test #3 above: 1 inch/minute, letters 7/16 inch tall, kerf < 0.024 inch

First, this 3 axis CNC gantry system was built on a lark, out of cardboard and mostly salvaged parts. Building it kept me sane through late December - though how sane is it to build machine tools out of cardboard? It functions; which is something short of saying it works. It's fun as a pen plotter, but not well suited to the kind of precision positioning this Solar CNC project needs. Still, it's good enough for these first trials, and getting some real use out of it is redeeming an otherwise strange and wasted December.

Mounted to the z-axis is "the cutting tool", which is the real focus of this project. It's a poor-man's laser cutter. If you understand how lasers work, you might rightfully object to associating laser beams with sun beams. I only mean to compare the burning and cutting results. Whether the photons are in phase or not doesn't matter so much when you're broke. A <$50 solar cutting head on a <$300 2- or 3D motion system is what we're after here - a humble, limited cutter that falls somewhere between a hand-wielded razor blade and a $20,000 VersaLASER VLS 6.60.

The familiar part of the cutting head is a lens to focus sunlight down to the smallest burning point we can achieve. I started with an old macro camera lens, which was about 27mm in diameter. This worked well sometimes. The great thing about that kind of lens is that it has multiple glass elements and optical coatings to produce sharp non-distorted photos, which also means it's good for producing a very small burning spot at it's focal point. It also had an aperture adjustment I might have been able to use as a burning power adjustment. But I wanted more power.

I ordered a couple bare glass lenses from Anchor Optics. In the video you see a single glass element spherical lens (coated plano-convex), which is simple and cheap, but not ideal for focusing to a small, sharp burning spot. Still, it worked. For a few dollars more, I also ordered a multiple-element lens (coated aspherical achromat), seen in the photo here. It works like the camera lens to make a very small spot size, but with a larger 52mm diameter it's putting more light into that spot, to good effect.

To get the sunlight to the lens, in a consistent downward orientation, we need the contraption above the lens. It's called a coelostat. Wish I could tell you with confidence how to pronounce that, but it's one of those latin-derived words that would sound one way if you tried to say it in latin, and another way if you apply modern conventions for pronouncing latin derived-words. I found a phonetic spelling that amounted to [see-low-stat]. "Coelo" refers to the sky, or the heavens. Look it up on Google and you'll find "Toto Coelo" an '80s band whose name would mean something like "all the heavens". They're the ones responsible for the unforgettable classic "I Eat Cannibals" - a song I remember hearing on the radio (or was it the roller rink) when it was fresh. As a kid, I didn't quite get the rest of the lyrics, but the concept of eating cannibals posed a perplexing problem of classification, which I innocently thought was part of the reason one would enjoy the song. I mean, if you eat cannibals, you're a step above a cannibal, you're higher on the food chain, yet cannibals are people too, which makes you an ordinary cannibal if you eat them. I'm fairly certain that the then-facinating problem of whether a person who eats people who eat people is the same thing as a person who eats people without prejudice contributed to my later affinity for logic and linguistics. That's what you get on the Web: Google fact checks that turn into freaky-odd tangents. Some neologician is gonna name it webindipity. (Googled that - no results yet, thankfully. Forget it quick - this whole paragraph.)

Back to the coelostat: According to Duane Johnson at redrok.com, who makes solar tracker electronics using ordinary LED's as light sensors, I've designed a type 3 coelostat. I worked on other designs, but one night I looked at this photo of the Swedish 1-m Solar Telescope in the Canary Islands, and realized how simple that arrangement would be to reproduce in cardboard.

My design still needs some hobby servos, or worm gear drives, to engage the geared rings visible in the photo. Adjusting it by hand lets you cut for a couple minutes or so before you need to stop and adjust it again. Once it's motorized, the next step would be to install some LEDs (as light sensors, ala redrok.org) along the edges of the light path, and probably elsewhere, along with some other circuitry and an Arduino, to get an intelligent tracking system working. After that, it would be even better to replace the Arduino with a cheaper logic circuit.

Let's pause the project here, and consider some context. I just wanna surf. This grand Solar CNC vision (which is grander than what this page suggests) is really all about making surfboards - mostly for myself. Consider my motives and this comes off as total self-absorbed thumb twiddling.

This thing should already exist - I should be able pick one up for $300 at the toy store - there next to the chemistry sets and cheap telescopes. Any little boy knows you can fry ants* with sunlight and a magnifying glass. A CNC machine built around that same primal urge is an obvious next step. Grownups have laser cutters to make plaques and trophies. Schoolyard kids swipe the the teacher's magnifying glass and burn bugs at recess. Somewhere in between we should have this affordable intermediate tool, and do something worthwhile with it.

Maybe it's just that previous attempts produced inadequate and unpublished results. If that's to be the fate of this project, at least we'll have this page to document that.

But I think it might work.

If it does work, we'll have the equivalent of a low-Watt, large-format laser cutter, suitable for cutting cardboard, paper, and maybe some other thin materials, like plastics or textiles, for roughly the cost of a RepRap - maybe less. Like the RepRap, it's an inferior - but not too inferior - version of the "real thing", at a price that makes the small compromises in functionality and resolution well worth it.

Initial test results:

Speed is the main issue. The kerf is small - in the range of 0.01 inch to 0.025 inch, or 0.25mm to 0.6mm - and the cut edges are clean enough for my purposes. Cutting all the way through 4mm thick corrugated might take more than one pass, or an up-down sawing motion - assuming I can get enough of the burning light cone into the slot of the first layer cut. Air assist might help.

Maintaining focus on the intended cutting plane is crucial. In this test, much of the uncut area would have likely been cut at the tested speeds, if only the machine were rigid and square.

The long photo to the right records an attempted speed test, from 0.2 to 4.0 inches per minute, cutting 0.5 inch spans at selected speeds in that range. The test run spans two scraps of corrugated cardboard.

In the circle cutting test in the video, I was able to cut through the top face of the cardboard at 0.7 ipm. In the speed test here, there's evidence that 1.2 ipm will work. You'll see in the photo to the right that after the initial line tapers off and disappears, it reappears twice and fades. Those were points where I paused the test run and manually readjusted the focus. Where I did this near the start of the 1.2 ipm test span, the first layer is again cut through for a short span. It seems the cut tapers off because the system goes out of focus, since the taper happens within the constant 1.2 ipm span. I attempted the same focus adjustment within the 1.6 ipm span, with less success.

So after two tests on a rickety setup, the known top-face-cutting maximum speed is up to 1.2 inches per minute, with indications that it can go higher, if only the focus can be maintained.

At five or ten inches per minute, this would start to be a useful machine. I'm hoping to get it to 20 ipm or more. I should be calculating, estimating, and projecting, instead of hoping. So here's a first sketch of an estimate:

The heat needed to ignite paper, divided by the heat passing into the burning spot in one second, gives seconds to ignite (hopefully cut through) paper under a stationary burning spot. Multiply that by the number of adjacent spots that would need to be burned to make a one inch line, which gives the total seconds to burn through, or cut, a 1 inch line, as a series of burned-through spots. That's seconds per inch, which can be inverted to inches per second, and converted to the familiar inches per minute figure we want.

The key part of the estimate is the minimum time it takes to burn, or cut, through the first layer of corrugated cardboard under the minimum-diameter burning spot. Why only the the first layer? Because cutting through the fluted layer is a project in itself. Besides, the top face layer represents normal paper media in general, and maybe some textiles. Here's the math so far for time to burn one spot:

A little research and some guesses, and we get some numbers to play with.

At that rate of burning spots, you could burn a 61.6 inch line of spots in one minute. An inch or so per second? Improbable with the current setup. But not too wildly improbable. What did I leave out?

This too-simple estimate ignores diffusion of heat to surrounding paper and air. And the insolation figure needs to be adjusted down probably, to account for only that part of the solar radiation that is able to pass the optics, and that contributes significantly to heating cellulosic media.

The second beer and I are not quite up to the calculus at the moment, but the relevant Wikipedia pages look like they will make sense later, so the estimate formula will be refined.

But it's likely I'll just go push the envelope with the machine as it is first, gunning for a new record, pushing for 1.5 ipm.

See estimate notes if you want to try to evaluate how far off my figures are so far.

* Please do not fry ants, except in self defense