Debugging continues, correction for seconds hand, design issue - April
In this month's
installment Buchanan creates a correction for the functioning of the seconds
hand. Until this point the seconds hand would move in two seconds steps as
would be expected since the pendulums are a two second period. After the the
machine underwent a complete finishing job the seconds hand began to run
erratically. This is demonstrated by the video below.
So what happened? This is an interesting case study. Throughout the
project certain adjustments had to be made when problems were encountered.
One of these was the fact the we were having too much inertia due to the
mass of the dual Wagner remontoire.
The reason for
this problem is the moving mass from the fairly large and heavy twin Wagner
remontoire as the cages cycle, especially at the point where the cycle ends
and the remontoire is stopped. The remontoire provide the power to the
escapement, the pendulums regulate the escapement.
pendulums have a two-second period there is a lot of time for feedback from
the cycling of the remontoire to disturb the second hand. There is also
the fact that there is a built in cushion for the remontoire release cam to
prevent severe jolting of the remontoire cage when it is stopped at the end
of its cycle. This was needed because of the mass of the mechanism coming to
a sudden stop. Furthermore, there was also a spring cushion built into the
two drive arbors to the twin escapement wheels. Though these features were
needed to make the remontoire cycling safer and smoother, this affected the
defined time interval needed for the second hand to advance at precise
two-second intervals, and this is clearly and painfully revealed in the
thing about this is even though the second hand is clearly erratic, if one
looks closely at the actual escapement wheels, there is no erratic behavior.
In other words, the clock keeps time, but the seconds display is erratic.
All of the devices we invented to make the clock work properly did work, but it
is simply the fact that the second hand is actually reading out the slight
movements of the cushioning springs.
Why did this not come out before now? Because before the final
finishing there was enough drag or in better parlance, friction, to slow
down the the recoil and cushioning elements to allow the seconds hand to
operate without notice. But once those elements were made to operate with
much less friction, they overran the two second interval. The answer is not
to make those elements less efficient, to reintroduce friction, but to see
how to make the system work as it should.
The video shows the erratic behavior of the seconds hand. It seems to coast
along instead of steadily ticking off two second intervals in line with the
two second pendulums. Sometimes two seconds, sometimes more, or less even
some backwards movements! Look carefully at the escapement wheels to the
right, one does not see any erratic behavior here. So the clock will run
correctly even with the odd looking seconds hand, but of course, this cannot
be allowed to remain.
The next day (the next day!) Buchanan came up with an
elegant but somewhat delicate and complex solution. Not unlike what he has
managed to do with other systems that were retrofitted in mid-flight. Other
examples are the adding of the minutes dial to the sidereal time
complication as well as the entire retrograde movement of that module.
The correction for the entire wheel turn / locking of the calendar system.
The changeover from weight to spring drive, and many others I cannot recall
at this point.
Buchanan will make the pallet
work exactly as on the grasshopper escapement so that if the balance arc is
sufficient for the grasshopper to run, the dead beat will also run. The
pallet will be gravity biased towards the pad and have a spring biased beat
setting in the other direction so that if a pallet ever crashes there will
be shock absorbing to avoid damage.
So what has been made here is a what looks like a supplemental
escapement to control the seconds hand, but is actually a count wheel. One might ask, if the
count wheel is attached to the same arbor as the current seconds hand why
would it too not move erratically? The answer is that there is a spiral
spring that will contact to the arbor at one end and the count wheel at the
other, so even when there is erratic feedback to the seconds arbor the
spring will absorb this while keeping a clockwise bias on the count wheel and that wheel is controlled by the
pendulums which are unaffected.
There is another serendipitous consequence to this design. We now have the
chance to turn the seconds readout from a two seconds period to a true seconds jump.
he conversion from a
two second pendulum to a one second readout on the dial is accomplished with two pair of jewel pallets. Each pair has
has pallets of unequal length. There is a video later on this page that
demonstrates how this works.
We have converted a two seconds
clock to a one second readout. It is the
reverse of what is sometimes seen in a one-half second pendulum clock with a one second
readout on the dial. This type of escapement complication is called a "coup perdu"
which is French for 'lost beat'; every other beat of the pendulum is lost so
as to get a one second movement on the dial. Here we have a gained beat.
Instead of the hand moving once every two seconds, it will move
twice. What is the French translation for our new type of escapement of a
'gained beat', perhaps "battement gagné"? This certainly qualifies as an
additional complication, just as would a coup perdu.
The other positive development from this arrangement, besides having a jump
second as one would expect from a clock is that one can actually hear the
ticking. The Harrison grasshopper style escapement is practically silent.
So to recap, this new count wheel has nothing to do with the control of
timekeeping of the machine, that remains with the grasshopper escapement and
pendulums. What this new arrangement does is convert a two second period
into a one second dial readout while eliminating the erratic feedback to
that seconds arbor from disturbing the seconds hand. Since the pressure on
the pallets are negligible as it is only the tiny coil that drives the
escapement wheel, no oil will be needed on this system.
So it is important to understand that while this may look like a
conventional escapement where the escape wheel is powered by the time train
and imparts an impulse and locking to the escapement and thus the pendulums,
here the direction of power is reversed and the escapement is moved by
pendulums that are powered by an independent source (the Wagner remontoire
cycled by the main spring barrels). That escapement provides locking to the
count wheel at intervals determined by the four jewel pallets. The count
wheel is turned by the current seconds hand through the spiral buffer spring
mounted between the seconds arbor and the count wheel. There is no impulse
from the count wheel it is only being locked by the escapement at one second
The white paper mock up shows where the proposed escapement system for the
control of the seconds hand will be located.
Another view of the white
paper mock up where the proposed escapement system for the
control of the seconds hand will be located.
This drawing better shows the shape that the escapement will have.
Each pallet stone will be in a brass carrier that is
adjustable so the time interval between ticks can be adjusted in relation to
the time/arc movement sine wave. The spacing of the seconds around the 360
degree circle is set by the radius of the pallet stones. There will also be
a beat adjustment and a spring safety.
The spoke style of the escape wheel is impressive and is a unique design not
duplicated elsewhere in the machine. It will be beautifully complex and may
be the most difficult to make, other than the Harrison escapement pair that
have the swirled spoke layout. Those were made in September 2008. This will
be "the last wheel" made in this project
Even though the wheel will be made using CAD-CAM, one will later see next
significant amount of hand finishing that must take place to make the
corners sharp. A cutting tool, by it nature, will always form a rounded
inner corner since it is round.
The red line is the actual profile and the blue and
green lines are the path the centre of the tip of the cutter travels so you
can see there is an allowance for the thickness of the cutter. The repeated
parallel lines are each cut at an increasing depth.
These three photos show the fabrication of the escapement wheel. The last
photo almost looks like a set of small diamonds sunk into the brass blank.
If one looks carefully, the inner corners do show a slight curvature of the
cutting tool. These will all later be finished by hand to match the sharp
corners in the rest of the wheel work in the machine. The wheel diameter is
2.25" (6 cm); it will be located in a an area where it will be easily seen
and will move at a fairly rapid rate of once per minute giving further
The mill is now put to work making the escapement and pallet holders.
Looks like I will have to do a quick re-do of the
former “last wheel”. This refers to photos and a video taken in
April 2019 of the
last wheel made in the orrery module, exactly two years ago. At the time
this was thought to be the last wheel that would be made for the project. It
is amazing that the one more ‘last wheel’ would be needed exactly two years
later. This highlights just how many things are needed to be done besides
just the fabrication work. Some of the time has been used in the
spring drive conversion but the majority has been in the final finishing and
debugging process; of which this was a part.
These two photos show the main parts straight from the mill.
The four photos above show the evolution of the pallet jewels, first photo in
the rough, second photo shaped into rectangles, third a curvature is
beginning to be lapped to contour the surface and fourth the jewels
installed. Notice that they are of unequal lengths and look a lot like
fangs. In fact Buchanan originally called the escapement the
“Contrarie Duo Coup Perdu Dent de Rat pallet.“ Loosely
translated as 'contrary (opposite of) a coup perdu with rat (fang)
pallets'? Dent fits in, because the slower two second pendulums were used on
a Dent clock made in the 1970's called the Concorde because it had a
frame that looked a lot like the fuselage of the supersonic commercial jet
named the Concorde flown at
the time and it inspired my choice of the slower period pendulums. One can see an example of the Dent Concorde clock here.
Note the pair of uneven jewel pallets.
This video shows how a two second period can be converted to a one
second readout on the dial It is the reverse of what is commonly known in
horology as a Coup Perdu escapement where a 1/2 second pendulum is made to
show a one second readout on the dial. Coup Perdue is French for lost beat,
where every other beat of the 1/2 second pendulum is skipped to get the one
jump second readout from the seconds hand on the dial. The action on the
video is uneven because it is the first trial, the escapement is not moving
at a constant speed and before adjustments were made on the pallets to get a
perfect one jump second. It is simply a concept demonstration.
The first photo shows the grooved area where the inner coil of the spring is
located. The second photo shows the skeletonized spring barrel where the
spring will be housed.
The spring is now slipped into the groove shown on the first photo above,
next the spring inserted into the spring barrel groove.
The new count wheel is now assembled. The pink, bronze, wheel is the
original seconds drive wheel. Note that the buffer spring is NOT firmly
attached at either end to the arbor or spring barrel, the significance of
this is shown in the video below.
This video shows the coil buffer spring that eliminates the erratic
rotation of the center seconds arbor through this spring and provides
rotational torque to the seconds escape wheel which is controlled by the
What is being demonstrated is the safety/setting feature in this
design. If the operator tries to quickly spin the seconds hand around it will slip
either on the arbor or the inner spring barrel in either direction. But if
left alone or turned slowly it grips in either direction. In practical use,
it is important that it grips the seconds hand to follow the escapement
wheel in a clockwise direction. It would have been easier to design and
build and require far less imagination to have simply had the inner coil
permanently attached to the original erratic seconds arbor and the endpoint
of the outer coil to the escape wheel - however we would have lost the
safety feature and it would have made setting this hand to time very
This is another example of the many safety features built into the
clock to prevent damage from operator carelessness or ignorance and these
features also make the machine more user-friendly. One hundred years from
now the operator will have less knowledge than we do now! I hope to
save as much information in both print and electronically to help the future
operator correctly care for the machine but nothing can be guaranteed.
The first photo shows a view of the escapement pallets from the rear. Next
the yellow oval shows the swing limiter, needed in case the pendulums
encounter an 'over swing' which could happen if the operator when starting
the clock pushes a bit too hard. Another example of the safety features
built into the machine.
The completed escapement. To the left is the antifriction wheel which
receives the impulse from the pendulum. To the right is a knurled knob that
allows fine control of the beat, through this one can get an exact match of
the jump seconds dial hand over each seconds partition on the dial. The jewel
pallets are also adjustable for depthing, but should not need any intervention
under normal circumstances, but is useful should a replacement ever needed.
yellow arrow points to one of the pallet pairs, the escapement wheel can bee seen
above and the curved escapement below.
the completed reverse coup perdu seconds correction system is installed. In the
foreground is the knurled knob, background the red jewel pallets, and a portion
of the escapement wheel. Next month this will be finished.
Total parts count to date: 7634. This should be the final count for the machine,
less the case, and other miscellaneous items as keys and tools.