Norwalk Havoc Robot League, February 2021: The Culmination of All of My Worst Nightmares

When we left off the previous episode, I’d just gotten the new attachments cut by Big Blue Saw and gotten them back. So with but a week to go until the competition, it was time to put all these pieces together and prepare the bot!

These pieces were cut from the same Hardox 450 (HX450) stock that I’ve had Overhaul bits made from since 2018. Basically, it turns out Kloeckner Metals, through whom I had Overhaul weldments-to-be cut out, does NOT reuse partially cut material.

If you buy one custom-cut part, you basically bought the whole plate. As a result, I asked them to slice everything else up into 12×24 rectangles so I can chew on them for years to come. The 4mm thickness in particular has seen application across a few 30Haul changes and even Roll Cake.

The nice thing about genuine grass-fed, wild-caught SSAB Hardox steel is that it’s very uniform in thickness and has a coating applied to it that grinds and sands off very easily to reveal clean metal. So much AR-grade steel is naturally bowed and comes covered in tough (abrasion-resistant, almost…) mill scale.

Above, I’ve already cleaned the pieces with a flappy wheel and have made initial tack welds to hold the shape together.

I did all of this with good ol’ Limewelder, which I discovered recently actually kicks some ass on 240 volts, so much that I’m going to get 0.035 wire accessories for it. I started off with 0.030 because of the need to do light sheet metal work on Vantruck, but on 240 volts it becomes the bottleneck for how much steel I can lay down. There’s some exciting Limewelder work to come yet.

More metal being smeared upon itself here. I generally decline to call myself a welder, as being a welder implies you’re well trained and practiced in the art and science. I prefer to say I host exciting meet and greets for metals.

The parts were joined at the tab-slot interfaces first to lock them in place, I suppose being a pseudo plug weld. Large fillet edges were done last and in a few stages to discourage warping. In this 4mm material, I can pretty much just turn Limewelder up all the way and draw lines – it barely has enough power to just make large stringer beads in this thickness, fillet welding in one pass. Any thicker and I’ve had to bust out my unsteady random-walk weaving, making everything worse.

You know what they say, a painter’s welding is a grinder I ate…. …..or something like that. The completed assemblies were primed and then painted in my favorite generic black robot spraypaint, Harbor Freight Spray-On Bedliner. I can’t imagine anyone actually using that stuff for a truck bed, but it does make a nice matte and lightly textured surface on robot parts that seems a bit more durable than the usual Rust-a-Color.

Hey, remember the “Open Loop Aluminum Donut”? Well, to my utter amazement, all seven holes of the bolt pattern went in without any encouragement. While I’m disappointed at not fitting digital scales to the Benchmaster, Master of Benches yet, I’m also happy to know I can always fall back on “well fuck that” machining techniques.

When all is said and done, here’s 30Haul set up in “generic” mode for Norwalk Havoc. I figure I’d change modules depending on who I was fighting first. The crash kit comprises a labeled spare hardware sorty-bin and a tote full of “Probably Most of Another 30Haul”.

This might be the first appearance of Overhaul 3 on this website, in real life. Because I basically stopped updating as WordPress died more and more, then picked up afterwards with moving and vans as I skipped BattleBots 2020, the entire design and build sequence of Overhaul 3 has been skipped. This is probably my biggest and most menacingly lurking back-blog that I have to do, and I already have the framework for it set up, but just have not mustered the willpower to begin writing yet.

The almost 1 to 1 relation between this 30haul’s geometry and that of Overhaul 3 can be pretty well seen here. No wonder I’m trying so hard to robot again, eh?

While I take forever to get to the write-up, you can see Overhaul get beaten by Sadbot over and over.

We now teleport a day ahead to somewhere in the dank industrial crotch of Norwalk, Connecticut. 30Haul’s been checked in and through the inspection and functional test now, so I get to walk around and check out the BRAND! NEW! NHRL building.

It basically expanded across the street from 50 Day which has hosted the insect-weight events. This is a brand new 30lber-rated cage, of a rather hybrid construction. It’s significantly sturdier than the NERC Motorama box, but is still made of wood for most of the floor sections.

NHRL’s thing is having a giant, slow house bot in the arena (which I support the concept thereof very fiercely), and I also identify with the general “there are no rules” interpretation of rules, especially as it pertains to multibots. In other words, this is not a place for the overly competitive to get uppity about following proper tournament procedures and fair judging. This is a quintessential #PostmodernRobotics experience.

A whole lot of gear and prep has gone into this facility – here’s a photo of the streaming battlestation.

So how did the robot do?

Horrible. Absolutely terrible by record, but I learned everything I set out to learn which, as I looked forward to this event as a shameless buffet bar sample for Overhaul, means I won without winning anything.

There’s not a split-up-by-match video available of the February NHRL, but there is a single stream capture that’s TWELVE HOURS LONG. I can’t actually be bothered to go through it to find the segments of video featuring 30Haul (entered as “This robot is not Overhaul”), so let’s get down to the details.

My first match was against Big Ripto of Motorama fame. Because _of course_ it was. I had no particular strategy going into this one, as it was more to figure one out. I’ve seen enough people do the keep-away dance with vertical spinners, and Overhaul 2’s “Wedges of Limited Liability” were made in 2018 as a way to decrease the attack surface against the likes of Witch Doctor and Warhawk (to varying degrees of success). But the 4 wheel drive vertical disc approach really came into vogue in the 2019 and 2020 seasons, which Overhaul had to sit out. Everything’s gotta be a Bite Force now, huh. In general, that time period was also when I was the most out of the sport as well, so I haven’t really had any actual match driving in over a year.

In all regards, this match went “fairly well” solely because Big Ripto suffered some mechanical and electrical problems. Kyle had issues with the belt drive on the disc becoming entangled, and it also melted a few wires. After the first few collisions, 30Haul had trouble driving much – because the Vex wheels had basically come apart and become little paddles.

On top of that, one of the good hits Ripto got on me actually managed to blow off the entire header on the 6-FET Brushless Rage. Why? It was the only one I somehow forgot to put the double sided mounting tape between the signal and power boards. The others were fine. This was, of course, on one of the drive sides, so between “flappy paddles” and dragging the left side, the rest of the match was a lot of Powerful Hugging.

The “Big Hit” pushed in the rear frame rail quite substantially, and this force transferred through the battery mounting bracket into the motor controller enclosures.

I was able to replace the closely-shaved Brushless Rage and flatten the frame rail enough for the top and bottom plates to fit back on. I learned next that 30Haul’s Losers Bracket match was going to be against Rum Ham, an undercutter kind of design. It’s a heavy hitter, but has historically been a little unreliable on the weapon. Nevertheless, it was a good chance to really test my assumptions when it comes to “T-rex Mode”.

So I began converting 30Haul over to T-Rex Mode. In real life (and for Overhaul) I would have this assembly pre-built and ready to be swapped in, or Bot 2 configured for it as soon as we know the first rounds of matchups. Just like Overhaul 3, the top half of the bot pulls off for independent service.

For the bottom half, I added the Snout using all of the available rubber bumper holes.

And the two halves go back together.

Well, it turns out 30Haul was about 11 ounces overweight in this configuration. I had to remove the (heavier than expected) Vex wheels on the outside, leaving just the chariot spokes. While we’d generally let even 3 or 4 ounces slide at most events, I personally felt almost a full pound over was just too egregious.

I was pretty happy with 30Haul’s durability this match, as I was able to keep the snout pointed at Rum Ham and cause it to destabilize a few times. I went for full speed connections as well – in a match with higher stakes, I’d be more likely to keep on the opponent, whereas I waited several times for Rum Ham to spin back up. Again, just seeing what goes wrong where so I can more effectively address it for Overhaul

I wasn’t able to leverage the situation, however, because one of the hits caught the very tip of the wedge and bent it forwards, which hampered 30Haul’s mobility a whole lot. It wasn’t _not_ driving, but was certainly not able to effectively maneuver.

In retrospect, a bottom brace/truss plate would have been more helpful than internal webs, and that’s one of the mods I intend to make right away to this design. I was otherwise very happy with the physics and the “rise through” functionality of the bottom forks. You can see that one of my “worst case mistakes” did occur, where Rum Ham caught the very side of the wedge and popped off a few of the rubber mounts on the corner.

Alas, due to my inability to follow through with the face-bouncing I did with Rum Ham, 30Haul was declared the loser of that round. I put it back together in “Sportsman Mode” for the 30lber rumble that never happened because the event had been running almost 14 hours at that point.

The next Norwalk Havoc with 30lbers is in May. So what’s the big plan?

I’m revising the Snout design to be stiffer on the corners, trading the internal webs for a U-shaped bottom truss shape. The metal is rigid enough on this scale that I don’t think the center webs contributed much.

Next up, after watching enough Battlebots and seeing enough matches at this point, I’m going to make 30Haul an entry in the “fork wars” currently ravaging the combat robot landscape. The joke is that ground-scraping forks grew a foot over the course of the 2020 season, as one of the only ways to ensure a vertical spinner doesn’t get under you, is to get under it first. Keeping the opponent away and under control is just as essential as being able to take (or deliver) the hit. There’s some amount of rumbling that BattleBots might restrict the “ground game” as it’s called, but we’ll see how that shakes out.

I’m also beginning to like the Vex wheels less and less, because while they initially offered the bouncy compliance I was looking for, once they start disintegrating, they become more liabilities. The spokes will tear through, leaving the big gaps in the perimeter that then flap in the wind. I decided to try and find some thick foam rubber I can use to make at-scale foamy wheels, not dissimilar to Overhaul’s.

All this and more in the next robot updates! Maybe when someone splits up the February stream capture I’ll link to individual videos.

A Return to Scale Model Testing Your BattleBots?! Überclocker/30Haul and Norwalk Havoc

We’re going to time-skip to the recent past and away from non-stop van content for a little while. It’s time to get back into some robots!

This isn’t to say I’m out of van tales – in fact, it’s actually gotten worse…. much worse…. but there’s plenty of news to report on the robot front and I’d like to do so before I become an off-brand automotive blog. I mean, more so than this site already is?!

So the event to be seen at if you’re a kool robot kid as of late is the Norwalk Havoc series which has been running as a beetleweight-centric event since 2018. I’ve mentioned it in passing a few times, usually in the context of Roll Cake and mechanical abject art I’ve made using its prize trophy. This event has been responsible for more or less creating an entire new generation of first-time competitors. It’s nice to have someone who is willing to bleed immense amounts of money onto the sport; the best way to make a million dollars doing robot fighting is, after all, to start with ten million.

After a while of doing ants and beetles, they decided to enter the 12/30lb “big arena” scene in late 2020 with the inaugural event taking place in February 2021. Well, with another BattleBots on the horizon and an Overhaul that was basically ready, I decided – hey, might as well use 30Haul to test a few more production changes before I make an even bigger financial mistake.

And so we begin. 30Haul was basically undamaged from the Before Times of 2019, and having followed the evolution of BattleBots Season 5 in 2020 from afar (as I decided that traveling out there just to hang around and mess with 10⁸ different COVID-related restrictions just wasn’t worth the potential to get detained), there were both some changes I wanted to make to the bot itself and some scale-model testing to be performed for Overhaul’s own future attachments.

First off, ever since the transition from “Uberclocker-shaped” to “Overhaul-shaped” in 2016, I’ve not had the ability to pull off Überclocker’s signature spin move. The long spring-loaded, roller-tipped legs let the bot move fairly well while holding onto an opponent. After I changed the design to become “30Haul”, the spring-loaded legs went away in favor of various wedge attachments to counter spinners (which I maintain are the cause of the decline of the sport).

Part of the architecture change of the latest Überclocker version 5 in 2019 with its broad multipurpose breadboarded front clip was so I can more conceptually detach the bot’s shape from its configurations. Overhaul v2 and its attendant 30lb version were both shaped with the frame rails extending past the lift axle’s fore-aft dividing plane, like a mild U shape. This limited what I could put there without making it vulnerable to damage, especially from the sides.

I specifically made the new “pontoon anchors” with pivot points space such that I could use either single-piece forks between them (like it currently has) or put something wider that straddles them on the outside, such as a new set of roller legs. I decided to push forward with adding these for the February Norwalk Havoc, since even though it was a “full contact” event, it would be good to have the Sportsmans-style configuration available (for the odd bot out these days that doesn’t have a small vertical spinner….)

And so, following the general pattern of the single-piece long wedge forks and using their construction sketch as a template, I started sketching out the profile:

These new roller legs were to stick out another half inch or so than the forks. Even with the longer wedge forks I ended up making, 30Haul was still prone to tipping forward if I lifted something too quickly (a syndrome Overhaul 3 itself also currently exhibits!). The longer extension meant these would really be limited to Sportsmans-style matches, but hey, nothing wrong with that.

After generating a profile, I mirrored the bodies to make the two side plates. In Inventor at least, my preferred strategy for making weldments is a multi-body part with fab files made using derived parts from the individual bodies. It’s only slightly terroristic but one of the rare habits I crossed over from Solidworks.

Completed leg design with web infill as two more disjoint bodies, and added fixturing tabs. Another reason I grew into doing this was the ease of doing cross-part (cross-body, I suppose) references for geometry such as these tabs, without actually having them cross distinct part files in an assembly. That rarely holds together long, for me at least.

Add a few nuts and bolts and here we are. The roller will be machined from Delrin, as per usual. When 30Haul is in this configuration, the rubber mounted pontoon anchors are instead going to be thru-bolted with Delrin spacers, keeping the compliance only in the (more rigid) single rubber bump stop.

Part of the success of Überclocker’s support legs is their spring compliance – freely hanging legs with just a hard stop make the whole multi-bot complex easier to tip forward, something Overhaul is struggling with now, as there is nothing to soften the initial rotation and the momentum of the lift will kick the center of gravity over the legs.

I expect similar changes to occur for Overhaul’s wedge forks – they’ll always have some amount of elastic compliance, versus just being freely swinging like they are now.

Here’s how it looks in this configuration. The (as-is) forks have a lot less prominence, but not really much worse than Überclockers-past were.

With that configuration behind me, I turned back to “spinner mitigation” strategies. Overhaul had historically featured two steel wedge assemblies I called the pontoons, named largely after the features on Overhaul 1 which were spawned from the then mega-Ron (and subsequently Sawblaze) contribution to the design.

Over the course of the 2016 and 2018 seasons of Battlebots, and from Motorama experience, I began to look away from this design. Sure, they were very effective in their job, but because they only covered a small portion of the bot, I had to drive very precisely to leverage them with the otherwise undesirable side effect of sending the weapon into the rest of the bot.

They also don’t distribute forces well, pulling on one side of the chassis. Recall that Overhaul 2018’s grudge match with Valkyrie debuted the “DETHPLOW” design which spanned both sides of the bot. I made short “T-rex” arms to fit behind them, and I do with I could have gotten to use this thing more during the actual tournament, because it did resolve the “focusing energy onto one side of the bot” problem…. but obviously not the “Fling Valkyrie into your face” problem.

I decided after watching the game evolve during the 2019 and 2020 seasons to switch to a more common bot-spanning wedge, a design I usually called the “Tombstone Snout”. It’s the present year, horizontal spinners are no longer your major concern. A rigid and angled piece of steel is sufficient to overcome most of them. It’s the verticals you really have to worry about.

I spent a little while thinking of ways to make the arms rise through the Snout. I didn’t want an open area with short arms any more, as that’s a lot of ingress surface for any opponent. So I decided to make narrow slots for a set of T-rex arms to rise up and through. For now, though, I’d concentrate on the (rather simple) flat geometry. As with Overhaul’s complex faceted wedges, I began with a wireframe sketch pile made of both 2D and 3D sketches.

From there, surfaces were made to turn the sketch edges into 2D facets.

I then apply a Thicken Surface operation to get the solid plate models. Depending on where the face is, it might be a thicken-towards-inside or outside, and determines if the piece is a driving component during assembly (place and fixture me first, basically).

I picked a good place to put the “T-rex slots” that don’t require repositioning the gear hub. Also notice the 45 degree forward-canted backstop on the top edge. That is your Valkyrie Preventer. Quite a few bots have this kind of rail feature on their front empennages, and I tried to live-add some to 30Haul before its match with BEAM (And they held on great until the end, which is when….. the thing happened).

The next step is to add the keying slots and tabs. It’s taken me a while, but I finally over the past 3 or so years became more comfortable with fixturing for welding. Previously a lot of my designs featured full-edge stitching of tabs and slots. While it was quick to put together for welding, it either 1. introduced a bunch of protrusions or 2. made the weld bead density high enough that I’d lose a lot of material strength to the heat-affected zone. Nowdays I tend to only key stuff together at the very corners and edges where they’d otherwise shift the most, or are hardest to fixture.

The final design after adding and adjusting some clearances with fillets. Notice how I trimmed off the outer corners of the wedge to slope downards to The Snout region – I wanted to prevent a glancing blow coming from the rear of this assembly, say I drove up next to or turned into some spinner the wrong way, from just blowing the whole thing off the front of the bot.

Next up was generating the new T-Rex arms. I went back to derive off the “master sketch” defining the bot for this one. These are going to be skinny on purpose, just giving me a modicum of lifting ability.

I knocked these out pretty quickly – edgy teeth for edginess, and a downward-facing tang that will allow me to use the bottom set of arms as a clamp of sorts if the actual clamp arm itself falls off.

Here’s what they’ll look like hiding inside the wedge. They live about 1/8″ below the surface.

Yes, I risk them not going back down if they are hit while raised, but at some point, you have to come down to the good ol’ LTFD: Let’s Throw Falafels Down, or so they say, and not attack an active weapon with the arms up.

The decidedly less threatening looking T-rex arms shown in a raised position. Immediately behind the center “barrier strip” is the singular standoff that keeps them together as well as serves as the pin joint for the clamp actuator.

Other small kibbles I needed to make for 30Haul – chain guide blocks, similar to those on Overhaul. The drive chains were sized for slightly loose tension, and they never fell off I guess but had become so sloppy I think they’d definitely fall off in any serious match. The guide just pinches them together very slightly.

Last up was changing the wheel hub design to one more reflective of Overhaul’s inner and outer wheels. 30Haul’s wheels were designed to be sacrificed and the bot geometry set in support of driving with one or more wheels out. In fact, at one of the Orlando Maker Faire matches, it lost a wheel almost right away to a flipper shot and I drove the entire match thereafter barely noticing.

That’s the kind of performance I want from Overhaul as well.

30Haul’s wheels tended to fall off as a single block since I made them use very long spacers that spanned both Vex wheels. Overhaul now has some very solid wheel hubs and chomky axles for the inner pair, and the outer (and optional) pair is attached via composite fiberglass-filled nylon bolts.

I changed the wheel and hub design for 30Haul to use aluminum standoffs for the inner pair and nylon standoffs with regular nylon screws. Hopefully this means the outer set has a tendency to shed first.

With all that CAD through, I sent out some metal cutting orders to my now hometown heroes Big Blue Saw (a proud sponsor of Overhaul 3!). Luckily, being nearby and all, I can just drop materials in my possession off instead of ordering through the website. I had leftover Hardox 450 from Overhaul’s build that I had the steel company cut up into 12″ x 24″ plates, one of which was perfect for one 30Haul’s worth of improvements.

Now, onto the actual bot.

It didn’t really need much work at all after Orlando Maker Faire, but it has been sitting on a shelf for a good year or so, and I slowly stole parts out of it (such as the TBS Crossfire rig) for other stuff in the works. Famous ending to many of my projects. So it was good to just take everything apart and give it the once-over.

What spare parts I didn’t make prior to Orlando Maker Faire, both from timing and the lack of heavy-hitter opponents, I went and knocked out now. This primarily involved more wheel hub parts, which were now a lot easier because I had a lathe more correctly sized for making them.

Some of the newly redesigned wheel hubs going together – this is the aluminum and real hardware layer, on the inner wheels. Notice there’s still a nylon 3D printed Fancy Spacer between the metal? I’d still rather these inner spacers bend and break off one by one versus take out the entire sprocket, if it went that far.

One wheel has aluminum threaded standoffs, the other unthreaded nylon spacers.

And they go together thusly, identical from the outside.

I finally get to make a _robot part_ with New Tinylathe, you say!? Well, New Tinylathe is also now just seeing its first year in my possession, so it’s about time. I had to make a pretty solid spacer to handle the torque of lifting, from the big lift gear to the T-rex arms since they’re spaced more tightly together and made of thinner material. The base part itself is an aluminum flanged donut with an 8-hole pattern. That is conceptually simple.

What’s less simple is the fact that I hadn’t picked up any metrology tools for the Benchmaster, Master of Benches yet, not that it deserves any. As mentioned before, I’ve gone back to my “vernacular machining” habits learned on trashy student-stop machinery at MIT and back in high school with friend’s shops. Running things into each other (carefully, with respect) is your edge finder. Open-loop holding the dials against backlash, and using Sharpie marks (who has time for layout dye?) is your “readout”. Actually using the “Normal” and “Loose” clearance fits for hardware is magic.

Trigonometry and symmetry are your friends as well. I managed to knock out this 8-bolt pattern (Of which only 7 are filled, as one of them presents accessibility problems for servicing – so fuck you) largely by feel. I’ll be damned if any of these go through.

I don’t need the entire donut for strength and would like the ounces back, so I had to cut out the center to make a spool-like piece. I did this by plunging multiple times with this insert cutoff tool, then just abusing it to make very shallow turning cuts at the bottom of the radius.

A couple of days later, the Big Blue Saw parts show up, so it’s time for some happy welding fun before I ship off to Connecticut!