Today was a wonderful day filled with little round things.

I’ve begun working on Ãœberclocker’s minor parts as I await another run on the waterjet (and more aluminum) to finish the fr0k. This is slightly more work than I anticipated, and with my days taken up by both the Media Lab and the MITERS-alumni-generated giant-solar-mirror-squeegeebot-building startup, free bot time is actually getting fewer and farther between. There’s nothing meaningful at that website yet. However, should you need giant squeegeebots for cleaning your multi-square-metre shiny, smooth surfaces later on, please give us a ring.

Here’s a completed “front leg” assembly, composed of two waterjet-cut pieces and some standoffs. I have become a fan of the “2D assembled shapes” building method since most of the work is done by the waterjet cutter. I only had to countersink the side plates (and a good 5-axis waterjet can even do that o_O)

With the new tooling setup on the lathe, I can make a simple standoff in under 30 seconds. A screw machine can do this in under 3 seconds, but I’m also not making 10,000 standoffs in a single breath.

The corners on this will probably be chamfered inwards later on in order to allow traversing of the Robot Battles stage. The plastic roller is 1/8″ smaller  than I intended since a 7/8″ UHMW round was sitting in a bin, but I would rather it be slightly smaller than dragging on the ground constantly.

I’m beginning to get a sense of how absurdly huge Uberclocker is going to be.

Next up were wheel hubs. The front and rear wheels are identical this time, being indirectly driven. I also cored out a pulley for each hub. 2.5″ Colson wheels press directly on the hub OD, and little bushings fit on the ID.

I remember when making little round things like this was a OMG SRS BIZNESS and took 40 miles of driving and all day to do. Oh, wait, that was last year…

Here’s a completed “coilover” thingie. I’m really not sure what to call it – it’s not a shock absorber of any sort, just a springy plunger thing.

And it latches onto the front support accordingly. The back end will be fixed to the frame and the front supports will float freely on an axle. The spring allows for compliance with the floor obstacles at Robot Battles, but when the bot raises an opponent, they will deflect more and push back with correspondingly more force, hopefully allowing the bot to balance. I can upgrade to stiffer springs if they prove to be too weak.

I also made a motor-side pulley with hub. The drive belt on each side will wrap around all three pulleys and a tensioning roller in a serpentine configuration. I have yet to design a belt tensioner, since I don’t know exactly how long the belt needs to be, and ordered a pair just based on the combined pitch lengths of the pulley system.

The cored-out pulley presses onto its own aluminum hub which is threaded on the inside to mount on the drill motor output shaft. I managed to fuck up this part twice – once by drilling through the wrong side and the other by mistaking a diameter for a radius and cutting twice as deep as I needed to.

Fortunately, MITERS is resplendent in 1″ aluminum rods.

I now have an appreciation of free-machining steels.

Here’s a completed axle-standoff, made from 3/8″ steel. They’re threaded on the ends so screws can attach through the frame and lock the axles in place while holding the frame apart the right distance.

I had to make this part twice also – the first try was with some 3/8″ “printer rod”, which is a sort of weird medium-carbon steel. This must have been a serious printer, because the rod was actually case hardened and uber-polished.

Meaning: No tools I had could touch the thing radially. I actually had to cut off a 3″ section on an abrasive saw to begin machining.

…like that. This is a nice picture, too. If you like long-exposure shots of burning metallic compounds, also check my Fiarwørks Day picture album.

Regardless, even after that, it was still an enormous pain to face to the correct length, chipping the tip off one of the HSS bits (a quick regrinding brought it back to life). It also ate a drill bit I tried to end-drill it with.

I gave up on that and dug around for some better material, and came up with a 3/8″ cold roll steel rod, which went quickly and quietly. CRS not being much stronger than 2024 aluminum, I wondered why I didn’t just make it with an aluminum rod.

…right, because aluminum galls like a motherfucker. A dash of 400 grit sandpaper to the 3/8″ steel rod and the wheel slipped right on.

Lastly, in non-round-thing adventures, I finished processing a part of the fr0k assembly. This is the rear “connecting block” between the two arm towers, and was one of the few pieces that I could still salvage from the ill-fated 1/2″ aluminum plate. The reason is because nothing critical is actually on the front face of the part.

Double-sided features on parts are a pain, since I had to flip the thing over and re-zero everything to make the holes on the opposite side.

Stay tuned for more!

So I thought this weekend would be filled with progress on Ãœberclocker. After all, I didn’t spend 4 hours babysitting the waterjet for nothing, right?

…right?

Wrong.

I don’t know what force of nature the robot gods took hold over that machine, but 60% of the parts are a total loss and need to be recut. I have managed to track the problem down to a burp in file translation (from a DXF to machine instructions) *somewhere*, *somehow*. It doesn’t like my multi-part files.

That won’t buff out. Anyway, I found that a few of the parts had shifted holes. “Hmm, maybe the giant 12 x 28″ slab of .5″ thick aluminum somehow moved… a quarter inch”. However, a quick scan with calipers later found that all the parts had their internal holes shifted consistently about a quarter inch. In fact, the entire profile of the parts was shifted rightward. I didn’t notice this upon retrieving the parts from the tank because it was so consistent (and they were masked by a paper cover)

The machine had cut all the holes and internal cutouts, but then somehow started 1/4″ too soon on the contiguous outline, taking with it all the parts.

Some of the cutouts, such as the arm spacers and small parts attached to internal cutouts, are salvageable, since they don’t rely on the top surface for placement. The arm towers, main fr0ks, and side fr0ks are a writeoff. I would actually prefer to redo the entire thing because I still have time.

I suspected file error, but figured the machine occasionally burps anyway. All machines do – they’re never flawless. But when it happened again cutting Uberclocker’s top and bottom plates… In this case, it routed a path through half the holes, the top plate profile, the last half of the holes, then the bottom plate profile. Somewhere between the “top plate profile” and “bottom plate profile”, it shifted a quarter inch right again. The bottom plate on the right is beyond reasonable recovery, since I’d have to move almost all the holes. Fortunately, I took out the top plate after it was done cutting – otherwise it would have had a quarter inch trimmed off.

I’m going to stop running multi-part files for now, as a sort of damage control.  It’ll take a bit longer to run the jobs, but the machine shouldn’t be playing any trickery. Here’s hoping it’s a one-time thing.

tl;dr waterjet hates charles.

Here’s some more pics of epic partfail.

The mirrored part has an equal right-shift in the profile.

The small internal parts are fine because they were part of an inner loop. I’ll just manually mill the corner blocks on the right. The trussed spacers are salvageable.

These arm towers are a total loss.

Hmm, a hammer might solve that.

Someone truly determined can probably salvage the parts, but I’d rather mooch  aluminum off the ML guys and recut these (I actually did buy this mirrored aluminum myself last fall)

Anyways, with the waterjet mad at me, all I could do was start on the parts which didn’t require it.

. ..like two (of four) drive wheels!

These things took about an hour or so cumulative time. A short job, but interrupted by…

…a MITERS toolgasm!

One of the guys found a bucket of old tooling hanging around his room. The story is that it came from a gunsmith’s shop.

This meant *good* tooling. Taps of every imaginable thread (ever heard of a 5/16-30 or 1/4-24 thread?!) with two, three, four, and spiral flutes, cobalt steel lathe tool blanks, hand-ground tool bits, carbide tipped bits, a bunch of small endmills (including the coveted 1/8″ endmill), countersinks by the handful, and… keyway cutters! A few reamers joined the mix also.

It was like picking through Chex Mix, but with 9000% more iron!

I quickly dug up some organizer boxen and sorted the bits.

Work will continue on Uberclocker, but I’ll have to find more 1/2″ aluminum to remake the lost parts. I should have plenty of opportunities to bootleg the waterjet again as fab week at the Media Lab approaches
.