Archive for February, 2017


Motorama 2017: The Event Report; Or, How Not to Scale-Model Test Your BattleBots

Feb 26, 2017 in Bots, Events, Überclocker 4

And we’re back! I must say, in a way, I miss the abject chaos (read: spinners) of the full-contact weight classes, but it is glaringly clear that I need to get my strategy back in shape. In all, this event was a good wake-up call for me if I want to play the BattleBots #season3 game seriously, but that’s for a later analysis. Here’s how things went down, starting with the finishing of Clocker a few days before.

One of my last to-dos was making spare armor wedges. I’d already waterjet-cut the plates, so they just needed to be cleaned and welded. These wedges represent a simplification of the design used on Overhaul that I would like to transfer. They’re simpler, reducing the number of facets and panels by half*,while also retaining the same lower-edge durability with a (higher mounted) gusset. However, they are missing the “Jersey barrier” double-angle front that Overhaul has, and this will be important later.

So there are four wedges – two are made from regular cold-roll mild steel, and the other two from 4mm AR500 plate. I’m really expecting to run the AR500 plate as primaries, and only ditch out to the mild if they get (somehow) demolished. I suspect there wouldn’t be much left of the bot if that were the case, but it’s good to have options! The 4mm plate one weighs several ounces more than the mild steel, owing to higher plate thickness (.125″ vs .140″) so I’ll definitely have to free up weight for it.

I jigged the whole thing up since it tabs together into itself and tack-welded the panels together using a TIG welder, before switching to the good ol’ spray-and-pray MIG welder to blend the outside seams together and drop a huge interior fillet into whatever edges I could on the inside. I am still the only person I know who tacks assemblies together using a TIG welder, and then switches to using a MIG welder. I write this off as me having zero patience for welding, but needing the initial assembly to be straight, so I do it with the precise near-zero-force application of a TIG welder.

*Note that Clocker doesn’t have forward- or side-facing wubbies like Overhaul, so if those features are being added back, it would increase the plate count, but still not to the point  where I had them for #season2

Free up weight? Where the hell else can I do that from!? It seems like Clocker’s been pretty well dieted, but a few weeks prior I had started thinking of do I really need semi-infinite drive power? in the form of possibly replacing the AXi motors. They work great, yes, but are definitely overpowered and therefore heavier than I need. I decided to swap to a set of 42mm SK3 outrunners, which would reduce me by around 4 ounces per motor, allowing me to use the AR500 wedges as the heaviest configuration. Power-wise, the SK3 outrunners would have been just fine. They also pair up with the pinions of the 4:1 P60 gearboxes from BaneBots I ordered (due to the higher Kv) and bolt to the motor plate with no modifications.  This is a great combo – I highly recommend it as a plug-and-play 30lber-scale brushless drive rig now.

The motors were basically the last thing to arrive before I had to leave, so I decided to hold off swapping the parts in until we got to the event.


The following image shows the totality of the glory of America:



On Thursday night, we packed Literally All the robots into vantruck, along with a sizeable amount of tools, support equipment, and other miscellanea. I planned to get there early-ish Friday to help set up and also to aid in Antweight & Fairyweight tournament logistics. Along with me were SawBlaze and Overhaul for display at the front of the audience section.

Sadly, this trip as-photographed did not happen, but that is an entire other story that has to be told separately. Long story short, the haulage minus SawBlaze and Overhaul were reshuffled into Mikuvan. This is a great story, I guarantee you (if you stalk me on the Internet, you already know it, so no spoilers!)

Alright, so it’s like 2PM on Friday now when I get there and everything is horrible and nothing matters. Let’s swap the motors onto Clocker:

Boy, those ESCs – spares left over from Overhaul and Sadbot, Dlux 160A HV units – are now officially overkill too. That’s what happens when you make a parts-bin robot. With the motor reduction, I was able to make weight using the AR500 wedges. Also in the same disassembly service were the floor scrubber tires:


Here’s a better look at them. I liked how they handled in the test box – still just a little light on traction, but very predictable. I brought along the Forsch (black) 60A wheels also, but decided to run these first since the Forsch ones felt a little more stiff.

Fast forward to Saturday and….

I feel like I’m at some kind of  career fair or anime convention. The people-ocean density was staggering; this is the largest Motorama Robot Conflict historically, and the largest year-by-year growth (over 50%). A lot of new faces, probably 25% of builders, and also quite a few returning legends. It’s a good problem to have.

In the interest of not dying, the 3lbers (beetlewights) were basically running in a parallel event with an 8 foot arena just off screen to the left, with only large bots – 12lbers, 30lbers, and 30lb Sportsman’s – running in the big arena.  Given the sheer number of beetles, it was the only way!

What’s great is MassDestruction helped spawn several ‘newb-vets’ who cut (….blunted?) their teeth in the MassD arena over the course of the last year.  These are two of Alex Hattori‘s robots. At this time last year, he had a 30lber made of two steel bars welded to a cast iron pot, and since then he’s cleaned house at like, every MassD ever, I swear.



Some other remarkable bots forged at MassDestruction from guys who work at, uh, MarkForged. Crap, my sponsor is beating me at my own game! What do I do!?

Another one of my favorites return – this is Pitter Patter, a 30lb shuffler (actually 45lb in the weight class) which way back in the olden days of Motorama 2015 was the original design model for Overhaul 1′s shuffle drives, which were basically a direct knock of this thing! For this version, the saw got smaller, but the shufflers got way faster… like 3000 RPM fast. This thing was cookin’ it in the arena.

Basically, you’re not getting anywhere NEAR the whole story just from these few photos. I remember when robot tournaments were this big, from the momentum of the first run of BattleBots, and I hope I see the 2nd Great Awakening of robots progress further still.

Anyways, onto my matches! This is Glasgow Kiss.

Topologically, it’s a good mockup of the Cobalt match. This is okay too! I’d actually hoped for a vertical spinner opponent so I can practice my anticipated strategy of using the ünicorn. However, I’ll gladly try to practice my horizontal-fending tactics too. The high level plan is to come into his weapon tangentially using the AR500 wedges and bounce him around, ideally towards walls, and try to corral into corners. More or less the same plan as for when I fought Cobalt.

I mounted the ünicorn anyway in case it could be used – I wasn’t counting on trying to swipe the belt pulley, as it’s too far inwards.

So how did this match go? Uhhh…

Well that’s not very typical at all.

Let’s watch the match video to find out what happend!

Alright, so my strategy starts out working fairly well. I’d say about 0:30 is when things start going awry. While I get a few more good tangential shots in, Glasgow Kiss is able to get one or two shots in which climb up the wedges and take out the clamp actuator and main lift gear.

At 0:49 I make a pretty bad driving error and end up plowing directly into the blade, so the forks and clamp are pretty much done by then – you’ll see me raise them to try and keep them up and out of the way.

The last big connection throws both of us apart across the arena, and I’ve lost all drive power by now so I tap out.

What Andrew (driver of Glasgow Kiss) does well is pivot the bot on the blade axis – in part a consequence of it being so heavy – such that it’s hard to just ‘get around the back of’ or execute similar strategies. He does this several times to leak away from Clocker’s grasp succesfully, leaving me to chase while he spins back up.

If you watch closely, you can see Clocker has some maneuverability issues right away. One of them is the bot’s right side having a tendency to stop and not reverse, which means I missed a few in-place turns. This occurred to me as strange – I mentally wrote it off to the smaller brushless motors in the drive cogging on start, but it definitely didn’t occur in test box driving. The heat of the match kept me moving, though, and I elected to try and drive around the problem, exercising the tactics I outlined in how2brushless at the bottom.

So Clocker seemed to be in one piece still at the end. Time to appraise the damage:

Check out the gear carnage. This gear is made from 7075 aluminum. It’s a nice and rigid alloy, one of the strongest by tensile strength aluminums, but it’s really best used in bulk such as gearboxes or bearing blocks and the like, not in thin sections. The gear is fairly heavily webbed out for weight, so it cracked through readily instead of bending. A 6061 gear would have bent and I would have had a chance to sledgehammer it back to something resembling flat.


Glasgow Kiss machined off most of this corner here when I was turned around. I’ve thought about making plastic corner hoopy-jiggles before, but haven’t been compelled to yet. As a part of a comprehensive horizontal weapon defense strategy, it might be worthwhile to do for Clocker using some 1/4″ UHMW or a thinner spring steel.

D’oh. I think the cross-arena impact stripped all the #6-32 threads from the end of the gearbox, so I lost drive on this side. On the other side, the chain jumped between the drive sprocket and the rear wheel sprocket.

You know what was awesome though? The AR500 wedges, on both sides, are practically untouched. Lightly divoted, but they were still flat to the ground. I did write off two of the lower wubbles on each side which had some tearing damage beginning.

But you know what – this setup went head to head with one of the biggest 30lb weapons a dozen times and isn’t much worse for the wear. What it really showed me is that Clocker’s frame and armor is perhaps overly built for the weight class now that geometry is compensating up front for frame thickness.

By near complete accident I’d say, the ünicorn came THIS CLOSE to piking the pulley and belt.

Alright, it’s time to fix everything up. Both sides of the bot had to be disassembled to replace the drive motor studs with longer ones. Since the P60 motor plate screws don’t go all the way through, there was some thread left which I could use with longer #6-32 bolts.

It looks like the frame was tweaked about 1/16″ in a parallelogram shape, from a similar corner hit on the rear right side (opposite the well-machined one), so the left side drive sprockets became offset enough to cause problems.

Getting the damaged lifter parts off was an adventure that took a long time. I’m now heavily rethinking the clamp collars on live shaft approach. It was fine in the Sportsman’s class where Clocker never took any real damage there, but with everything twanged up, there was hearty use of deadblow mallets, aluminum pusher tubes (to avoid marring the shaft), screwdrivers, etc.

What I couldn’t save were the clamp actuator and lift gear. I had thought about machining another lift gear the week before, but it remained just a thought. While I had a newly assembled and painted clamp arm ready, I didn’t bring spares for the clamp actuator. Without a backup clamp actuator – since Glasgow Kiss had basically wiped all the internals out also – I had to push everything back together in “spatula mode”, just with the lower forks and around 120 useful degrees of gear. Once again showing the difference between Sportsman’s and the full contact weight classes – just like in BattleBots, you should really be prepared to build 2.5 robots, one full set of spares and another for the things which break the most often.

So I delay my next match (and run down that delay as far as I can) to get spatula mode together. When I finally hustled into the arena, though, I discovered that Clocker could only spin in place or turn right. I clearly had wired one of the drive motors backwards, but what? Moving only channel 1 in my elevon-mixed (single-stick driving, basically) radio only caused the left side of the bot to move, with no response from the right side. However, it could obviously spin in place; without a motor being backwards, it means it could drive straight forward or backwards, but only turn right with 1 channel.

Without more time, I had to forfeit my match against Shaka, who, I will point out, somehow went 2/2 at this tournament using only forfeits. It won its matches by forfeit, but had endemic electronics problems which caused it also to lose by forfeit… I am told that in testing shortly after our non-match, it blew up.

Back in the pits, it took me a little more investigation to discover that my Hobbyking radio had somehow lost a mix. When you configure a radio for single-stick driving (or Delta Wing, Elevon, V-tail, etc. for aircraft), you assign mixes to tell channel outputs to listen to certain combinations of stick inputs. Here’s what a typical simple elevon mix looks like for my Hobbyking T6A-v2 transmitter:

There’s two mixes involved – one to tell Channel 1 to move with Channel 2, which on a typical radio is the vertical throw of the right-hand joystick. This means pushing forward on the stick sends the same signal to both outputs on the receiver, so the robot drives forward.

The other mix is to tell Channel 2 to move the opposite of Channel 1, which on a typical radio is the horizontal throw of the joystick. This means if you push stick right, one side of the bot moves forward and the other moves backwards, and is accomplished by setting the mix percentage to be -100 in both directions (do the opposite no matter which direction the stick is moved)

For me, the latter mix – the one outlined in Miku Pink – was NOT responding, despite showing correctly! This meant moving Channel 1 resulted in no opposite motion, just the bot pulling right. This was exactly the behavior seen in the arena, and I would never have discovered it if I had not accidentally put a motor in backwards.

I said the maneuverability tics Clocker showed in its first match will come into play later. I’m now 99% sure that this issue affected the match, and I tried to dynamically drive through it since I try to avoid stationary directional changes (turning in place) due to the brushless drive. A non-working Elevon mix will still kind of work if you move Channel 2 first – it will simply add and subtract Channel 1′s value from one side. In this case, it left the bot prone to pulling right, which is exactly what I saw.

How did I discover this was the problem? Well, I simply had it resend all the settings to the radio without touching a single one and it resolved itself. My radio literally lost a mix from its memory between Friday and Saturday for reasons unknown, even to the point where it convinced its software that the mix was still present.

I must say, I am not even mad. This is an impressive failure mode that I’ve literally never seen before, ever. Before anyone dishes on Hobbyking radios, though, I personally have owned a half-dozen (I keep accidentally giving them to newbies or random students and then getting another one) and also worked with hundreds back in my 2.007 days when they were the radio of choice for the class, and this is the first one I’ve ever seen DROP A SICK MIX like that.

With Clocker out of the tournament and the radio issue solved (!?), I waited for the 30lb rumble to join in on, where I basically overdrove the arm past the end of the gear immediately….. so I simply ran around as a wedge corralling bots in corners until the Vex sprockets’ teeth all came off!

My chain gliders probably wore  enough in that 5 minutes of crazy driving to make the chain skip on the sprocket (since it doesn’t have that great wrap angle), and the power of the brushless drive proceedd to machine the teeth off in short order. Ah well – it was a great rumble anyway. At one point I had every bot except Translationally Inconsistent, who kept slithering away sideways, piled in one corner.

Once I find a good video of it, I shall update the post to include it.

What’s great to see is that the 60A wheels hardly wore. Obviously this is both good and bad, since it means I could have traded hardness for more traction. For the 30lbers, I might go back to the 50A compound – Clocker in previous incarnations has run 50A wheels and I’ve been satisfied. Now is when pouring a few full-size wheels for Overhaul to try and drive around would be a next step.

We part with some shots of gourmet damage from one of Jamison’s loser’s bracket matches against Triggo. megatRON was upgraded to have an AR500 impactor disc on the end instead of a saw, and having that house brought down on you is capable of some serious damage:

this kills the triggo :c

Check out the 1/8″ heat-treated chromoly-steel shell rim also, from the same weapon:

This thing is not trivial; megatRON was actually one of my more feared potential matches because I have relatively weak top side defenses. Expect potentially interesting changes to Sawblaze for #season3 perhaps?!

Speaking of which, what takeaways for Overhaul do we have here besides the obvious bring a spare of the thing you don’t think you need spares of. Or three.

  • DAMN, THAT WAS A GOOD MATCH THOUGH. Honestly, if I had the choice of losing like that to Cobalt, versus the way I did via #setscrewghazi, I’d have picked the former in a hurry. I would have had enough spares to bring Overhaul back online quickly anyway, and it would have made for a much better show and much better test of the bot.
  • I’m highly satisfied with the AR500 wedges. So happy. It deflected the hits from Glasgow Kiss with ease, and also seems to have done its job of transferring the energy into the floor. AR500 has become a bit of a crack epidemic in robot fighting recently as more of it is readily sourced along with laser/waterjet services to handle it. It’s a nice alloy, really – heat treated to the high 40s Rockwell C already, and easy to weld with conventional consumables.
  • Good deflection is also a curse, because you aren’t in control of where the big beating-stick goes afterwards. I’m more convinced than ever – besides by this hit – that the double angle on the front of Overhaul’s pontoons is an absolute necessity. I designed without them for Clocker for simplicity and to see if I’m just being alarmist, but what the single slope let Glasgow Kiss do is deflect its own way upwards and clean house in the clamp actuator. I will need to think about how to  how to retain or improve this design for Overhaul, and to add it to Clocker.
  • I think it might be time for a scoop, for both Clocker and Overhaul. You know how Overhaul has the short arms that I used against Cobalt? Imagine those becoming vestigial and ending behind a angled steel plow which could nest in between the wedges on their inside slopes, making the front of the bot more contiguous. The remnants of this design can be seen in the forward-angled plate that resides on OH1′s forks.
  • It’s more clear than ever that a self-reinforcing geometry trumps material thickness outright. If scaled down directly without changes, Clocker would have 0.75″ thick frame rails, which it clearly doesn’t. It has 0.5″ thick, heavily-machined out side rails with 1/4″ thick cross-bracing plates, and that left the match against Glasgow Kiss needing a single screw extraction and maybe a hit from a good ol’ Engineering Hammer. What this actually means is I spent much of the 6 hour drive back from Harrisburg trying to rationalize that maybe I do need to have Overhaul’s frame remachined again. I’d be able to optimize for the geometry of the side rails. It would shed a lot of weight which can go into other systems I was running out of weight for, and really, based on how deeply Overhaul’s frame rails are pocketed, it’s almost useless to be made from 1.5″ thick stock. But UUUUUUUUGGGGGGGGGGHHHHHHHHHHHH.
  • I’m really, really itching to leave the clamp collars behind when it comes to power transmission to the forks. I think when it comes to fork improvements, just adding cross-bracing to Overhaul is enough, and I way more favor the 8-bolts-to-remove-an-arm setup on it right now for serviceability. I can replace a full set of arms and the clamp actuator on Overhaul faster than I could get the damaged forks off Clocker.

I would love the opportunity to test these hypotheses on a 30lb scale again in less than 1 year, especially because I (think) #season3 is still going down this year. Even if I can’t prove my hypotheses in short order, this was all good stuff to know!

It’s Motorama 2017 Time! Überclocker Changes and Upgrades

Feb 15, 2017 in Bots, Events, Überclocker 4

Since Franklin Institute this past year, I’ve been spending quite some time thinking about what changes I need to make to Überclocker for the annual winter robot party, Motorama. It’s the largest event on the east coast for years running, and the ONLY one left with full-contact 30lbers; I’ve gone basically every year since 2013, and sporadically before that (2008, 2010). This year is slated to be some kind of BattleBots #season2 reunion (where Season 2 was called “Motorama, the TV Show” since so many builders who regularly participate ended up on teams!), and there are some of us who are taking the opportunity to do some… scale model testing. Quick! What is the Reynolds Number of a flying Tombstone!?

So here’s what has been going on during the past few weeks! In my summary of the Franklin event, I identified a couple of strategic issues with Clocker which would also concern Overhaul for #season3, given that they’re built so alike.

The first was having everything ‘line up’ in the front. While it also included making the pontoons more adjustable, higher priority on my mind was making sure the arms have enough constraint that they don’t just splay out. We saw this happen in the Overhaul vs. Beta match where I caught Beta with one arm over and one arm under, so the arms because angularly misaligned. Recall that Clocker 3 and before all have multiple spanning elements holding the forks together; alignment was never a problem with that, but Overhaul didn’t have those elements primarily for an aesthetic reason (to maximize the forkiness). While Clocker sort of did have those constraints at Franklin, it was just one spacer stack, and that was quickly lopped off by megatRON.


So I’m gonna add more, duh! These two additional spacer & tie rod stacks are located out-of-plane with the one at the end, which will yield much greater torsional stiffness.

One issue is that another 18″ of threaded rods and aluminum tubes will put the bot back overweight a few ounces. Nonetheless, I intend to just build everything out to my desires, and then try to weight-cut from there, rather than compromising early. To pre-compensate a little, I decided to order replacement smaller drive motors. The AXi motors are great, but they are dramatically overkill for power, and I can definitely afford to lose a few ounces. Going to the 42mm outrunners will save me about 3 ounces a side, which alone might be enough. In order to utilize the smaller, higher-Kv motors, I also decided to order a pair of Banebots P60s in the 4:1 ratio instead of my current 3:1s, which should allow me to keep about the same speed.

Now, the 2nd big strategic weakness I want to experiment with is where it gets a little interesting.

I mentioned in the FI post about minimizing my defensive cross-section when it comes to vertical weapons. Those things – including drums, drumlet-drumettes (smaller in diameter and width) and vertical discs/blades are actually what I fear the most designwise, because they do two things to you in a match. One is flip your bot over, from which you need to recover (and which would take precious seconds where you’re vulnerable to followup attack from a good driver), but the more insidious one is ruining any straight edges you might have had where the weapon hits. A small KE weapon will put all of its energy into your material like a singularity; it will deform wedges and protrusions, basically preventing you from having an advantage again. All it takes it one fuckup, as Clocker’s match with Duck Yeah and even better example Blacksmith vs. Minotaur show. Notice how Blacksmith more or less has control of the match before it gets dinged once.

So to counter these kinds of weapons, you would have to do two things. Number one is keep them away from you, and number two is present as small of an area for them to touch you in as possible. There’s a lot of precedent in the sport with “keep away sticks”, including one used to great effect on Icewave last time on the show. To reduce my “vulnerable edge length”, then, I basically had to distill the front wedges down to points.

I take that back: this got interesting very quickly.



So that’s revision 1 of the design. See those perforations? That’s for if I mess up and somehow manage to plant these into someone’s weapon instead of besides it. This design is intended to be cut out of AR500 grade steel, which is extremely rigid and springy but won’t stretch that much, so it will preferentially break at the postage stamp line. It’s like an active salamander tail system.

The saw teeth on top are the real bad idea here. Instead of a keepaway stick, I wondered what would happen if I made it a part of the offensive strategy. Most of these little vertical weapons have rubber belts attaching them to their motors. What if I just went straight for that with a very sharp stick? Stab into the gap between weapon and robot frame until you damage the belt or take it right off. That would take some serious driving and luck to pull off, which lured me to the idea further.

There was only one thing I didn’t like, and it was one of those “come back to what you CADed up last night in the morning and think again”. One of my complaints in the FI recap was getting stuck on MegatRON when we charged at each other. These extra-long death-shanks are attached just as the pontoons are, so if I run up on someone else’s wedge I can just as easily prevent Clocker from getting back off. Which is serious bad news when it comes to avoiding a vertical disc/drum spinner, since now they can just turn slightly for a broadside.

This led to revision 2:

Yeah. “It looks like a sawfish-unicorn”, or an Overwhal. That’s right, I decided to affix it to the clamp arm instead, exactly in the fashion shown.

This position I liked a whole lot better for two reasons. One is that it’s implicitly height-adjustable, and can actually be a manipulator weapon of its own. Clocker’s top clamp arm is not trivial – it is designed and built for about 500-600 pounds of closing force. It will lift a lot of things on its own, and is more finely positionable than the lower forks. It’s also more durable with its leadscrew attachment, but the leadscrew anchor is also a mechanical fuse for if things go very wrong and it gets the uppercut treatment – it will break away and probably fling the clamp arm backwards and out of the way, leaving the forks still usable. If I attached this to the forks, and they get bent, then my life becomes very difficult.

The second is a takeoff from the height adjustability. I realized that offensive unicorn strategy #2 was that now I can reach around weapons and bring the house down on their retreating sides, where the disc necessarily disappears back into the robot. With crafty positioning (or a lot of flailing) I could pretty easily literally throw a wrench in the works and shove a wad of AR500 directly between the robot and its own weapon. This would probably result in a very sad unicorn horn and ideally more sad opponent; for me, that’s why the postage stamp holes are kept, so not only will it break away on a successful landing, but will also do it and leave me a 2nd chance if I miss.

Strategically now, I can keep the clamp arm closed and all the way down and use the horn as a keep-away stick of minimum attackable cross section, and also manipulate bots from afar, or get it caught in something else like exposed drive wheels.

….and if you thought it looks silly in the CAD, it looks 10 times as silly in real life. I actually want to make another one of a different length now!

This is another ‘attachment of several ounces’ which would necessitate shedding weight elsewhere, which I will find. One thing I designed up previously but never implemented in real life was a set of light wedges, to be made of sheet 1/8″ aluminum bent into shape. I’m going to go ahead and make them, since they’ll cut around 1 pound off the bot each (Those steel wedges are HEAVY!)

I won’t need that much reduction in weight to use the horn of course, so maybe the configuration will add something else interesting to make up the weight, or just ballast. There is literally no point in weighing less than 30.00 pounds.

So that does it for major design changes. Moving onto more minor quibbles, I wanted to go back and have a look at the wheels again. The custom 50A cast urethane wheels worked beautifully at Franklin, and I now had a bucket of Simpact 60 and Forsch URS-2160 (McMaster part number 8644K24), both 60A urethanes with much higher tear strength ratings, to try.

Now that I was confident in the process, I revisited the hub design. I just designed the first hubs with circular thru-holes for rubber retainment so they could print without support, but the circular holes caused the diameter of the hub to start getting large. I didn’t have much more than the 1/8″ tread pattern’s worth of tread thickness per wheel. With a more rigid rubber, I might be able to increase the relative thickness of the tread portion.

I updated the hub to look pretty much like a scooter or skate wheel core – through-slots replace the circular holes, and the walls are thinner. This brought inwards the OD about 1/4″, which is great!

I also wanted to play with another tire geometry. A little earlier in the year, when Big Chuck’s Robot Warehouse was just being set up, we rented a floor grinder to strip the wooden floor of the decades of industrial grunge that had set up colonies within it:

Well gee, as soon as I saw that, how could I not clone the design in one of my own wheels? So if you’ve never used a floor grinder, a big sanding drum gets shoved over this flappy-wheel. To install it, you lock the rotation of the flappy-wheel and gently rotate the drum over it (in thise case, clockwise) while pushing it on. The flappy-wheel is effectively a huge sprag clutch with the drum as the outer race and the flaps as the ratchets/sprags. When the wheel is spun by the motor and the drum gets loaded against a floor, the flaps get forced open a little from the transmitted force of friction, causing them to push out against the drum harder, which causes more friction.

After I finished going “Well huh, that’s kinda smart’ I realized this design would get very good traction in one direction as each flap gets forcibly planted into the arena floor. The reverse direction might not be as spectacualr. Before I got ahead of myself with anisotropic traction designs, I decided to just imitate the flappiness  in my current tire design.

That’s the same helical tread sweep, just with many more slits of a greater depth and narrower width. I did this for “easy” (change the number and size of swept features) for the time being. I’d like to play with a straight-cut geometry like the floor scrubber in the future.

Printing this damn thing was an ordeal. Unlike the previous wheels, the molds could no longer be printed upright without support structures due to the way the helical threads are placed. Furthermore, the deeper tread features also prevent demolding in two halves, so I had to split the mold into quarters. Attempt #1 with supports was basically a no-go, since it was almost impossible to get them out cleanly and not leave little strands and hairs everywhere.

I next tried to print the mold wedges “pointy side up”, making a flat face on the outside of the circle for them to sit on. This was okay, but the nylon warped just enough on each print to make the edges not seal at all – this was attempts #2 and #3. I guess I could have made an Onyx mold too, since it has virtually no deflection, but by that time I’d mentally moved on.

The fourth and successful attempt was a single-piece mold which simply had the upper lip chopped off. I don’t even know why I thought the upper lip was needed now. Just fill to the top and be done!

That’s the model with the lip removed.

So how do you demold this damn thing? It was risky, but I decided the one-piece mold was okay because of the spiral nature of the tread allowed me to helically demold the cured wheel from the mold. And this ended up being completely true! I back out the hex bushing a half-inch, take a wrench, and untighten the wheel right out!

This worked quite well. Here are the two first wheels to emerge with the new material and tread! I stuffed the leftover mixed rubber into an old wheel/hub combination, because wheels are wheels. Notice the white core of the two new ones – they’re made from plain ABS, since I wasn’t about to waste the Onyx material on something I wasn’t sure could ever be removed from the mold. They’ll be on standby as low-priority spares nontheless.

Next up, the Forsch Polymers URS-5160. Forsch is one of those “Call Billy” companies that I always complain about – just go look at their 1997-chic website! Except this time, I was literally told that I had to call Billy (over in BILLING no less) and FAX him the order, then MAIL them a check. Credit card? Paypal? Pffff.       

Oh, Billy also leaves at 2pm each day, so I gave up after 3 days of failing to get in contact with him because I might have trouble waking up before 2pm on most days. Luckily, someone clued me in that McMaster’s general purpose pourable 60A urethane is manufactured by Forsch, otherwise I would have given up completely.

So why the hunt for a product which tries so desperately to not be purchased by anyone? Well, it advertises around 25% more tear strength and ultimate tensile strength than Smooth-on’s Simpact 60. Smooth-on is geared towards being easy to use – everything is made 1:1 or 1:2 mix ratios, so it doubtlessly sacrifies some strength and performance for convenience. I figured that polyurethanes worked like tacos – the shadier and harder to find that a Mexican restuarant is, the better the food. This has been almost bulletproof in my experience. I made it a point to obtain a Forsch product and use it like Robot Jesus Himself intended.

What I really want to try getting my hands on is the URS-2450, which has basically the same tear strength but in a 50A durometer. May Billy and I finally meet in the grand arena of procurement soon.

I cut new wedges out from AR500 plate. These were what they were meant to be, but I couldn’t get the material in 1/8″ (or 4mm-ish) thickness in time before Franklin. This was actually cut from one of Jamison’s spare plates left over from Sawblaze. I’m preparing them for welding here by grinding the incredibly thick scale they all seem to come with off.

Stay tuned for more, though with Motorama now 2 days away, I might just be updating after the fact! Still to come are the making of the pontoons, the spare lighter drive motors, and maybe a little bit of wheel testing!

12 O’Clocker & MassDestruction 6: Where I Rebuild a Bot After the Event is Done

Feb 09, 2017 in Bots, Events, Twelve O'Clocker


I feel like this website has become the Life of Charles, what with real editorials and non-stop round-the-clock van coverage and my tenuous professional aspirations… This is not the man I know. Where has he gone? *looks at own hands*

But now I’m back, with some new developments for Überclocker in preparation for Motorama coming up next week, as well as 12 O’Clocker stories to tell first. This bouncy little thing has been going to events and demos since 2013 with hardly any changes – just switching motors, basically. It’s gotten sufficiently worn down to a stub in the past few months that I decided to do a full teardown rebuild with some new parts!

To tell this story, we go back to the dark days of MassDestruction #5, like 3 months ago… Wait, CAN YOU BELIEVE THERE’S BEEN 5 OF THESE THINGS ALREADY? THAT’S MORE SEASON THAN BATTLEBOTS please take me back greg ;~;

This MassD, I took a more organizational role, helping judge and run matches. However, this didn’t prevent me from putting 12 O’Clocker (at the time, my only working bot -_-) into the arena in the somewhat informal 12lb Sportsman’s Class, where pretty neat matches like this occurred. MassDestruction has become a popular regional attraction; word has gotten out, and we pretty much filled out the Artisans’ Asylum event room to capacity. Like, look at this photo.

This is “filming a music video using the flashmob mosh pit at your post-phlegmpunk band’s free unannounced concert” level stuff. What’s better is that the builder population is getting more and more towards being newbie-dominated. This is a great problem to have.

12 O’Clocker came in 2nd place (out of like….3?) at this event, which was great, but it did take some damage. For the deterioriating ABS motor mounts that retained the lift motor finally gave out completely, wrenching the drill casing apart under its own torque:

Oooooh, that’s not good. I finished the tournament using a found drill motor given to me by an Artisans’ member, unceremoniously hot glue MIG-welded into the remaining mounting block pieces. At some point in the final against Snek Plissken, I also lost the lift motors which turned out to be one of the logic capacitors on the old RageBridge 1 units in 12 O’Clocker just breaking off the board. I also ended up demolishing another motor pinion just like at Momocon; the most recent set of motors for 12 O’Clocker came from some 12V Ryobi drill motors, and it seems like they were not up to the task of being run at ~20 volts.

Fast forward another 2 months, and MassDestruction the SIXTH! was on the horizon.  With the promise of more rematches with Alex Horne’s not-Sewer-Snake, I decided on a quick tuneup by replacing the broken ABS lift motor mounts with MarkForged Onyx prints because of course I did. New drive motors were also on the docket.

The Rage Panel slides out from the bottom, so I took the bottom plate off, which also let me do a hardware inspection on parts of the bot I rarely touch after finishing. This level of surgery was also needed to finally detach those ABS blocks.

So new drive motors were a bit of a conundrum. When 12 O’Clocker was built, it was still common to find generic cordless drill motors with 9-tooth pinion gears and 36:1 reduction (two 6:1 stages, 9 tooth sun, 18 tooth planets, and 45 tooth ring) gearboxes. Nowadays, it’s almost impossible to find these kinds of gearboxes, with 24:1 being the most common such as being found in all of the Harbor Freight 18v drills and most other rebrands. The 24:1 boxes have a 4:1 first stage using a 15 tooth pinion.

Trouble is, 12 O’Clocker was already geared to go fast, and dropping the gear ratio another 50% would have made it impractically fast and probably burnt out the motors in short order. Those Ryobi drill motors that I kept slipping the pinion on were attempts to find more 9-tooth motor pinions to fit the existing gearboxes.


After some haunting online, I found that one of my usual Amazon suspects uxcell sold 18v rated 550-sized motors with steel pinions already installed. Well imagine that, a prepared artificially flavored drill motor!  So I got a bunch to play with. They certainly look like 550-size motors and quack like them. The cans are a little thin, pretty typical of a mature Chinese genericized product… I can pick them up with a screwdriver. Every possible area of cost cutting has been well optimized!  The bronze shaft bearings obviously have no oil in them, since they are a little rattly, but a drop of motor oil in each solved that.

What I did notice was that the pinions’ press fits weren’t that tight. It was actually easy to undo them with a flat-blade screwdriver alone. To pre-emptively avoid embarrassing public gear slips, I took the pinions off and repressed them with a healthy dose of lime-flavored Loctite.

You know what – I’m just all giddy at the fact that they’re motor-shaped at all.

At some point in one of its tournaments, 12 O’Clocker either fell off a Dragon Con stage and landed on its main sprocket, or I got beaned by some flying robot, because the sprocket developed a flat spot which caused the chain tension to vary cyclically, leading to some lost chain moments.

In a moment of either desperation or brilliance, I decided to use my Harbor Freight slide hammer kit with a hook end to pull the sprocket rim back out like you would pound a very reticient dent. I bought this originally for van repair, but it looks like it works for robot dent pulling too!

Putting things back together, sans battery. The Ragebridge 1 with the missing capacitor had that repaired; the capacitor ripped out a logic power trace when it fell off, so it just turned the controller off. All the caps were securely Goop’d in place after the replacement surgery. If you’re using a Rage or a Rage 2, you should do this just in case also.

The biggest problem plaguing 12 O’Clocker was its battery, which I balanced once in 2013 and never again since. The cells had drifted far enough apart since then such that two of them flatlined at MassD #5, and I could no longer revive them. This meant I had to cut the battery open and undo the cell joints to the point where I could pull the two dead cells out and replace them with fresh cells. I closed the battery up again after this replacement (the green tape is new and covers the modded solder joints) with some thicker heat shrink, and making my charger do 5 overnight balance-charge-to-discharge cycles evened the cells out.

One last mod before MassD #6 was the permanent resolution of the clamp motor coming loose. The threads in the face of these Pololu 25D HP motors had completely stripped, so the motor was really just holding on by the electroweak interaction at the end. To remedy this quickly, I just took the faceplate off, slammed a #6-32 tap through them both, and then countersunk the original mounting holes in my actuator body. #zerosigmas is best Sigmas!

If you use these motors, or any of the similar motors from Servocity or Kitbots (or the straight shit from eBay), make sure to also take the motor off, clean the area, and use blue Loctite or similar threadlocker on the reassembly path. The motor does like to also wiggle loose – this is what the “battle hardening” mod offered by Kitbots helps prevent.

So anyways, it’s the morning of 1/28. Time to….

literally all the robots

This is what I trained for.

That’s Overhaul, Sawblaze, and two lift carts in the back. With space for another smaller heavyweight, or a dozen 30lbers and tools & equipment. And probably like 27 people. Some times it’s nice to just bring the U.S.S. BROWN C. STENNIS to an event.

This time, the event was held at the Charles River Museum of Industry, in one of their large event rooms. I once again helped with event logistics including box setup and judging. Overhaul and Sawblaze were brought along for visual stimulation, which was unfortunately because the event room has neither loading dock nor wheelchair ramp, and was, of course, a New England First Floor – 6 feet up the stairs.

Running 12 O’Clocker – especially when things started breaking – and half a robot show at the same time was a unique and singular experience. I will never do it again.

Have some 12 O’Clocker matches!

The match against Don’t Step on Snek, a.k.a Snek Plissken, a.k.a Sewer Snek… god dammit Alex, pick a name already!

By this point, 12 O’Clocker had lost basically all of its forks. They finally reached their fatigue limit at this event, one by one breaking off, until I had basically a big spatula. In the match prior, the right side motor pinion slipped its press fit as I had feared, so I went into the final match (also against Alex) one-motored. Which is fine, since Alex at this point had also started to run out of motors. The finals match was such a headdesking, facepalming occasion that I’m not even going to bother finding a video.

Poor 12 O’Clocker before the finals with the forks arranged the best I can, so SOMETHING AT ALL is still sticking out ):

Well, that’s it for the event. I broke the damn thing so much that I felt like I might as well use the momentum of the event to make spare parts. As I needed to also waterjet-cut spares for Überclocker, I threw on replacement forks for 12 O’Clocker in the same run.

Tearing down the bot completely up front to replace the fork components! This is where I discovered that despite my best attempts at anti-seize grease usage, the lift sprocket’s hub had galled onto the aluminum tube shaft, so the slide hammer was needed again just to break those two apart. I reamed all the shaft collars out again and cleaned up the aluminum shaft surface. This time, I tightened all the collars as much as I could – no longer relying on clutching the lift sprocket for torque limitng, but just setting the RageBridge current limit low enough that running into itself will not cause problems.

The new forks are slightly modified from the current design by adding more meat to the areas where the tie rods pass through. This was previously where they broke, so I made sure to add at least a majority of the cross-sectional area found in the rest of the fork.

By the way, this tube-removing service is also a problem with Clocker, especially after everything got twanged far up its own ass at the Franklin Institute event. I’m going to reconsider using a live shaft with shaft collar hubs to the forks for this reason, possibly considering a more Overhaul-like tie rod and central hub approach. Otherwise, I’m going to make an attachment for the slide hammer specifically for this purpose!

And here’s the refreshed 12 O’Clocker! Hopefully a staple of many demos to come.