In our last episode, Pop Quiz moved. I finally confirmed that the drive gearboxes are at least adequate to move the bot around as a weapon platform, possibly even doing some light pushing. Today, however, was a day of epic progress on Ãœberclocker, since I was able to get on the waterjet cutter. In the same session, I also cut Pop Quiz’s top and bottom armor plates out of a raw carbon fiber panel. Drivetrain parts for LOLriokart are currently in limbo due to a bank fart, but some additional components are on the way, and said fart should be resolved during Monday’s regular financial industry business hours.
And a week of nothing happened. Multiple reasons for this: epic partying at the Media Lab, waiting on parts, and deprivation of waterjet access.
However, July is approaching, and so my time will be slightly more constrained with the beginning of a new internship (alongside my Media Lab job). So I’d better get some bot work done while I still can, eh?
The big news item for today is: Pop Quiz 2 drivetrain test is a success.
Drive pods installed! Each motor is epoxied into place into its mount (not very serviceable), and each drive pod is retained by two screws and the “channels of drive pod holding” machined into the chassis, like the design.
And here is Pop Quiz 1.91666666667. I duct taped all the parts the bot would conceivably need, plus some extra things to get it to somewhere around 1 pound. There were things missing, such as a giant blade, but overall it was ready for a test drive.
I approve. The bot is actually a little bit too squirrelly. It’s also running on 11.1 volts out of 14.8, so it will be even faster when done (I doubt the ounce or two that I’m short will actually make a difference). The BB controllers seem to like the motors enough, and I didn’t get any thermal problems from either controller or motor. I’ll probably just dual-rate the drive down on the transmitter… or just drive really fast. It’s somewhat slower than Test Bot just by the feel, but because the bot is so much smaller, is harder to gauge the heading.
I have a feeling that frankenmotors will be an integral part of my bots from here on.
With family visiting and Media Lab partying for the past two weeks, there’s been a drop in bot productivity. But with new parts for Ãœberclocker arriving, it’s time to get back to Pop Quiz so I’m not switching between “big bot” and “little bot” mode. So here’s Pop Quiz updates from the past two weeks, as well as some miscellanea that’s been going on.
How do you carve a UHMW circle out of a square? Simple, a hole saw. I used a 1 3/8″ hole saw to make the internal hole, then drilled the perimeter screw holes, then finished it off with a 2″ hole saw on the outside.
Next came throwing the ring on the lathe and shaving the OD down to 1.75″. Not a critical dimension, so I took a bit of liberty with proper machining procedure.
The completed motor ring thing. The screw holes are for #2-56 screws, four through the top and four through the bottom. This binds the motor to both the top and bottom plates, which is much better than the single-supported motors of Pop Quiz 1.
Next came the task of assembling the motor magnets. I decided again (for some reason) to use two half-magnets to make one magnet, which was problematic for installation. Of course, I neglected to make any sort of spacer….
…so I tried a host of random objects wedged between the magnets, most of which didn’t work very well. Only when I got around halfway did I realize that the blade out of a utility knife was a nicely sized shim for the magnets.
So the finished magnet ring has “less than consistent” spacing. I hope this doesn’t negatively affect the operation too much.
Slamming everything together gives the outer appearance of the finished motor. I have yet to wind the motor itself – this will happen soon.
Now that the motor is pretty much complete, let’s go back to working on the drivetrain!
I gave up on finding .75″ x 1.5″ aluminum box extrusion since it seems to be an odd size. I didn’t need all of the cross section, but figured nobody made .75″ x 1″ C-channel. The only place I could find it was online, and I’d have to wait a few days for shipping.
So what’s the solution? DIY aluminum channel! I could actually make the piece exactly the height I needed to minimize postprocessing that way. Here’s the .9375″ x .75″ x .125″ wall aluminum channel emerging from a rectangular stock piece. Essentially, it was ramming a 3/8″ endmill into the thing until I was satisfied with the depth, then finishing off the sides to 1/2″ wide.
While the channel was still attached to the barstock, I turned it around and machined the little indent that will hold the motors.
Finished and ready for sectioning into individual drive pod frames! The finish on the internal channel is absolutely crazy.
Meanwhile, the carbon fiber panel that will become the bot’s top and bottom awaits. I’m not yet sure how to go about carving said top and bottom plates out of this yet. The Media Lab’s PCB prototyping machine looks promising, since it can use thin carbide bits and already machines fiberglass circuit board panels, and I would only have to feed it a drawing file.
I cut off four chunks of “drive pod stock” and milled them down to 3/8″. Here are the unfinished frames sitting in their final mounting spots.
After drilling the requisite holes, here are some of the drive pod frames with axles installed. The axles are 1/8″ steel rod, and is just shoved into the axle hole – they will be retained by the motor on one side and the bot frame on the other.
A finished drive pod, cf. design drawing. The milling machine allowed spot-on placement of the gears, so it actually runs smoothly. The motor itself will be retained with adhesive, hopefully of a non-brittle variety so it can take impact. That means there really isn’t any changing parts with this design – something I decided to just go with due to design and part constraints.
Let’s hope that doesn’t bite me in the ass.
Drive pod test mounted. The wheel has clearance issues with the frame, and rubs against it significantly. This is due to the squish-fit that I used to place the wheels on the drive hubs, making their diameter significantly larger.
I’ll mill off the portion of frame between the channel to allow for wheel clearance later.
And after a night of mad milling (is THIS why they made CNC with automatic part and tool changing?), all four drive pods completed. Pop Quiz is getting close to the “Looks like a robot… kinda” stage. I hope to test drive the platform soon – I’m really not sure about the torque capability of these motors, and it’s better to find out that I suck at life now than when Dragon Con is next weekend.
Work continues on PQ2 today with initial construction of drivetrain parts and the finishing of the frame. I woke up bright and early at 3PM and headed to the hardware store to get more body screws for Pop Quiz (and a handful of 4-40s for the fixture), as well as a 1/8″ Dremel spiral bit. It was a little sketchy as an 1/8″ endmill, but UHMW is soft enough for it to work.
And off we go. Full bore on the spindle, about 2700 RPM, is still way too slow for something this small, so I fed very slowly. I Dremelmilled the four unoccupied slots first, moved all the fixture screws to those, then finished the rest. No telling how much my zeroes were screwed by playing around with fixture screws, but it’s not enough for me to care.
It turns out I milled a bit short on the left side yesterday, so I had to shave off an additional 1/16″ using the Dremelmill. Should have taken the extra 15 seconds to switch to a real endmill, as the finish imparted by the Dremelmill was less than spectacular.
After the slots were milled, it was time to drill the front holes. This involved angling the mill head 45 degrees and projecting all my measurements to a 45 degree angled surface. Other than having to remember that all X-directions were âˆš2 times more than what I think they are, it went smoothly.
Done. I did accidentally drill too deep and deflected the bit on the aluminum fixture on one hole, but it doesn’t ruin anything.
Finished frame. Overall, not too bad for playing around with the fixture as I was working, not using real endmills for finishing, and UHMW being the amorphous jelly-like material that it is.
While the mill was still warm, I rebored the motor can to accept the bearings. A .375″ endmill made for a very nice “Loctite Finish” hole which the bearings pressed into.
Time to see if I can actually make the drive pods or not. The drive hubs are made from little rounds of brass. Brass is a wonderful substance, as it falls away with the lightest poke of the tool… and I could take off the entire excess amount on a single pass.
A finished drive hub! I drilled the gear with a #14 drill, which, at .182 diameter, is a nice press fit onto the .1875″ shoulder on the hub. The rest of the hub is .25″ in diameter. The bore is 1/8″, and will ride on a Convenient Section of CD-Drive Read Head Guide Rodâ„¢, which is a polished length of 1/8″ diameter hardened steel.
Now THIS is tool abuse. Yes, that’s an X-acto knife shoved onto the toolpost. What else would I do if I wanted to slit a soft rubber rod at a precise length?
Each Pop Quiz wheel is a half inch round of rubber with a .25″ bore, cut off at 5/16″ long.
A completed drive wheel. A 1/4″ drill bit would make a much-smaller-than-1/4″ hole in the rubber, which means I had to cram the wheel on. This is better than having to glue it down or something, and also makes the wheel easy to change.
With the 3 foot section of rubber I ordered, I could make a shitton of little wheels. The rubber used is 50A durometer “Santoprene”, which was cheap at McMaster.
How do you drill a triangular hole?
This is what happenes when you squish rubber rod in a lathe chuck a little too far, then drill through it. The compressed material springs back and makes a neato shape.
State of the bot as of June 1st. I went ahead and did two drive wheels before running out of brass rod. The motor just needs winding work now, and the frame is complete.
I ordered a carbon fiber sheet and the micro receiver from RobotMarketPlace. I need to get ahold of a 1/16″ to .093″ (or thereabouts) sheet of titanium to cut the blade out, and have been stalking EBay for proper candidates. Other than that, Pop Quiz 2 is on schedule with a projected mid-June completion date.
I have begun parts-gathering for Ãœberclocker, and made a McMaster order for all the random-ass bushings, shoulder screws, and other odd hardware I can’t scrounge. This somehow totalled up to be over $50 in random-ass bushings, shoulder screws, and odd hardware. Great Robot Jesus.
But before we start, here’s the toolgasm of the day.
I suddenly feel very insignificant.
Anyways, onto Pop Quiz.
Hey, drivetrain parts! These things are tiny. Very, very tiny. Like, holy crap that’s tiny. The gears are metric 0.4 module, which is roughly 64 pitch. The total reduction is 3:1, from an 8 tooth pinion to a 24 tooth main gear. Fortunately, the pinions have 1.5mm bores, and the drive motors have 1.5mm shafts. There is no way I can bore one of those out.
The drive gears are a little better – 3/8″ in diameter, so somewhat reasonable. They will be bored to 3/16″ and mounted on a custom hub-shaft thing. I’m not sure yet how I’ll manage mounting them on the lathe. And I’m wondering if sub-thousandths precision is required for these things…
The bearings and shoulder screws are for the weapon motor.
Onto the frame. This is the fixture for the UHMW, a Convenient 5″ Square Aluminum Plate. Eight holes are drilled into the plate corresponding to the locations of the drive pod retaining slots seen in the first rendering. I decided against locating them along the perimeter, corresponding to the top and bottom cover holes, because they would be too close to the edge of the aluminum.
5 inches also happens to be the maximum opening width of the vise. Whew.
I’ve known that Bridgeport mills could wiggle their heads in a spherical envelope, but didn’t figure out that it was controllable (i.e. I don’t have to hold a 100+ pound head assembly and jiggle bolts at the same time) until tonight, when I discovered that the head is mounted on worm gear drives which allow for easy, controllable tilting with a wrench.
Milling the front slope on PQ wasn’t going to be painful after all.
If the head joints are servo-controlled, does this constitute a 5-axis mill?
If the rotary table base and tilting table of a conventional 5-axis mill are then added, is it a 7 axis mill?
What kind of creepy things can you machine with that?
Mounted, zeroed, squared up, and ready to go. This is a 6″ square slab of UHMW that will be reduced to a 5.25″ square slab. Then it will be milled down to .375″ thick, from half an inch (Actually .485″ as measured – UHMW manufacturing tolerances are an absurd +/- 0.025 on the thickness)
Uh, oops. I seem to have milled my fixture screws. There goes the hex drive.
Now a 5.25″ square piece of aluminum with a 45 degree slope front.
And now the planing down begins. For this, I removed each fixture screw as the cutter got close to it, then reinstalled it after I cut past the screw hole. This didn’t seem to affect the finish, but is definitely not the right way to do things.
UHMW forms very strange looking, continuously spiraling curls because it is such a soft and malleable material that the last few thousandths just deform away from the cutter instead of being diced.
And the entire slab has been reduced to 3/8″ thickness.
Next was hollowing the slab out. With a 1/4″ endmill and an obscene spindle speed, I blazed a square around the internals, leaving a 3/8″ thick wall. However, I had to route around the fixture screws. This is no problem, since once I get an 1/8″ endmill the area will be detailed and recut anyway.
After cleaning and cutting away the flashing (UHMW also loves to leave lots of edge flashing), here’s the result. The sides are straight. The illusion of waviness comes from the edges, which have been massacred with an Xacto blade.
I haven’t finished out the corners, nor the slots of drive pod holding, since I need to scrounge an 1/8″ endmill or something similar (maybe a spiral bit).
Ahh, the poor fixture screws, milled from all directions. I think I’ll have to buy a stack of 4-40 screws in order to finish this.
After the frame is done, it will be time to proceed onto the drive pods! I really need to see if this crazy semi-direct-drive idea with frankenmotors will work. There’s still time to redesign the whole bot.