Motorama 2014: How Candy Paint & Gold Teeth Happened in an Hour; Überclocker Wins 2nd Place in Sportsmans

In the spirit of Motorama always landing on or near Valentine’s Day, here’s some robot loving.

That’s Jamisons‘ new somewhat-secret-but-not-really 30lber megatRON.  This bot was finished in the minutes before departure, as Jamison and I both had teaching responsibilities and other factors that somehow all decided to congregate this past week.

I spent much of Friday evening checking over Überclocker and making sure it was in working condition. After last Dragon*Con, the drive completely stopped working (it had gone one-sided in the first 30lber rumble). I hoped that the motors weren’t damaged and that, like at last Motorama, it was a solder joint issue, since the DeWalt motors really have no good place to solder to.

I’ve known this to be a problem, so I bought some 3/16″ flag terminals right after DC to remedy the problem. That’s how you’re supposed to connect to these motors.

A single large zip tie wrapped around the whole can, overlapping the terminals to keep them in, and I should have no more of these issues. The two solder-covered tabs above the insulated flag terminals are where the wires were attached before. This is an insecure location that’s prone to flexing.

Fast forward a few hours to the event! The trip down was “ordinary” – with Mikuvan in good repair after the last-minute alternator save (with a Big Chuck’s Automotive Blog entry to come about that), the only issue was having to slow down when snow hit as we got close to Harrisburg.

Here’s Candy Paint being last minute saved. I foisted this effort off to Adam, and a new tagalong who is a visiting student in one of the labs that operate out of the IDC, Joaquin. Joaquin had started helping out everyone with their builds (he’s from Chile – he’s technically here on summer vacation), and at 11:30PM the night before had decided he wanted to come along.

The setup from the original plan had changed to the following:

  • Instead of running 8S, I brought along two 5000mAh 3S Traxxas packs that were in my post-DARPA stash to run 6S. Unfortunately, one of the cells ended up having a broken inteconnect tab, so it was taken out and then the bot ran on 5S.
  • The 160A dLUX controller seemingly did not want to run my motor at all. Weird, but it’s also a relatively smart controller, so it might have thought my 4 turn Delta-terminated motor was a short. We brought along two spare Sentilon 100A units which are my go-tos for brick-stupid controller that can actually flow some current.

Besides those two hacks, the bot ran everything as intended.

Here’s the MIT conglomeration pit table before everything started becoming scattered and messy.

To finish Candy Paint, some field tactics had to be deployed, such as when swapping the connectors on the batteries…

The bot basically right after it was finally through inspection and ready for first matches. Not bad, eh?

During the first match, the weapon motor controller decided to fail short after a solid impact that also flipped the bot a few times (luckily landing on its wheels). The reasons for the failure are unknown – these ESCs are known to work great, then spontaneously decide to peace on the next power cycle. The pulsed short current caused one of the Traxxas batteries to fuse a cell tab; for the next match, that was cut out.

And it generated this cloud.

No, just kidding. That’s from HyperActive, in what must be one of the most spectacular kills in all of robot-time, in its match against overhead saw bot Gloomy (which looks a lot like megatRON). Here is the video (and another angle).

I have mixed feelings about this match. On the one hand – holy crap. On the other, now every well-meaning but naive member of the “You should put, like, a circular saw on it so it can cut through the other guy!” crowd now actually have a leg to stand on.

The aftermath of that fire. Gloomy’s weapon came straight down on the 1/8″ Lexan top armor, which gave way to the saw quickly – the titanium sparks seen in the video all came off the center weapon pivot. And right under it was the lipoly battery of course…

Clocker suffered a bit of damage in its second fight as the upper clamp actuator decided to disgorge itself. The construction of this actuator didn’t really leave much side support, so if it got pushed or torque sideways, the T-nut joints slipped out, which is what happened here. The real issue was I didn’t retighten all the screws from Dragon*Con. After I retightened everything, I didn’t have another coming-apart problem, though the motor eventually stopped working.

Another problem with Clocker was that I didn’t have spare wheels. In my rush to get Candy Paint ready, I neglected to order a spare pile of McMasterBots wheels, and only had one more spare. The wheel on Clocker that was the worst ended up getting that last spare, and the rest… well, by the end of my 3rd match with Nyx, they were gone.

So out comes the VHB tape. This was an idea cooked up by Dane – VHB double sided tape is anchored down with superglue, then stretched very tightly for multiple wraps, then anchored to itself with more superglue.

Now, I this this is a brilliant idea for smaller bots, or bots with less drive power, but Clocker’s all drivetrain so these tape treads shed within a few seconds of the pushing matches getting heated. For my last matches, it was like driving the bot on ice.

Candy Paint lost its last match when a wire came loose from the very ad-hoc wiring harness and got snipped by the big shredder holes in the pulley. Oops!

Here’s the final damage tally to Clocker after its last finals match (with Upheaval, no less). The amount of bouncing Clocker did broke off one of the end rollers, and the outriggers themselves are also extremely bent up (hard to see here). The clamp motor isn’t responding, but the actuator is still mechanically good. Seems like it’s due for a full overhaul between now and Dragon*Con…

I’m also considering switching the McMasterbots 40A wheels to some Banebots wheels. I can make a quick-change hex hub in the same manner as Candy Paint and make the wheels far easier to swap, and have many spares on hand. Right now, I have to hand-machine and modify each of the wheels, and as this and past tournaments have clearly shown, Clocker blitzes through wheels – rears especially.

I am extremely satisfied with Clocker’s performance this time. Much like Dragon*Con 2013, it came in second place, being out-robotted instead of something completely stupid and negligent happening. I also now have settled the score from that event with Nyx. As usual, it put on a very visually stimulating show.

Here’s the matches:

vs. Knuckle Buster

vs. Piston Wax

vs. Nyx

vs. Upheaval (two matches for the championship)

Candy Paint went 1/2 – I’ve uploaded the matches where something remotely exciting happened.

vs. Hot Fuzz

Next, it fought Cathi (a ring spinner), wherein Joaquin accidently threw the radio’s “Throttle cut” switch so the weapon didn’t spin up, then Magnificent Poncho (a pusher), whereupon it spent most of the match upside down until the blade came off. It turns out that the blade motor was wired to spin in the untighten direction! Oops.

Finally, in a MIT-on-MIT grudge match, it faced the Atomic Bumble Prime.

One additional overlooked issue was that the blade was a double bevel profile (i.e. looks like <         > from the side). Any hit to an off-perpendicular surface will tend to throw the bot upwards, which clearly happened. This could have been taken care of with 5 minutes of grinder hits, but it was just not done.

Candy Paint’s future is a little uncertain right now. The bot is mechanically sound, took no damage past cutting its own wires, and I can easily rewire it to conform to my original expectations. However, outside of Motorama, there isn’t a local event that handles full 30lbers with weapons. I could put a sanding disc on it and bring it to Dragon*Con. Otherwise, Candy Paint will probably remain idled until next Motorama, unless I feel like putting it together properly and going to some basement to grenade a few old TVs and printers.

That does it for this event! Admittedly rushed preparations ended up working out somewhat, and Clocker has proven itself again to be a pretty damn reliable bot. And I’m not broken down somewhere in the middle of New Jersey.

I now turn my attention back to daily operations of the IDC fab space, as well as my new season of 2.00gokart students. Expect some updates on that as the semester progresses! The class is now somewhat legendary among freshmen and sophomores.

Candy Paint and the Hail Mary Finish: Last Update before Motorama

Over the past few weeks, I was supposed to complete Candy Paint & Gold Teeth – especially taking this past week to do so. However, real life has a way of trying to happen all at once to me instead of in reasonably scheduled chunks. This often manifests itself in trivial instances such as a delivery guy showing up with a pallet jack at the same time I’m trying to answer a student question about machining, all the while my phone starts ringing. During all other times, I am a lazy bum.

This past week, it manifested itself in the entire shop being room-swapped, almost like out of some bad redecorating show:

Yes, that is the Shopbot, in all its 5 x 10 foot glory, mounted on a pallet jack and a trash can rolley base. With zip ties.

Oddly enough, I didn’t think this one up.

The former EE/laser cutter room was turned into the combination fab shop and Shopbot room…

..and the former shop had all of the EE equipment, rapid prototyping tools, and 3D printers stuffed into it!

Overall, it’s much more cozy instead of being a big patch of empty space. We did this to consolidate shop space in order to free up the massive area formerly occupied by the Shopbot for more researchers to have space. It was completely necessary, and (of course) this past weekend was the only window of opportunity: any later in the term and it would interfere with classes, and start of summer is too late because of the space needs of new researchers.

Besides that, the following also happened:

  • 2.00gokart Season 3 has started! This warrants its own post, but right after the shop was moved, I had to give orientations/machine training sessions to all of the 20 BRAND! NEW! excited students.
  • the Department of Facilities insisted that this week was exactly when they needed to wax the floors, so many of these things (but THANK ROBOT JESUS NOT THE SHOPBOT) had to be temporarily moved back out.
  • Mikuvan ate an alternator on Monday. This is reserved for another episode of Big Chuck’s Automove Blog, but this was a several-hours repair job yesterday after waiting for parts.
  • I had to teach a main-2.007 Solidworks lecture on short notice, which was a few hours of preparation.

What does this mean? Well, for one, I have almost nothing penciled in for next week right now, so I should just be able to sleep the whole time!

This post will recap Candy Paint’s progress up until last night. It’ll be short but pictureful. As it stands, there is a slim chance of the bot being completed right before check-in and inspections begin, so I’m just rolling triage – what can be done will be done.

First, the welded frame:

Hot. Damn.

I passed this welding job onto my friend Jack, who is the shopmaster of the D-LAB facility across the hallway. I wouldn’t have stood a chance – my only experiences with welding aluminum have always ended up in puddles.

This frame taught a lot of lessons – namely, I’m never going to do it again. I expected some “taco” deformation of the whole thing, but it was actually quite straight. What I found out later on was that many internal features had moved or warped out of place seemingly magically, and almost nothing fit the way it was supposed to. There was to be much Dremelling in my future.  I’m actually not sure why I wanted to go welded-frame in the first place… something about trying out new techniques. The frame was suboptimally designed to be welded – tight corners, very mismatched material thicknesses, and holes very close to edges, since my experience in design-for-welding is slight at best.

One of the first things I did was grind off some of those strength-giving fillets because the top of the bot needed to be flush – again, a design-for-not-welding guy trying to do design-for-welding. This will weaken all the joints, so I’m hoping the underside fillets and tab/slot mates make up for it.

A rough grit zirconia flap wheel made short work of the aluminum.

The finished result. I then took this downstairs to the giant 20 ton hydraulic press for some gentle frame straightening, using the bar that I messed up on the rolling machine as a flat jig. With this done, the next step was to post-process the holes:

One thing I am now aware of that is done is weld a blank frame first, then post-machine everything from a single datum. All of my features ended up moving around or warping – the big center hole’s bolt circle somehow shrunk almost .02″, necessitating Dremelling to remedy. I’m glad I invested in those carbide burrs.

With the frame ready – or almost, with an hour’s more Dremelling than I had intended – for drivetrain and other parts installation, it was time to make those other parts.

The center spindle is made from a single piece of 4340 steel. I single-point machined the giant 1″-14 threads.

I machined the giant center block from a 2 x 4″ brick of billet using the EZ-TRAK CNC mill in the auto shop. Here it is completed with one of the tapered roller bearings and spindle installed!

Waterjetting is easy, but billets are satisfying.

Stator and partially completed stator hub. The stator is a copier motor pull unit I had, the same size as the one in Kitmotter and one of the Razermotors. I’ll be rewinding this with only a few turns of ungodly huge wire.

It took a while to install the wheels, because all of the slots had shifted a little, warped a little, or shrunk a little. But this is what the bot looks like with all 4 wheels!

This is remarkably straight and level – there’s a tiny gap under one wheel. Oh well – after the first hit, I’m sure nothing will be straight ever again.

Installing the big center spindle block. This is a pretty integral portion of the bot, and it fastens onto all three major frame rails inside.

The little offset pocket is a zeroing error on the CNC.

Pretend-o-bot #1! The spindle isn’t constrained here, it’s just sitting in one bearing. This thing spins for a minute if I whip it up to speed…

I stripped all the windings off the motor and used a string to measure the length of new winding needed.

This motor will be spinning north of 15,000 RPM to drive the weapon bar at around 2500. To get this speed, I needed to make a very hot winding – I calculated that 4 turns, Delta-terminated, will be sufficient.

To make 4 turns and not waste all the copper space, I had to resort to very weird measures:

This is my “9x Hobbykinging Rig” from building the Chibikart motors. It places 9 strands of #28 magnet wire in parallel to simulate an easier-to-deal-with 18 gauge winding. “Hobbykinging” refers to the tactic employed on most Chinese r/c motors of using many parallel strands of fine wire to wind motors, as it’s easier on workers than trying to bend thick wire and fit it properly. Done right, it can achieve a higher copper fill than a single thick wire, but there is a diminishing returns point if the individual wires are too thin (such that the enamel insulation starts making up a sigificant portion of total cross sectional area)

To make the windings I needed, I calculated I had more than enough space for four runs of this. So, I wrapped the Hobbykinging rig 4 times around a 12 foot long table to yield 36 parallel strands of #28 – roughly equivalent to a 12 gauge winding.

This is what the winding looks like.

I actually enjoyed winding this motor immensely. It was so easy! Just sling the giant bundle into the teeth and pull. And only for 4 turns instead of 30-40 like the hub motors!

I wanna go work for Hobbyking!

The completed winding.

The  next step was to terminate the motor in Delta by bunching a start and adjacent end of another phase together (e.g. Start of A phase, end of B phase get bound together). I just ran the bundle out using heat shrink, then torched the ends to destroy the enamel coating, then used my Battery Abuser to tin the ends.

After this, I potted the bundles in epoxy to secure the windings.

Moving on to mechanical work again, it was time to assemble the drivetrain.

The sprockets were cut out a while back with the waterjet load. Here they are installed on the motors – and after some manual chamfer-filing while the motors were run under load.

A good amount of Dremelling and hammering was needed to move the mounting surfaces back to where they were supposed to be. The motors will hopefully be secure with their front and rear mounting brackets!

One side’s drive installed. Check out my little chain tensioner blocks – these are made from Delrin and are pushed into the chain with a set screw drilled into the frame.

Each side, after installation, got a 20 minute long run-in so the chain could carve its own path into some of the weld fillets. I came around and cranked the tensioner screw down little by little as time went on.

Now with both sides!

Most of the remaining work on this bot is just putting things in. I have to machine a (simple) shaft for the weapon motor, then it can be test spun. Electronics and batteries will most likely be installed ad-hoc.

I’ll leave the updates to Clocker to another post – so this will be the last one before Motorama. Now, back to the shop….