Friday, April 10, 2015

Raygun XIII

I think why "sleeping on a project" works is that it is a sorting device.

At some point in developing a project you will encounter problems with multiple variables. Too many variables to be able to hold in mind at the same time and thus be able to weigh against each other properly.

Thing is, every time you put down a project to work on something else for a while, when you pick it up again, you get a semi-randomized sort of those multiple variables. Eventually, you'll pick up a set that can be solved -- and then you simply adjust the variables that got left out until they, too, work out.

Well, I'm not there on the Raygun yet.

I got client approval on the mock-up. Furthermore, she likes the idea of surface details like screws and bolts. So I can simplify the problem of fastening the body parts together.

What is hanging me up now is trying to balance assembly complexity, machining time, size of stock, alignment issues, access issues, strength of final assembly, maintenance access, and...worst of to move from "hero" model to cheaper duplicates.

In the simplest terms, the more parts, the less waste metal and the cheaper the stock. Also, the entire thing is potentially too large to be cut out in one machine session. This gets subtler; smoothing and polishing will be a lot easier if the larger planes can be separated; if the side detailing is a separate part, for instance, it is easier to smooth the side under it. Same for the radar dish.

But the more parts, the more fasteners that have to be planned, screw holes to tap. And it also begins to involve more machining steps, to cut the various slots and holes and other elements. And all of that takes away from the savings in time and materials.

And they aren't obvious trade-offs. Take the fin. If I split the fin from the jet tube, then it can be made from a thinner piece of stock -- but then I have to solve how to connect it to the tube. If it is made in one part, then I have to align the material twice in order to cut front and back, and I need a really long drill bit to make the hole for the wires. In two pieces, alignment is easy and I can mill a slot before assembly...but then I need ways to fasten it together.

And some of the parts just aren't working yet. The barrel is a big problem right now. The acrylic rod needs to be nearly full-width all the way back the the LED to act as a light pipe. But I'm not sure it is strong enough to hold the front end elements. So there needs to be metal involved. But because of the geometry, that probably means hand-cutting some pretty large threads. Which would be easier on aluminium, but because it is probably a hollow tube, steel might be required for strength...

Another hangup is that everyone is out of stock of the surface mount transducers I want to try. But worse...even if I had the transducers on hand, how can I simulate the acoustic behavior of the aluminium shell without having, well, the main body cut out already?

Well, at least the lightpipe seems to be workable:

And the trouble is, enough of this project is still guesswork. And these are not good processes for iterative corrections. I have to get most of it right on the first try.

And I just don't have the comfort level yet in all of these materials and processes. I remember way back at TheaterWorks, when we were trying to problem-solve while loading in a set, we'd sometimes go back to the shop to stare at a piece of stock lumber. That would let us engage the gut judgement that comes from having built a whole lot of scenery (some which failed, and thus helped us learn the edges of the envelop). I don't have that yet for CNC.

It's odd. I can know there are unknown unknowns. I can even estimate their size, based on years of having, in fact, had to deal with problems I lacked the skills to anticipate. My estimate is that the unknown unknowns on this project won't cause any critical setbacks, and can simply be rolled into the known unknowns.

But right now, I am far too aware of trying to balance out pros and cons when I simply don't have the background to parameterize either properly. And that makes it very hard to proceed with confidence on my assembly plan.

No comments:

Post a Comment