POUVILLON RESTORATION PROJECT - October 2011
Calculation and fabrication of remaining cam contours and spokes.
Note that the format of this section of the web
site is designed for the newer wide screen monitors, those using the
conventional 3:2 ratio monitors may need to move the horizontal scrollbar at
the bottom of the screen to see the entire page.
I have asked Buchanan to provide what I call a
'forensic' report. That is to record his observations as he goes along. I
will provide the .MP3 audio file
for each segment but just in case your security settings will not allow you
to open this file I have also transcribed each session. My additional
comments will be inserted into the text from time to time and this will be
in red text. Buchanan refers to each photo by
the number of that photo which can be followed by each photo above the
captioned text. The .mp3 audio file will appear
in blue text.MP3. Click on this
text and you can then follow along with the audio file by scrolling downward
through the photos as they are narrated one by one in the voice of the
Photo 36 001.
We have the one year arbor with the two cam blanks fitted for the
sunrise/sunset shutters. In the upper portion of the cam you can see the two
levers resting against the peripheral edges of the cam blanks.
Photo 36 003. We have another view of the two cam blanks and the
operating levers resting on the upper edges.
Photo 36 004.
Here we are marking out the seventy three spaces on the cam blank prior to
transferring our test measurements. Just below the steel ruler between the
‘2’ and the
‘3’ we have a blade inserted into a slot against which we index the
seventy three teeth of the one year wheel. This gives me my seventy three
divisions. You’ll notice that the scribe lines are off-center to the center
arbor. This is because the two black levers that read the cams are actually
mounted off center to the center line of the arbor. If you refer back to the
earlier photographs you’ll see that drop down pillar and each operating cam
is located on an arbor projecting from each side of the drop down pillar. So
I’ve put a bush on the center pivot of this cam to give us the correct
spacing. The line is straight because the arc of the lever is at right
angles to the axis of the one year cam; whereas the arc of the equation of
time cam, the pivot was parallel to the one year arbor, which meant that we
had to have a curved line to give the correct positioning of our
Photo 36 005. We have the fully marked up cam. What I’ve done is
I’ve marked out every alternative tooth so we’ve marked them out one, three,
five, seven, nine and so on all the way around it just gives us a little
clearer spacing. On our error correction chart we still will do the full
seventy three readings to check our errors.
Photo 36 006.
Here we have the beginning of the error correction chart; we have three sets
of columns to fit in all our readings. When we go to the left hand side of
the page the first column is the tooth number of the one year feed wheel or
the star wheel with the fancy shaped spokes numbered one through seventy
three. The second column gives us the actual time of sunrise for each one of
those seventy three positions. Then the third, narrow column which at this
stage is empty is my ‘test one’ column. I’ll show you in the following
photographs how we arrive at the readings for the test one.
Photo 36 007. Here we have a
caliper lifting the sunrise lever. The bottom has a hook which attaches to
the one year arbor, and we adjust the thumbwheel to open or close the
caliper until we have the correct reading on the sunrise shutter for any
particular day. Once we have this reading correctly, I’ll then take the
caliper and transfer this measurement onto our cam blank.
Photo 36 012.
Here we have the sunrise/sunset dial removed from the mechanism. When
attempting to set the shutters to mark out the cams I found that the dial
was virtually illegible so we have proceeded to clean it up.
Doesn’t this silhouette look a bit
like a bear cub’s face? The second photo shows this dial cleaned up.
Photo 36 014.
Here we have the caliper with the hook end around the center arbor of the
cam and the upper notch section which holds the sunrise/sunset lever is now
on scribe line ‘1’ and we make a
scribe mark on the number ‘1’
Photo 36 015. One can just the
scribe line across the number ‘1’
Photo 36 016.
Here we have a slightly clearer view of the first scribe mark on the sunrise
cam blank. Pouvillon-37-000.MP3.
Photo 37 001. We have a cam blank for the sunset shutter dial. You’ll
notice that we have a series of radial scribe lines offset from the center
to take into account the off center mounting of the reading lever. You’ll
notice we’ve also numbered each radial line according to each step of the
Photo 37 002.
We have the cam cut out initially. The station numbers are every second
scribe line and the outer ring numbers are the number of minutes error on
the actual dial once the cam is installed.
Photo 37 003 is the cam blank and
also our error chart. If you go to the left hand side of the first page
you’ll see in the first column we have the year wheel tooth number; in other
words the number of steps to rotate the cam around one year. In the second
column the target value is what we want the hand to read on the dial for
each station. So on station one it should be 7 hours and 44 minutes. When we
actually performed test one in the third column, you see that it was 7 hours
and 30 minutes which means we have an error of approximately 14 minutes.
This is then noted on the cam blank and we remove metal according to the
error. And then we install the cam blank into the clock a second time and we
test again. So if you look at our first column it should read 744. It read
730 we have an error of fourteen in test one. In test two it has been
reduced to twelve. This continues until we are within our desired tolerance.
Photo 37 004.
We’ve now proceeded to test five using the same method.
Photo 37 005. We can see we now have certain sections of the cam
that are up to accuracy and a small error remaining on certain portions
which we continue to remove.
Photo 37 006
is the final is the final shape of the cam and you can see the original
blank size to get some idea of the amount of metal removed in the filing and
testing. Photo 37 007 is the
sunset cam and the equation of time cam.
Photo 37 008
is the completed cam and the seven tests which if you take 73 stations times
seven tests, it means on each cam we have performed 511 test files until
we’ve gotten the desired accuracy.
Remember that this exercise was done
for four of the five cams in the annular cam pack.
Photo 40 001
shows a base pillar of the main frame of the clock with a cutout to clear
the new count wheel or the count wheel, I believe, was fitted after the
original pins mounted in the
face of the strike wheel.
Photo 40 002.
Another photograph showing the cutout in the splayed base of the rear main
clock pillar to clear the count wheel. And, of course, in the foreground we
see the blue pins
that are now unused on the strike great wheel.
Photo 40 003.
We’re trying to show here provision made in the pillar,
actually a horizontal beam,
going across the base of the clock to clear the pins projecting from the face of the
strike great wheel.
horizontal support beam is made from a different material from the rest of
the steel used in the Phase One clock frame. It looks, in fact, to be plated
and is identical to the material used in the small pillar located behind the
fly fan for which we have found no function for at this time. see Photo 47
010. As found, someone, whose crude work was obviously not Pouvillon’s, had
tried to use that pillar to trip the Easter calculator on a daily basis from
the strike count wheel. The entire design was wrong, as the calculator is
tripped annually. Pouvillon had added these two plated components to the
clock at a much later date.
Photo 40 004.
We need to look carefully at the darker portion on the right hand side of
the picture just below the notch in the count wheel where you can see a
reflection in a frame showing an undrilled pivot hole, or a mark on the
frame corresponds to the front frame where Mr. Pouvillon has placed a pivot
and on the rear frame it has remained undrilled because it has never been
Photo 40 005.
Here we can see a better view of the center punch mark in the frame member
right in the center of the photograph.
Photo 40 006. Another pivot point
that is undrilled on the rear frame. I believe the front and rear frame were
made in the traditional manner where both frames were clamped directly onto
each other and marked out and cut out. But, of course, this point on the
rear frame has no pivot so it was left undrilled, only marked.
In blue circle.
The first photo
shows the length of day/length of night cam. Since any 24 hour period can
only be divided between the day and night, these two dials can be read off
this single cam.
The next photo shows one of the cams controlling the sun rise/sunset
shutters on the log book page used to test and cut the cam to its correct
profile. Unlike the length of day/length of night function, the
shutters each need a separate cam for the sunrise and the sunset function.
See video below.
The third photo is the second cam for the shutters.
photo shows the three cams mounted on the one year step arbor. The equation
cam is hidden by the larger, first shutter cam. One can see the three levers
acting upon the surfaces of these three cams to perform the length of
day/length of night dial function as well as the raising and lowering of the
Next the kidney shaped cam is being
crossed out on Buchanan’s scroll saw. A similar procedure was done on the
remaining cams and the result is shown in the next photo.
The four photos above show the beauty of the cam pack.
It holds a central position within the clock and the fact that it is 900
to the rest of the movement only accentuates this part’s importance. We have
tried to reverse-engineer this assembly as close as we could to what
Pouvillon would have done. It completely fits into all existing holes and
spaces without any additional alteration to the original artifact. Given its
visual appeal one could see how this part ‘went missing’ sometime in the
past; perhaps when the clock was in pieces at the charity shop,
see biography at the beginning of this
Photo 42 010,
last photo above,
is a photograph showing the cam pack assembled into the clock we still have
to fit the snail, finally, that will trip the Easter calculator.
Photo 42 012.
Another view of the cam pack and the operating levers which you can see
resting on the top of the cams. Photo
42 013. Another view if the cam pack fitted to the clock.
Shown here are the diagrammatical sketches of the lunar
dial and the Planisphere dial. We will see in latter documentation that
these two complications had to be devised at the same time as their feeds
Photo 44 001.
Here is a view of the feed that drives the main orrery. Of interest to note
is that the bevel gear fitted to the second arbor driven off the strike
great wheel is only pinned to the arbor, there’s no flange on the arbor for
it to but up against as in all of the parts of the clock pertaining to
the actual time or strike train,
as was done in all of the Phase One
construction, see blue arrow. It’s meshing with
a second bevel and a vertical arbor mounted on a cock. This cock is
interesting because it supports the planisphere dial. We can also see the
pinion just above the vertical drive bevel that actually drives the
planisphere. This pinion is separate to the actual bevel gear and I am
inclined to think it is a later addition,
see white arrow and Photo 45 002, below.
Photo 44 002 shows us the
universal supported on a cock on the second frame pillar and this is to
provide a ‘kink’ in the vertical arbor to clear one of the time train
arbors. The cock that supports the arbor just below the universal joint also
carried the lunar dial. Also
note the circled area where this cock is joined to the Phase One cross frame
pillar. It is done by having a flat filed onto the frame pillar with a
simple rectangular metal piece fitted. There is no effort to blend this
piece in as would have been the case if this piece was conceived as a whole
in the Phase One design.
Photo 44 003
is another view of the vertical arbor and the bend introduced by Mr.
Pouvillon to bypass the time train arbor. I think that had he planned this
addition initially, he would have made a slight change in the size or
position of this time train arbor or also changed the position of the strike
arbor where the bevel gear drive is to prevent a misalignment or an
obstruction that is causing him to use a universal in this drive.
Photo 44 004. Another view of the
lower orrery drive bevels not shown.
Photo 44 005. A picture of the universal joint halfway up the orrery
Some more conclusions as to
sequence of construction from the restorer. All these photos evidence
the vertical drive to the orrery. As one can see the two cocks that support
the vertical arbor are the minimum required to fulfil their function. Both
cocks carry a later complication. A look at the orrery platform will show
that this has no provisions for latter projections that support the mystery
dial. In other words after the main movement, (Phase One), the orrery was next. No other
complications could have existed without it and all are attached in some way
or driven from it. When the clock is finally restored I think we will do a
sequence of assembly photos to prove this point.
Pouvillon-45-000.MP3. Photo 45 001.
We have the planisphere
mechanism mounted on the cock supporting the lower end of the bevel drive to
Photo 45 002. We have a view from the opposite side of the
planisphere mechanism. We can see the pinion on the vertical arbor driving a
gear wheel on a second arbor. You can also see that it is carried by a cock
screwed to the cock mount carrying the bevel gears. And, of course, this
arbor carries the drive up to the lunar dial.
Photo 45 003. Here we have the
drive consisting of a train of three small gears to the lunar dial. It’s
also mounted on a separate, attached sub-cock and is part of the lunar drive
train. If we remove the lunar drive train then there is no support for the
upper end of this arbor, the second vertical arbor, and the planisphere
won’t operate unless one mounted a single cock on the upper lunar dial
support to carry the upper end of this arbor. It’s also a question as to why
this arbor is so long. I think that both the planisphere dial and the lunar
dial were constructed simultaneously. Otherwise there would have been parts
discarded if one complication was made individually before the other.
A few more conclusions as to
sequence of construction from the restorer.
The planisphere, from the construction point of view seems to be
interrelated with the lunar dial. In the first two photos you will see the
24 hour arbor is pivoted, at the lower end, in the planisphere frame and in
the third photo, the upper end of the 24 hour arbor, in the lunar frame
cock. Now which came first?
The 24 hour arbor is required for both dials. There was either a temporary
cock for the top or bottom pivot or Pouvillon built them simultaneously.
This is the first time this question has presented itself. So Pouvillon is
able to integrate two complications as we have said before, but this is the
first time we have seen it done between complications. Up until now each
stage of construction is a stand-alone unit. If you remove the planisphere
the lunar dial will not function and if the lunar dial is removed the
planisphere will not function, although a simple cock would solve the
problem in both cases.
The kinked universal drive to the orrery is, I think, a clear indication
that he had not planned to fit the orrery originally.
This means that the orrery was not a
planned part of the main clock frame containing the time and strike trains,
what we call the Phase One sequence of the clock’s construction. The
comports with our earlier observations that Pouvillon first created the
steel-framed two train clock as a stand-alone unit before ever conceiving of
using it to create his astronomical clock.