Thursday, June 11, 2015

Fusion Cooking

So I'm officially behind on the Raygun. My target was to start cutting metal this week. Work showed up instead, good work -- did one morning gig hanging lights (actually, spent most of the time driving the scissor lift around) for a clean $100 in cash.

From the money from the most recent grenade order I dropped fifty on a new tool holder from Grizzly. It looks good, all solid (and very oily!) metal, but I've experienced some unusually soft steel before. Such as the bench vise I got from OSH that has disabused me from ever buying a vise from that store again.

But even taking all this into account the CAD is going terribly slow. Part of that is learning the software. The greater part is having to solve so many design issues. Metal is a bit unforgiving. I can make a few adjustments for poor fit-up or other errors, but it is slow work correcting mistakes made in metal. Metal is hard, slow to shape and requiring power tools to do so. And expensive. I really need to anticipate as many issues as possible and solve them before I start cutting metal.

It is engineering. It is a balancing act. Trying to balance aesthetics, strength, and manufacturability. Out of the many things you have to do to engineer something properly is to compensate for various sources of error. You design in steps that will catch and correct (such as my current M40 lathing scheme, which holds 4/1000 in reserve for a finishing pass; this picks up and corrects various places where tools like the knurl or the parting tool deform the grenade, throwing up ridges of excess metal behind them).

A subtler one is how I run a de-burr tool around the edge of the hole before inserting the button; this produces an expansion space for the metal to flow into when I crush the plug in with the press. In previous grenades, this edge was forced into the button pin, pinching it.

The other major tool in dealing with the inevitable errors in measurement, tool wobble, manufacturing variance, etc., is to not design around perfection in the first place. I really had this demonstrated for me during discussions of the Apollo Program and the absurd suggestion that "the landings" were faked in a film studio. One of the claims the believers in a hoax make is that the necessary positional accuracy of the original launch is impossible to achieve. The slightest fraction of a degree off at launch, they argue, and the spacecraft would be multiple kilometers off target by the time it reaches the Moon.

Which is why the NASA engineers -- like any engineers -- did no such thing. They simply pointed the spacecraft "good enough" to get within the circle of error of the next designed-in correction stage. One equivalent in design for assembly is you use a slot or an over-sized hole on the upper piece, so the bolt will still go in and hold everything properly even if there is some slop in machining or alignment. You don't design so the holes must be perfectly aligned...unless there are other design goals requiring it.

But this has to be of course designed in. Even at the simplest level, design for manufacture means you have to solve endless "duck, wolf, bag of grain" problems, where part C is bolted over the bolt holding part A to part B -- fixing the assembly order and making it impossible to place bolt holes for C on the inside of A.

My raygun nozzle is currently there. The reflector dish is CNC'd out of plate stock, top and bottom pass (which themselves brings up the question of how good my alignment will be). On the final pass I'll machine a hole all the way through, and push a lathed spindle through that. The acrylic nozzle goes inside that spindle and all the way into the gun to reach the Cree that illuminates it, and ends up supporting the front assembly.

After working on it off and on for over a week, and covering several pages of graph paper with mechanical sketches, I've figured out the connections and assembly order. The spindle is one-way (it has a lip) and will have a tapped hole in the side behind the dish which holds a screw pressing into the acrylic. The dish is pushed against the spindle from the other side by the pressure of the 3d-printed insulator rings -- which also are floating freely on the rod -- and the whole Dagwood Sandwich is pinned together by a second screw at the gun end. Which, since the front end of the gun tapers, is on the outside of the shell, and since the insulator rings are there, is accessed for tightening by a hole in the insulator rings that you can fit an allen key through.

There are a lot of alternatives I had to reject, and a lot of secondary considerations. Could machine the spindle as part of the dish but it would look better lathed. And take a lot more metal to machine from a single block of material. The dish can't be lathed (wrong kind of curves) but it can be lathe-polished. Which means the assembly is designed so I can put both parts together and chuck that in the lathe before removing a metal central rod and putting the acrylic in. In addition, any error in the length of any of these parts is adjusted for at the gun end, where the final retainer is.

And, yes, the acrylic is a failure point. But if I added metal -- well, extending the spindle to the gun is possible but the thickness is determined by the aesthetics at the nozzle end, and still creates a potential weak point. Inserting a stainless steel tube strengthens it, but requires the acrylic nozzle have a step to cover the diameter change, and that becomes the point of failure. This way, replacing the nozzle in case of breakage is less onerous, and I'll simply have to put up with the fact that the reflector dish provides a good lever to put stress on it.

(And, yes, I could switch to polycarb, but it is tougher to machine and although stronger, more likely to leave dangerous edges when it fails under stress.)

I've more-or-less solved the front assembly, how the handle fits together, the fin assembly, the trigger guard. But I still have to finish designing the trigger assembly, the access plate/removeable power cell that will allow uncoiling the USB cable to recharge the internal LiPo, how to get the LED wired into the fin...and I haven't even touched the side panels with their switch, indicator lights, and swoosh.

And very few of these assembly details are reflected in the CAD as of yet -- which I fully expect is going to reveal other potential problems in fit-up and assembly, as well as mechanical weaknesses and structures which are un-machinable as they stand. (I'm not about to get into 4-axis CNC at this point, so I basically have to make shapes that can be attacked from two sides at a maximum, plus traditional machining -- milling and drilling -- at off-axis to these. And, yes; every threaded hole I add is a manual operation; hand-tapping.)

Oh and yeah. My cheap eats these days is a three-day curry; 1 "can" Coconut Milk (I like the "school-lunch milk" boxes from Kara; very rich and fatty), red curry paste (May Ploy brand; comes in a tub like miso paste), tofu and veggies to suit (I get locally-made extra-firm tofu sold bulk-style at Berkeley Bowl). Often Dynasty sliced water chestnuts, cut baby corn -- but I've also used carrots, broccoli florets, canned bamboo shoots, and fresh snow peas. Plus a can of Albacore tuna if I can afford it. Makes three servings...I brew up 1-2 cups of jasmine rice in the rice cooker and re-heat the curry for a meal. Goes nice with a rooibos chai (also bought in the bulk produce section).

No comments:

Post a Comment