Ed Whetstone - Kinetic Sculpture Project

(... several months later ...)

Independent Project: Kinetic Sculpture

Developing the Idea:

In my second semester for this course, I set out with a fairly ambitious goal - to design and build a pendulum clock or other gear-driven kinetic sculpture.

Clocks:

(image via wikimedia commons)

Clockwork art has a long and storied history, starting with the invention of of the first escapement gears as early as the 1050s in China.  An escapement is the mechanism which allows a power source - water, springs, motors, etc - to be divided into discrete units of motion in time.  This is the tell-tale tick-tock of a pendulum clock, which was the absolute pinnacle of timekeeping technology for two centuries.

A power source becomes motion becomes our measurement of time -- a fascinating idea with a lot of potential for artistic exploration.

Through gear trains, the speed of motion can be sped up or slowed down for different timescales, and adding additional power allows for chimes or animated figures -- this is why most grandfather clocks have more than one weight.  One drives the gear train, and the other drives the chimes -- a third might power just the midnight chimes.

By the way, the pendulum only supplies timekeeping, not power.  The gears are turned by the weights alone.

Today, making pendulum clocks tends to be the niche of ambitious woodworkers, but they’re definitely cool.  Here are a few by Clayton Boyer, who sells schematics of his clocks on his website:

clocks by Clayton Boyer

clocks by Clayton Boyer

Pendulum clocks require a lot of precision-cut, flat pieces.  Sounds like the perfect task to apply to a laser cutter.

Kinetic Sculptures

Ultimately, the complexity and precision involved with making a clock (combined with a pretty nasty workload this semester) has caused me to put these designs on the shelf for later.  Instead, I’ve decided to focus on a much simpler, manually-driven kinetic sculpture.

Artistic, sculptural clocks are one small subset of a larger category -- kinetic art.  In its most basic definition, kinetic art is art that moves.  Like clocks, that motion can be driven by a variety of power sources, like wind in many Anthony Howe pieces:

Or, with electric motors, like those by Bob Potts:

In my research on clocks, though, I stumbled onto the work of David C. Roy, who does beautiful spring-powered kinetic wall art.

One piece in particular really caught my attention:

The concept is so simple, but so aesthetically pleasing, that I knew I had to try a version of it myself.

Breaking Down the Problem

My first task was to make several key decisions:

• how do I power the device?
• what are the constituent parts necessary to build the piece?
• how do I weight the pieces so that they maintain their orientation?
• what pieces can I cut on the laser, and what other hardware will I need?
• how can I quickly iterate designs in order to find a pleasing pattern?

To help me with these decisions, I have a few personal goals to keep in mind as well (I’ll address whether or not I met these goals a little later on)

• simplify -  What is the simplest way to achieve this effect?  I can add bells and whistles later once a proof-of-concept is done
• Integrate off-the-shelf hardware - try not to get bogged down in what is possible with the laser -- there are a lot of problems that can be solved with a trip to the hardware store, even if I didn’t make the parts myself
• Iterate - try to develop one idea at a time, and make sure it works before moving on to the next step.  Don’t be afraid to make mistakes.

Next, I need to break down the problem.  Each piece consists of a number of “orbiters” around a central axis, independently weighted so that they maintain their orientation.  Simple enough conceptually!

Iterating in Software

For me, a design challenge like this lends itself to a programmatic approach.

I set to work experimenting with different patterns to try out as orbiters, via Rhino's curve tools and CurveBoolean:

 I wrote a Python tool in Maya which takes meshes (exported from Rhino in most cases) and automatically rigs and animates their rotation around a central axis.  The tool also had the ability to switch out backplate designs and assign shaders for color testing. Most importantly, I can dynamically assign the number of orbits and "layers" to optimize spacing.

Ultimately, the plan was to make a plug-and-playable system for iterating a wide variety of designs for these types of sculptures, and “prove out” each design before manufacture.  Time constraints made this untenable to fully implement, but for the future it would be fun to play with this idea. Unfortunately a laptop crash has completely wiped the tool from existence.

Here is a video of the end result -- several interesting patterns, a variable number of orbiters, and different types of layering.  The first part of the video is a different concept, which I may return to at some point, when I have more experience with gear-driven designs.

Bearing With Me for a Moment…

My design (and David Roy’s, incidentally) relies on gravity to maintain the orientation of the rotating “orbiters” -- this means I need a free-rotating axis at the center of each one, and a free-rotating axis at the center of the wheel.  This means using bearings.

The simplest kind of bearing is just a sleeve that you can put an axle or post through.  Not very exciting but effective in most clocks.  For my needs, though - I needed a solution which would give the orbiters zero “play”.  I did purchase some PFTE sleeve bearings to experiment with, but they were too wide to be useful in this application.

Moving up, you get into two types of ball bearings which might be suitable for my needs - first is a thrust bearing, or “lazy susan” bearing, which is two pieces “sandwiching” metal or nylon balls.  These are ostensibly meant for perpendicular loads, like a lazy susan.  In practice they could be used horizontally, so long as the weight they hold is small enough

The other type is a radial bearing, or wheel bearing.  These bearings are structured so that they hold a radial load, like a wheel, using a layer of balls between the axle and edge.

Both have tradeoffs in terms of functionality:

• radial bearings are going to be harder to integrate, but will spin easier in a horizontal position
• flush bearings will be easier to build with, but aren’t meant to spin with a radial load

I ended up getting a pack of industrial flush bearings and skateboard wheel bearings to try.

Prototype 1:

Considering the easier construction with the thrust bearings, my first task was to test out their feasibility.  I had a flew plexi panels that I wasn’t going to use for this project, so I quickly glued up a mockup to see how they rotated:

Two problems immediately arose - if the plexi got too heavy (three layers of .118 acrylic at around the size I was aiming for) the flush bearings needed a little convincing to rotate well

Problem the second: the bearings come pre-lubricated, and unless you’re very careful about cleaning off all the grease, the glue just peels away a layer of grease rather than sticking to the actual metal of the bearing.

With this in mind, I still thought it was worth the extra ease to attempt using the flush bearings for the first prototype.

these images show the final patterns that the sculpture is meant to oscillate between

the etching lets me quickly line up the pattern layer onto the weighted material, which worked quite well

these small pits allow me to insert a magnet, allowing quick fine-tuning of the weights

Assembly went well, but again I ran into some major problems:

First, I didn’t allow myself enough wiggle room for wobbling. - the heat of the laser has a tendency to warp thin acryclic, and the material isn’t perfectly flat to begin with.  My clearances weren’t nearly enough to allow for that warping.

Second, the glue on the flush bearings lasted exactly five days.  After that, a small tug just tore the pieces apart.

Third, the layered-acrylic approach added uneven weight, making maintaining orientation extremely difficult.

In theory, I could add weight in a fine-grained way with my magnets.  In practice, the first prototype was too much of a mess to be worth experimenting with. I still like the idea of this modular approach to the problem, though.

Prototypes 2 and 3:

The second prototype was mostly a cannibalizing of the first for parts to experiment with.

I came up with two solutions that actually met my earlier goals pretty well:

• replace the pattern layer with paint -- it weighs a lot less than a layer of acrylic
• use hardware in place of layered-up pieces of acrylic for posts
• wheel bearings in place of thrust bearings

Final Build:

using the acrylic masking as a paint mask worked remarkably well

after painting

peeling the remaining acrylic masking

assembly went well until the humidity caused the glue to fail - upon re-gluing many bearings were badly off-axis.

using adjustable hardware made aligning and leveling the posts much easier and more precise

the final assembled product!

Conclusions:

Basically, this project was a very educational failure.  The original concept just never coalesced, and there were still a lot of blind spots in my design and construction of the final piece (such as - no way to mount it on a wall yet!)

It almost works.  I ended up having to re-glue several of the wheel bearings, and they got mis-aligned.  If you look at the final product, you can see a distinct tilt to several of the orbits.  Luckily, the new approach to construction allows for a lot more clearance in this case, but it still doesn’t spin correctly.

In the end, this was a cool idea that just didn’t quite pan out.

Lessons learned

• Start as simple as possible when you’re out of your element - I spent a ton of time early in the semester researching and learning about clocks and clockwork, which ultimately was just too big of a problem for me to take on, given my lack of any real “making” experience, and significant lack of time for experimentation.
• if you’re working with expensive materials, mock things up in cardboard or MDF - I could have saved a ton of time and money if I’d been able to test things out without having to use the acrylic.  Of course, time was my major consideration, but honestly the first prototype should have been a cardboard mockup to test spacing and clearances.
• sometimes, doing things the hard way is the right way - This design really should be gear-driven, with a spring or electric motor.  It makes a lot more sense to force the orientation of the orbiters rather than letting gravity do the work.  It would be cleaner, more flexible, and ultimately simpler to assemble.  There’s a lot more up-front investment in expertise and design, though.
• using etching on acrylic to make masking areas is really great - If anything, this was the best takeaway from this whole process, and I plan to make significant use of it in the future. It's like custom paint-by-numbers on plastic. Wonderful!