Alexandre Pouchkine by François Junod
A Superb Contemporary Handwriting Automaton
This is not a new video, but this piece by François Junod is probably the most stunningly beautiful and complex automaton mechanism I have ever seen, contemporary or antique. In the context of what can be seen in this video it appears that there were no shortcuts, no compromises, and no cost-reductions – this is mechanical design done right.
But functionality was clearly not the only goal. The level of workmanship and finish detail on every piece part is unsurpassed and elevate this to a true work of art. I have been impressed by both the automaton and the video presentation since I first viewed it a couple of years ago, and I see a little more every time I watch it.
I thought it might be of interest to make some notes about how the mechanism appears to operate. I am no expert on anything, and I will make a disclaimer here by saying everything written below is my opinion or best guess and may not be accurate.
Points of Interest – Timeline
NOTE: Some sections of the video are only a fraction of a second in length. It may be helpful to open the video in a second browser if you are trying to follow along with the text.
0:10
It becomes apparent later in the video that this is part of a mechanism that moves the paper platen to a new position, allowing the next word or drawing to be written and stay within the limited range of arm movement.
0:15:24
This is a view of the stack of brass cam disks that rotates and drives polished steel followers, providing the required three degrees of pen motion. The cam stack also moves to different positions vertically, positioning different sectors under the cam followers to write different words or drawings.
It appears that the two cylindrical brass assemblies mounted atop the larger steel end plate each have a pair of small radial ball bearings. Two shafts pass through these bearing assemblies and provide the force to rotate the cam stack. The bearings allow the cam stack to move vertically on these shafts with very little friction for what is probably a fairly high rotational torque. The dual bearings allow a zero backlash rotation of the cam stack so that there is no chance of the spring tension on cam followers allowing the stack to jump ahead slightly under its own accord.
0:19:24
Here is a similar linear bearing assembly made from three equally spaced radial bearings. These appear to be used on each end of the cam stack for the central shaft, providing concentric rotation while still allowing vertical movement.
0:20
A beautifully made flyweight governor with a driven helical screw.
0:25
This is a view of two cam followers (steel) riding on two related disks in the stack. The three independent writing motions (elbow rotation, arm fore-aft, and pen lift) for a given message or drawing would require at least three independent cams or sectors of cams, all accessed simultaneously by the follower so that all motions are properly in sync.
It appears that this stack has all of the cams of a particular type of motion (pen lift for example) stacked adjacent to one another, and divided into three groups. The followers are then spaced to contact the respective cam in each group.
For example, it appears that the lower 19 cams in the stack are similar, and from the appearance are possibly used for the pen lift which is a more of a binary motion. The next group of cams above have edges that would indicate one of the analog motions, possibly the left-right arm movement. The top third of the stack appear to be linked to the fore-aft arm movement.
0:25:30
You have to be quick to catch this clip. It appears to be a drawing that gives some insight as to the function of the top cams. The drawing shows:
- cams 0-6 assigned to right drawing and left drawing
- cam 7 is labeled signature
- cams 8-10 are labeled verb
- cams 11-13 are labeled adverb
and I can’t read the rest. I don’t know how messages are constructed.
0:43:19
I’m not sure about this, but the spinning star wheel stopped by the pawl is very similar to the reel-stopping mechanism used in mechanical slot machines. This might be used for random selection of the message or drawing to be written.
0:46:30
Here is a short glimpse of the rear linkages that move the arm.
0:59:30
Another quick view of the writing motions at the back of the arm.
It appears that the fore-aft movement of the arm is supported by a linear bearing arrangement – the half-cylinder at the back of the arm appears to have two ball races on either side.
The thicker steel (stainless or nickel-plated perhaps) lever above this section has horizontal and vertical rollers that apparently drive the fore-aft arm movement while allowing left-right movement, which is driven from below by a similar dual-roller arrangement. The lowest roller appears to be part of the lower steel lever (the one with the visible engine-turned pattern) and drives the up-down movement of the arm.
This dual roller arrangement no doubt has very low friction, and if it is used with eccentric shafts through the rollers it could conceivably be tuned to have zero backlash.
1:16
Close-up of the pen-lift cams and follower.
1:18
Very close shot of arm mechanism moving fore-aft.
1:23
A short clip but a good view of all arm motions. The cam stack rotates faster than I would have guessed, which may mean that the cams probably store words or multiple short words, and not entire messages.
1:25
The cam stack and the flyweight governor coming to a halt, probably just prior to a vertical shift of the cams to write a new word or drawing.
1:33
The cam stack shifts vertically to access a new word or drawing. Note the aligned clearance notches in the cams that allow vertical movement past the cam follower.
1:40
This is another view of arm movement, and for the first time shows that the paper platen is moving as well between words. Precise pen movements and large pen movements do not go together; this design appears to have the arm range optimized for precision and limited to small movements that are within the range attainable with this size and shape of cam and follower. It appears that the paper shifts to be able to write the next word while staying within this limited arm range.
More of the astounding works by François Junod may be found here.
