. (After the escape turns it becomes the fourth tooth of #15.) Marked E on the drawing. We want the Escape wheel to have a little free run before it starts lifting of 0.5mm. Draw a circle centered on tip of the tooth F R0.5mm

    20. We now have two points on the lifting tab marked. Draw a line between these two points (the end of the magenta marker line & where the red R0.5 circle cuts the R32 OD of the escape wheel). Extend that line 0.5mm each end. Construct vertical lines from the outer edge of the pallet to these points. When the pallet is cut (I had mine laser cut) the lower edge will mark exactly where the lifting tab must be attached. The shape of the lifting tab is now defined. It will be 0.5mm wider both sides so the lifting tooth will not reach the edge.

    15 3 pallet banked detail.jpg

    Detailed drawing showing the lock pin half way between two teeth. The R0.5 circle drawn in the tip of the lifting tooth. The complete lifting tab is shown. The lifting edge goes from the tip of the line coming down from the pivot (that line has been rotated together with the pallet). The lifting edge then passes through where the R0.5 circle cuts the R32 outer diameter of the escape wheel. So the escape wheel will turn 0.5mm before it picks up the locking tab. The locking tab is made 0.5mm wider than the two construction points. When the pallet is cut (laser cut in my case) the outline of the locking tab is included. Then the actual locking tab can be bonded to the pallet exactly where it is required.

    21. Copy the lifting tab to the locked level drawing and rotate it 1.625° backwards to match the pallet of this drawing. Make sure that it does not hit the adjacent tooth.

    16 9 pallet locked.jpg

    22. The 9 pallet is created in exactly the same way except the lifting tab is on the lower half of the pallet and the flat face of the lock pin at S has the flat face facing downwards towards the elbow T. The pendulum is 0.65° to the left of the centre line. Draw the #9 pallet just the same as the # pallet with the lock pin having 0.5mm lock.

    23. The lifting tab is in contact with the fourth tooth past the centre line on the lower half of the escape wheel. Mark that position with a magenta line from the pallet pivot at R.

    17 9 pallet locked detail.jpg 17 9 pallet locked detail.jpg

    18 9 pallet banked.jpg

    24. Copy drawing to another level. Rotate Escape Wheel 6° clockwise. Rotate the pendulum 1.3° anti-clockwise. Draw a construction circle centered on the pallet pivot at R passing through the center of the R1 pallet pusher foot (circle not shown). Draw R1 circle tangent to the pendulum with its center on the construction circle. This is where the pendulum pusher will be. Measure the angle that the pallet has to be rotated to bring it into contact with the pendulum. In my case it was 1.625°.

    25. Rotate the pallet to be in contact with the pendulum. Also rotate the marker line drawn in #25.

    26. The lifting tab is created in precisely the same way as the one on the #3 pallet. It is lifted by Escape Wheel tooth 4 after the zero line at the bottom. Draw the R0.5mm circle on the tooth tip. Draw the face of the lifting tab as a line from the end of the magenta marker line and the intersection of the R0.5 circle and the R32 outer diameter of the escape wheel. In fact the magenta marker line passes very close to the second point. Mark the lifter tab making it 0.5mm longer each side.

    27. Copy the lifting tab to the locked drawing. Rotate it 1,625° clockwise. Check that it does not foul the adjacent teeth

    19 9 pallet banked detail.jpg

    28. Then spoke the Escape as you choose. The forces are very small. The lighter the Escape wheel is the easier it is for the drive weight to accelerate it each second. A lighter Escape wheel also reduces the noise when it locks.
  6. tok-tokkie

    tok-tokkie Registered User

    Nov 25, 2010
    Cape Town, South Africa
    Country Flag:
    In my description of how to lay out the pallets I have used the layout mandated by Grimthorpe. I will here show that you are, in fact, able to place them more or less as you please

    11 Grimthorpe Layout.jpg
    The opening diagram shows the escape wheel of R32 with the pallet pivots on a concentric R64 circle. Even that is wrong. That is the layout for a double three-legged escape (what I initially made). For the four-legged escape (which my thirty-legged escape is a variation of). Grimthorpe writes:

    Scaling that for a R32 escape wheel the pendulum pivot should be on R83.2 with the pallet pivots at slightly less. I have included much of his didactic description as a footnote but I can’t resist including this extract:
    I shudder to think what would he have said about my thirty-legged version. I will also challenge several of his other stipulations.

    alternative 3.jpg

    Here I have raised the pivot points up to the level of the pendulum pivot. The same teeth are used for the lock. This is arbitrary chosen simply so I can show graphically how freely the pivots can be positioned.

    The #3 pallet is in the same locked position so the tangent is shown passing through the original pivot. The two pallets could be cranked or one could be behind the escape wheel and the other in front of it.

    The #9 pallet has two teeth marked as the possible lifting teeth with the R0.5 free run circles drawn.

    alternative 4.jpg

    Here the #9 pallet is locked. I have chosen to use the left tooth as the lifter.

    The advantage of this layout is that the lock pin is proportionally closer to the pendulum pusher foot so its motion is also proportionally greater. So a lesser swing of the pendulum will unlock the pendulum even though we have maintained the same angle of impulse. The total swing of the pendulum could thus be reduced with a consequential reduction in the circular error. The distances AB & AD are (A is the pivot for #3 pallet; the drawing still has the labels on the original pivot points).


    So in the original layout the lock pin moves 43% as far as the pendulum pusher foot but in the revised version it moves 57% so unlocking will happen as a smaller pendulum swing.
    The pallet would need a lot more counterpoise to restore the impulse force to what it was.

    Alternative 1.jpg

    Here I have chosen to use the teeth closest to 3 o’clock & 9 o’clock. Fist drawing shows the #3 pallet locked so the pivot A is on the tangent to the R32 escape wheel. It slopes in by 3°. The pendulum is at the same angle as before with the pusher foot in contact. Four magenta construction lines for possible lifting pins are shown.

    The #9 pallet is in the banked position. The pivot at R is at the same height as on the actual escape wheel. This pallet would have been laid out in the locked position then copied to this drawing and rotated to bring it into contact with the pendulum. Three construction lines are shown for the lifter tab. Two tabs are shown as the uppermost construction line does not work out properly. I would use the middle line. Here the lifters are shown with the 0.5mm free run gap for the escape wheel before it collects the pallet.

    Alternative 2.jpg

    This is the escape wheel in the other position for determining the pallet layout. I would use the lower lifting tab on the #3 pallet.
    With these pallets there would need to be proper poise weights added for there to be any impulse. The escape I have made actually has counter-poise weights so I can reduce the impulse so as to limit the pendulum arc.

    What I am wanting to show is that you can choose many a tooth as the locking tooth then the pivot must lie on the tangent from that tooth and it can be at any distance you prefer as long as it is on that tangent line.

    In fact you could even use a pallet from each of these two examples.

    Here is an extract from Grimthorpe:

    I have placed the escape wheel with the uppermost tooth 3° past vertical. That is one quarter of the angular pitch of the teeth. Then I use any convenient pair of teeth for the lock and place the pivot on the tangent to that tooth. This ensures that there is no recoil when the pallet is pushed aside to unlock the escape wheel making it easy for the pendulum. What Grimthorpe is saying is that a little recoil is desirable for it makes the lock more definite. To do this place the top tooth at, say 6.5° but retain the original position for the pallet pivots. Now the lock pin for the #3 pallet must be moved to match the new position of the tooth. The tooth must land flat on the pin when lock takes place. This pin will be slightly tilted so that when unlocking takes place there will be a little recoil. But when the tooth lands on the pin there will be a slight force stopping the pallet’s rotation so the lock will not rely only on friction.
    Similarly for the #9 pallet the tooth will be a little higher and again there will be that slight restraining force on the pallet.

    The escape will still be in beat since we have moved both teeth by the same angle. I have used 0.5° for illustration. I have not done a CAD drawing to see if that is a reasonable angle.
  7. tok-tokkie

    tok-tokkie Registered User

    Nov 25, 2010
    Cape Town, South Africa
    Country Flag:
    Some extracts from Grimthorpe’s writing. (His book is available as a free pdf from both the Google scans of significant books and from The Gutembourg Project.)
    Initially for the four-legged escape.

    The scapewheel speaks for itself as to the long locking teeth; but it has two sets of lifting pins near the centre, pointing alternately backwards and forwards, one set lifting one pallet and the others the other, the pallets being in different planes, before and behind the wheel, with one stop S pointing forwards and the other S0 backwards. This cannot be done otherwise with an even-numbered wheel, and though it can with a five-legged one, that is an inferior construction in other respects, and is much more difficult to make rightly, and has not one advantage when it is made. I say this from experience now, as I did before from theory. For with the strange propensity of mankind for taking more trouble to do wrong than it would take to do right, some persons would persist in making the escapement with five legs. I thought it might possibly save a little force or weight on the train, but it does not even do that; nor anything else, except give more trouble to adjust: whereas the four-legged one is so easy that the very first I had made went at once perfectly without any alteration.

    Another great advantage of these escapements is that the length of the teeth or legs, and the largeness of their motion, make the pressure on the stops, or the work of unlocking by the pendulum, insensible; and therefore also they are incapable of holding up the pallets so as to cause ‘approximate tripping’ by any force that you can apply to the great wheel, provided the escapement is made properly. But with that same genius for doing things wrong when it is as easy to do them right, some persons have made the angle CSE, at the stop which is struck upwards, less than 90°, and then have said the escapement failed because it sometimes tripped, as it was pretty sure to do. For safety it is as well to put the up stop a little higher, and the down stop a little lower than their proper theoretical places, which are where both the angles would be exactly 90°.

    That determines the theoretical distance of the pallet arbors C from E the scapewheel centre. If its diameter is 4 in. that distance is 5.2, and that is the proper distance of the top of the pendulum spring above E. The pallet arbors must evidently be a little lower. Mr. Bloxam made them cranked (see p. 84) in order to get their common axis in a line with the top of the spring; but that is an unnecessary refinement, at least for these escapements, and is never done. And as no point in the pendulum really swings quite in a circle, I doubt if the friction of the pallets on it would be sensibly less for their being both made to describe the same circle by cranking their arbors: at any rate it is too insignificant to care about.

    Next a bit about the double three-legged escape

    The distance of the lifting pins from the centre should not be more than a 40th of EC, or else the angle of impulse 2 gamma will be larger than is found expedient. It is difficult to make the pallets light enough even with a small and the larger it is the lighter they must be. The length of their tails down to the beat pins is arbitrary, but I found the Westminster clock perform decidedly better with the pallet tails long than short. The length here shown does very well, and it looks neat to make the two parts reciprocally parallel. The pins should be placed so that the lifting may take place equally across the line of centres CE, because then it is done with the least friction. For this purpose the pins which lift the lower pallet must be set on the radii which run along the acting faces of the teeth, and the other set of pins half way between them, with reversed screws, as I said before.

    In this case the distance of the pendulum top from the scape-wheel centre evidently = diameter of scape-wheel. The lifting pins should not be farther from the centre than a 36th of this, or the pallets have to be inconveniently light and thin: the pins and arbor may be solid as a three-leaved pinion. They should be so placed that the one which is holding up a pallet and the one which is going to lift next may be vertically over each other, the third being on a level with the centre—i.e., they will stand on the radii which form the acting faces of the teeth of one wheel, as you see here.
  8. tok-tokkie

    tok-tokkie Registered User

    Nov 25, 2010
    Cape Town, South Africa
    Country Flag:
    I just stalls with the loading arrow going round & round. I rather depend on it to show the clock in action. I will post the link again & hope the software lets it remain as a simple YouTube link you can click on to activate rather than embed it as a running vid.

    That did not work so I will load it again in double quotes.
    " YouTube "
    Ah! That fooled the software. We only have 30 minutes in which we may edit a post so I can't fix the opening post.
  9. Phil Burman

    Phil Burman Registered User

    Mar 8, 2014
    Suffolk England
    Country Flag:
    Region Flag:
    #9 Phil Burman, Jan 17, 2018
    Last edited: Jan 17, 2018
    Thanks for a great write up tok, looks like it has been a long and interesting ride, all the background info makes for a great read. There is a great deal of technical info in there so it will take a few readings to assimilate it all.

    Is that an involute tooth or two that I see in the lower end of the train?

    Is the the requirement for lots of power (drive weight) inherent in the Grimthorpe design or can it be mitigated by possibly reducing the overall scale of the escapement for example?

  10. Allan Wolff

    Allan Wolff Moderator
    NAWCC Member

    Mar 17, 2005
    Tulsa, OK
    Country Flag:
    Region Flag:
    Amazing clock! Love the video and I an still absorbing all the technical information you provided. I always wondered why the original gravity escapements didn't have more "legs" and now you have proven that it can be done. Fantastic design; and it looks great too!
    Thank you for sharing,
  11. tok-tokkie

    tok-tokkie Registered User

    Nov 25, 2010
    Cape Town, South Africa
    Country Flag:
    There was a lot of trouble with the video. I uploaded it & made it Public but it acted as if it was Private. When I realized that it was not loading I did the second link & reported the problem to the Mods. The mod (Steven) asked if it was Public & it was only after that that I found it acted as Private - YouTube knows my IP address so it lets me view it so it is not obvious. I managed to make it public. I am sad the original post still can't show it. Without the video of it in action this thread is rather turgid. It has much too many words.

    Gear teeth. Originally it had all involute teeth. In wanting to reduce the drive weight I made the train in cycloidal teeth all with the same PCD & tooth count. I then did detailed measurements of the torque coming through each stage. It turned out to be pretty much the same. Being a mechanical engineer my background favoured involute but having been exposed to horology for a while on the internet I had come to like the elegant shape of clock wheels. So I decided to use involute for the first 12:1 and cycloidal for the later 60:1

    The original Grimthorpe design requires considerably more power because there is a big 'fan' which is a speed limiting governor for the escape arbor. The fan absorbs a lot of energy. But absolutely critically that fan acts as a friction damper to prevent the escape recoiling against the sudden force when it locks. It holds the escape arbor & wheel dead on the pallet locking face. There is that quote by Schoof about it which I give and the data from Trinity that the energy used is 12 times the energy delivered. In my design there is no 'fly' and my clock runs on very little weight (it is an 8 day clock). I enquired about the energy used in a thread that I linked higher up.

    Woodward fitted a friction damper to his W5 for exactly the same reason - eliminate bounce. His clock tripped repeatedly until he fitted it. A version of W5 is being made in Sydney Australia by the horological society there. They have modified the design by increasing the number of teeth on the count wheel. It has allowed them to eliminate the friction damper.
  12. jhe.1973

    jhe.1973 Registered User
    NAWCC Business

    Feb 12, 2011
    Arizona, USA
    Country Flag:
    Region Flag:
    Hello tok,

    Thanks so much for posting your clock. You have added so much to this forum that I knew your clock would be a worthwhile addition.

    WOW, is an understatement!

    Congratulations on what you accomplished here. I noticed right away how much quieter your clock is than the last gravity escapement version I heard. :clap:

    Thanks also for providing so much detail in your diagrams and text. I really must study it all soon!

    Just one question though. With that view out your windows, how do you tear yourself away to look at your clock?


  13. tok-tokkie

    tok-tokkie Registered User

    Nov 25, 2010
    Cape Town, South Africa
    Country Flag:
    The view is interesting. I am a lapsed surfer but said I wanted to be able to see the surf from my bed. It took 18 months to find this house & Antonia (wife) said to the realtor 'My husband will buy it tomorrow'. When I came here I said to the agent that it was exactly what I wanted & I would put in an offer. She asked 'Don't you want to see the rest of it?' My favorite surf spot was behind the stadium but it only works when the swell is really big so I could check it out from home before the stadium arrived. But there are three other spots so you just move on to see which is working with the current swell size. That is why we moved here.

    As to the view. The view is the backdrop but it is actually the action that attracts you - the ships moving in & out of the harbour or who is tied up. Huge waves crashing over the breakwater during a storm.
  14. John MacArthur

    John MacArthur Registered User
    NAWCC Member

    Feb 13, 2007
    Tokkie -- I too am impressed by your thinking "outside the box" on the gravity escapement, and by your exhaustive write-up on the design. And, to add to the list, I note that, re: the power of running clocks, two of my Graham regulators run on 33 micro-watts, and 27 micro-watts respectively. The one with a larger drum (33uW) seems to have a more comfortable amount of "over-run" on the dead faces of the pallets, and as such will probably run more reliably for longer between cleaning.

    Again, quite a remarkable write-up.
  15. clockkeeper

    clockkeeper Registered User

    Dec 7, 2010
    Keeper of the Clock
    Trinity College Cambridge
    Country Flag:
    #15 clockkeeper, Jan 20, 2018
    Last edited: Jan 20, 2018
    Very impressive! At 13μW your 30-legged escapement is a clear improvement. The Trinity Clock trin-hosts.trin.cam.ac.uk/clock now runs at 7mW (the 12mW that you quote was before we changed from 5-day to 8-day running, and I didn't change the going weight).

    By dimensional analysis I think that (all other things being equal) the driving weight goes as the fourth power of the "size" of the clock mechanism (incidentally the same exponent as the Dambusters man Barnes Wallis figured out for blowing up dams, that the mass of explosive needed goes as (dam height)^4 ). This means that power goes as the cube of size.

    So taking Ralph's 30-legged clock to be about 1/3 the size of the Trinity Clock mechanism I'd say that the 30-legged escapement is about twenty-times more efficient.

    Just an amusing aside: I think one reason for the extra power needed for tower clocks is that pigeons sometimes perch on the minute hand. The attached letters from Trinity College Cambridge, decades ago, tell a nice story.
    In 1944 there was a problem put down to excessive oiling - the clock had stopped at 7.45am. And in 1955 the stoppage at 7.44pm was put down to "so much friction in the mechanism". On both occasions Smith of Derby were called in to do a thorough maintenance. No fault was found with the clock. Knowing what we know now the stoppages were caused by pigeons sitting on the minute hand.
    In Feb 2011 we fitted an anti-pigeon wire on the Trinity Clock and we haven't had a problem since.
    The Trinity Clock

    TC_Clock_1944_letter.jpg TC_Clock_1955_letter.jpg
  16. tok-tokkie

    tok-tokkie Registered User

    Nov 25, 2010
    Cape Town, South Africa
    Country Flag:
    There is a silly typo in post #6. At the bottom I write about intentionally adding a little recoil as
    suggested by Grimthorpe. I use 6.5° as a suggestion. That should read 4.5°.

    Those familiar with the magnificent clocks made by Tekippe will know that he intentionally introduces a
    little recoil in his dead beat escape. Using a MicroSet he has been able to find the precise amount of
    recoil that automatically compensates for circular error. He writes:
    Sadly I don't think that technique will work with this gravity escape. Tekippe adds weight to simulate
    a change caused by inconsistency in the drive train. Adding drive weight increases the impulse in a
    dead beat escape. In a gravity escape the impulse is unaffected by adding drive weight. But the
    locking force is increased. That increases the load on the pendulum thus decreasing the swing. The
    converse of what happens in a dead beat escape.

    Tekippe clock discussed in this thread: https://mb.nawcc.org/threads/tekippe-precision-
    regulator.79463/#post-1073964 which gives a link to
    Video of Bernie Tekippe & one of his clocks.

    Here is a very interesting experiment on a gravity escape by Bryan Mumford (who makes the MicroSet).
    On a tower clock with gravity escape he found that increasing the weight of the gravity arms the
    amplitude of the pendulum was DECREASED - just the converse of what I find.
    Circular Error


    I am absolutely delighted to have impressed clockkeeper. clockkeeper = Dr Hugh Hunt of Trinty College,
    Cambridge who must be the world authority on gravity escapes since he has supervised post graduate
    students in analysing the gravity escape there.

    The back story to the bit about pigeons. They occasionally sit on the minute hand when it is horizontal.
    That adds to the load in the 45 minute area disrupting the clock or even stopping it as shown by the
    two letters. When the post grads were looking at the data they realised that it was pigeons causing
    the previously mysterious disruptions. I emailed clockkeeper and suggested the wires to keep the
    pigeons at bay. I had done it to the balcony handrail outside my bedroom to deter pigeons who came to
    get water from my lily pond but were also fouling the balcony. He fitted them and the problem has gone
  17. tok-tokkie

    tok-tokkie Registered User

    Nov 25, 2010
    Cape Town, South Africa
    Country Flag:
    #17 tok-tokkie, May 4, 2018
    Last edited: May 4, 2018
    I have just stumbled across this YouTube video of a gravity escape with separate pallets for the detent action (operating on a 4-legged escape wheel) and another pair of pallets for the impulse of the pendulum (operating on a 30-legged escape wheel). Made somewhere between 1910 & 1920.


    Very different to my design being both a bit of Grimthorp's 4-legged gravity escape and a bit of a 30-legged escape. So my design is not as original as I thought.
    From 2m34s on the video you can see the wheel on the 30-tooth escape wheel arbor meshing with the pinion on the 4-legged escape wheel which also carries the fan. The 30-legged arbor is also the seconds arbor.

    It retains the fly so still needs a lot of power. It has to be wound every hour. I don't understand what the advantage was of having two sets of pallets. Kinematically it is a half way stage between the usual 4-legged gravity escape and my 30-legged gravity escape.

Share This Page