Archive for the 'Done!' Category


The Overhaul 2018 Design & Build Series, Part 1: Because of The Implications

Mar 23, 2018 in BattleBots 2018, Bots, Overhaul 2

Here we go! This post is all about design and redesign. What I’ve decided to do is instead of making a master post of all of Overhaul’s problems and issues I wanted to address, I’ll just stream it as I go. This stems from sitting down to write such a post and then realizing that okay, I actually have a problem with damn near everything. I was 2000 words in and hadn’t even gotten out of the discussion of shortcomings.

One of the most important things you can learn as an engineer is proper project scoping, whether your own or on a team with others, and how to spot and pre-empt scope creep. Well if that’s the case, I just watched my oscilloscope grow legs and bail out of a window, which is a shame, since it cost a lot of money.

Yeah, there’s 1001 things I want to change about Overhaul, but I obviously don’t have the time-money product factor to make it all happen. So I whittled the list down to some of the topics I covered previously, as well as new ones that came about from observing Motorama.

There were both easy to execute changes such as simple part replacements, and also more involved design…… overhauls….. which needed my attention more from the start.

The easy stuff included

  • Higher gearing for less top speed and more control over its own inertia, experimeted on with Clocker.
  • Getting rid of the “rocker” drive configuration by using a smaller center wheel – I want all 6 wheels on the ground at all times if I can manage it, especially with the added torque of a higher gear ratio.

More difficult and needing frame mods and new subassemblies, we have

  • Moving to all Brushless Rages for controllers! This basically meant throwing out the whole electrical deck, which I felt was too over-engineered for Season 2 anyway with its full polycarbonate enclosure.
  • Changing the lift gearboxes to the BaneBots BB220 line, which I had used in a few consulting projects between 2016 and now and which was better suited for high-torque loads. Luckily, this is easy since they are the same physical size.
  • Designing up a bracing system for the frame rails where they are current unsupported between the center and front wheels
  • As seen previously on the first design post, moving to the new actuator design which is more compact and should be much more durable.

And finally, what needed comprehensive redesign:

  • A new clamp body (the Overhaul anteater-dolphin-horsecatbearpig-raptor head) to accommodate the new actuator design
  • Changing the arm and lift hub such that the arms are universal, not mirrored weldments. This wasn’t a problem during Season 2 so much as I wanted to maximize the swappability of parts – if I had to replace a left arm twice, I’d be out of those but not Rights. If the arms were simple flat weldments, they could stand in for each other.
  • Changing the armored wedge design to be more effective – and since Motorama, optimizing for a wedge-fight mode versus a spinner-mode
  • For the latter spinner fight mode, going to a full front-spanning plow arrangement.

Since I was already starting off on the actuator, I decided to move onto the next most logical place first, which is the things that actuator mounts to. When you’re doing almost completely freeform design, you have to anchor it some place. When I taught the EV design class, I called it “grounding” the design – basically, you have to start somewhere, so just pick one and come back to it later as the design evolves.

I had a few concepts sketched for the new lift hub. It basically had to be compatible with the existing arm and gear bolt pattern and be the correct width, so actually designing it was a matter of picking a fastening method which made sense and trying to get it lighter. By “making sense” I meant moving away from the engine head studs holding the previous setup together, one which I no longer favored.

Basically, the longer the threaded stud, the more you have to torque it down for overall rigidity in the things being fastened. I also couldn’t locate an easy source of replacement studs in a higher strength grade than approximately Gr. 5 in the 2 different lengths needed, whereas I could easily find very high strength bolts. Recall that the arms had to be flat now as a design goal, so the whole lift hub is wider anyway, so the bolts may be rather short (well, 3″ and 4″ versus 6-7″ of stud length)

So I made the decision to abandon modifying or adding spacers to the existing lift hub – more spacers, more places for preload loss – and make a whole new hub.

This is the object that resulted. It keeps the profile and spacings of the current hub, but becomes a steel tube weldment with threaded 3/8″ thick endcaps. The length of the barrels on each former arm is just added to the length of the hub, keeping the whole liftgear the same width.

Since I would most likely be the one welding all this up, I could add stupid shortcuts that real welders would probably shun me for.  That’s the reason for those weird little tabs on the bottom, so I can fixture the holes relative to each other easily!

After shadowing Overhaul 2′s principal welder Skunkadelia , and having done more welding in general for work related projects, I’m now more confident enough having picked up my welding skills again to design around it more. I used to despise welding, and still kind of do, but hey, somehow this robot is 50% welded steel shapes so I might as well learn to maintain it.

I kept mulling over the arm design as I decided to whip up a temporary Überclocker-esque wedge shape. Notice the lack of side-wubbie engagement at this point? I was thinking I could get away without them, but Motorama showed me quickly they were probably necessary – as many wubbies on deck as possible.

I included the sidebar in this screenshot to show the process of generating the surfaces and planes needed to define the completely-lacking-in-perpendicular-features wedges.

Adding the backstop plate in a similar fashion to what I was planning for Clocker. After raw sketch lines, I generate surfaces using the sketch profiles.

I then use a Thicken operation on each surface in turn to make them into solid “plates”. At this point, there’s no joinery, because I was just interested in pounding out the shape.

In continuing the “nearly a visual mockup” theme, I then moved onto the fork arms themselves. I decided to dispense with the “tube skeleton and welded plates” method for not really being beneficial over just being made of plate weldments. The tube cuts were always less precise than the waterjet- and laser-cut plates anyway, and part of the reason Overhaul 2 was missing the inter-fork bracing standoffs Clocker had was because none of the holes ended up lining up.

I also committed to having all of the new steel parts made of commonly-found 1/4″, 3/8″, and 1/2″ steel plate. High strength steels like AR400 are far harder to find in thicknesses thinner than that. I had a lead on 3/16″ material fine, but sourcing the 4mm plate last season for the fork and clamp sides was an adventure. Very few steel companies list it as a product, because who needs a 1/8″ thick dumptruck body or mining shovel? It also gave me the option of making things from normie steel (mild/hot and cold roll low carbon) to just use geometry over sheer material strength as a design rule.



The forkss are now fully made from interleaved plates. Here I’m playing “connect the dots” with tube bosses and gusset plates. I made this plate adaptive so I could still change the  sizes slightly and not have to manually regenerate it.

Here is where I decided to make the forks from plain mild steel. I had the opportunity and supply chain to use AR steel for them as well, but decided against it. Essentially, if the forks were very rigid, they had the potential to easily wreck everything upstream (the lift shaft and hub, the towers, etc.) if somethng hit them directly. With a few geometry changes, I could get the forks to be rigid enough to take the potential downward force from the clamp.

Again, with the basic shape of things established, I went onto the next piece.

I started thinking about “Limited Liability wedges” for Overhaul about the same time as when I cooked up the idea for Clocker. Overhaul’s front is a slightly different shape than Clocker. The chamfer that forms the sloping face is smaller and so the frame rails of the bot extend further forward.

This means there was only so much liability I could limit if I kept the tall rubber wubbies, since by design, they kept the old style wedges flush with the forks. While I probably could have hard-mounted them to keep everything low-profile, I wanted to keep that level of shock isolation and break-away behavior provided by the wubbies in case an attack on  a vertical weapon goes wrong.

I therefore switched the design to a shorter wubbie style , which through the angle of the sloped front, set the ‘intersection line’ of the slope with the ground back a good 3 inches or so. This finally let me have enough fork prominence to make it worthwhile.

Above is the quickly thought up revision 1 of the design, which was basically “make slope piece, add pokey thing”. The idea was fine, but obviously that pokey wasn’t going to last long as a piece single-supported from a plate.

I extended the slope and front vertical pieces off the edge and extended the “pokey thing” as a gusset piece. Now this looks better topologically.

With basically all the elements of the bot that I wanted to redesign in place, I mirrored things and imported the Season 2 clamp to get a first visual.


This let me have a sense of which components needed to be bumped and shifted. Overall, the forks are 1″ longer than they were in Season 2. I wanted more prominence in general in order to attack with them first. These wedges are there only to help stablize the bot in most lift instances.

My goal for the clamp profile was to shift it forward more and also make it a little lower profile. Overhaul’s Season 2 clamp design was almost exclusively to house the huge actuator, and it actually looked a little ungainly to me. Plus, the taller the clamp, the more likely it is to get stuck sideways (which the ears are actually placed to prevent).

I simply exported a copy of the clamp profile and ripped most of the features out of it, including the awkward bump of motor-saving.  The hole positions shifted around a lot with the geometry changes, of course, so they had to come and go as I needed them.

After a few tries, I found a near perfect alignment of holes that actually let me mount the actuator in a position I long wrote off – with the motor pointed forwards. This configuration was modified slightly to adjust the motor spacing. It allowed the use of a very short leadscrew to achieve the range of motion desired, so I was much less concerned now about leadscrew rigidity.

Now with 100% more ear. I re-imported and adjusted the ear model, keeping it as a folded single piece. The plan was still to make the clamp sides from 4mm AR steel. I also added an alignment tab for welding. This was the extent of the work I did before Motorama.

One of the first things I did after Motorama was go back and edit the heavy anti-spinner wedges. Clocker doesn’t have the two front wubbies, nor the two sides, and I became even more convinced that Overhaul’s triple-plane constraint was a good design choice. Future Überclocker will have a more design-true layout rather than being a cartoony model.

I made a multi-faceted edge that was just a flat plate where the side wubbies are mounted. Recall that Overhaul’s season 2 wedge pods had fully angled sides, and I had an interior gusset piece which mounted to the side wubbies. I decided to flatten this area out in order to give even less things to grab onto – the prominent gusseted back edge was pretty much an invitation to getting the whole thing ripped off forwards if I miss an approach. Reinforcement of the area would be taken care of by a cross piece (green outline near the front wubbies) and a flat lower gusset.

This was the result. I rather like it – it adds to the new edgier aesthetic of the bot and is much lower profile and less bulky looking than the Season 2 wedges. The 2nd-angle transition to the pointed backstop plates is also much sharper, hopefully adding a stronge upward vector to someone’s deflected shot.

At this point I started becoming weight-paranoid, and so I just rage-added every remaining major part of the bot I could think of. What’s not shown is a pattern of several dozen (if not over a hundred) socket cap screws that make up the frame hardware.

There was also beginning to be a large weight difference on the order of 7-8 pounds between the heavy anti-KE configuration and the lighter wedge-match and vertical weapons configuration. I was just going to let this play out, since I could always add some kind of ballast if needed or make a smarmy lawn-care attachment to fight Hypershock with, but I wanted to leave a healthy margin for the improtant parts.

Notice something about the pointy wedges? I swapped their sides on the bot design! This obviously also works in real life, and is partly the reason I chose the design I did. I could even go in lopsided if I mounted two left or two right ones. An “arms close” configuration like this is that I would imagine doing for a fight against another lifter/flipper, whereas forks apart are what I would take on vertical weapons with since I can try to make close flanking passes and try to get under one of their corners.

Finally, we get to the crux of my Motorama ruminations.

I pulled the CAD back up one day afterwards and went alright, that’s it… this is happening and I will make weight for it no matter what

Basically I took the existing heavy wedge design just completed and bridged the two halves with a single U-shaped front plate. Reinforcing features have not yet been added underneath, but they will be.

This added about 8lb to the front – a little less worse than I thought, really, and that was without selective weight reduction cutouts. I like this already.

As I said from my Motorama conclusion posts, having a full-span front wedge probably could have turned my tournament around. On Overhaul, it is also wubbie-supported in all 3 coordinate planes in both compression and tension as opposed to just a flat plane of wubbies. It would take a lot of me fucking up in order to lose this thing in battle. The team started nicknaming it the “dethplow”, so Dethplow it is.

The plow also allows me to address another Motorama quibble, which was having to back up and attempt to bring the fork down. I made another configuration which I termed “T-rex arms” since they are half the length of the full length forks.

They fit fully behind the plow when stowed, but if I have someone trapped, I no longer have to back up, but can clamp and lift as normal.

rawr i am oversaur-cathorsedolphin-shark…..bearpig

The clamped-opponent orientation does change a little since the tooth now has significant overbite, but whatever – spinner matches are a matter of survival, not looking good.

I tabbed everything together and added five longitudinal gussets and one transverse rail to brace up the front and underside. The total weight of the dethplow is around 30 pounds!

Next up on how to CAD an Overhaul: Moving through the other systems making the revisions I wanted to do.

Motorama 2018: How Not to Scale Model Test Your BattleBots, The Remix

Mar 18, 2018 in Bots, Events, Overhaul 2, Überclocker 4

I had originally intended to go to Motorama 2018 solely for #vantruckjustice and to serve as an event volunteer and purveyor of Ragebridges and Brushless Rages. But with the announcement of Season 3, it became clear to me that I really should take the opportunity to get some practice driving in with 30-haul a.k.a. Überclocker v4.

Not only was my list of “things I didn’t like about Overhaul” extensively long, but based on my experiences after Franklin Institute and Moto 2017, I had several mental strategies against KE weapons I wanted to try out. Better do it on the small scale where it’s less expensive, right!?

So onto modifying Clocker!

One of my recurring themes in the past few event reports where I ran Überclocker has been the idea of reducing my wedge cross section against vertical drum and disc style spinners. I’ve had a theory for a long time now that broad armored wedges/plows are actually a liability against those kinds of weapons, despite being more effective against horizontals. There’s nothing better to confirm my theory besides Blacksmith vs Minotaur – in which Blacksmith actually does quite well against Minotaur until the latter manages to land a square hit upon the front of Blacksmith.

Essentially, a vertical weapon will tend to bend up your defenses by hitting it at a single point along its bottom edge, effectively making the length of the plow/wedge useless especially if the vertical weaponed bot has a feeder leg of some sort.

One countering strategy is going fully vertical with your defenses, like a crossing vertical bar of steel or something, making sure you hit the weapon before any feeders are able to touch you. Whoops! used this reasonable effectively against totally-not-Minotaur for quite a few hits.

That doesn’t quite work for me, though, because Clocker/Overhaul both have lifting forks.


Another strategy is minimizing your cross section ot exposure to those weapons by being extra pointy, giving them less of a chance to hit something important. This is also a strategy that I began moving towards with other ‘wedge fights’ – a broad surface is, again, vulnerable to any imperfections in not only itself but also the floor. I wanted to explore this strategy with what I call the “Wedges of Limited Liability” seen above – basically turning the armor pods into little shanky forks. I designed them to follow the profile of the existing wedges, out of the same 4mm-ish AR500 material.

These are a few ounces lighter apiece than the regular ones, which is going to mean a couple of pounds at the Overhaul scale. So it was interesting to begin thinking of the configurations I could get – freeing up a few pounds on Overhaul could let me add other attachments or additional armor (e.g. if I had to face Beta again, I’d spend the extra pounds on top armor).

While I was at it, I also redesigned the normal heavy wedges in the style of Overhaul. I had thought about ways to retain the double-angle feature but significantly reduce the number of pieces needed to construct it. Overhaul’s wedges were rather complex and made of 9 individual pieces each. I came up with an idea of making the second angled facet into a ‘backstop’ of sorts, attaching directly to the outside surfaces through extended tabs that also acted as gussets.

In essence, the above is what Overhaul’s new wedges will look like, but with some geometry placement changes. Clocker’s front is a lot more tapered than Overhaul’s, meaning the ‘backstop’ begins too far back to be really useful here to protect the gear from another Glasgow Kissing. I was more interested in the construction and their potential behavior towards deflecting  hits in general, rather than specifically trying to address last year’s weakness.

One change that has been on the docket since Franklin late last year was changing Clocker’s gearing. Using the lowest RPM/v rated NTM 42mm motors was still too much – I rarely exceeded 50% stick travel while driving, and the constant burnout mode the motors ran in during each match made fine control actually rather difficult. To give the bot more control over its own inertia, I was going to go to 11:1 P60s (from 4:1) but with a bump in motor Kv from 650 to 750. While this reduces Clocker’s nominal top speed drastically to only 10mph from like 20, it was going to mean more speed in a useful range. A lot of my strategy relies on being able to carefully control my approach and orientation to opponents, after all.

The same changes will be carrying through to Overhaul, but less drastically – I’m changing only the external motor sprockets, from 15 tooth to 12 tooth, moving the system reduction from 8:1 to more like 10:1, which is what Sadbot has been running in testing and whose maneuverability characteristics I like more. It will reduce Overhaul’s nominal top speed from 19mph to 15mph.

By the way, the NTM Propdrive 42mm series have a Mabuchi RS-700 size bolt pattern and a 5mm shaft, which mates with the Banebots P60 700-series motor blocks, not the 500 size.

Since I finally blew up the clamp collar joint at Franklin, I drilled the sheared bolts out and replaced them with a pin drive. The holes in the gear were bored out, and the “pins” in the shaft collar are actually shoulder screws with their heads milled off! I literally tightened shoulder screws into these holes and then clamped against the screws on the mill and blazed the heads off.

The shorter 4238 size motors, in combination with the 2-stage gearboxes, actually end up at the same length as before! This package is fairly potent and is a brushless 30lb drivetran of choice currently. The same two beta-version Brushless Rages still run the bot, dating now all the way back to Dragon Con 2017 and carrying the bot to victory at Franklin, then several demos at MassDestructions.

Fast forward a few days and an uneventful (!) vantrucking trip later, and here we are at the event:

So I didn’t get a chance to actually weld up the new heavy wedges – they’re shown sitting in a pile next to the bot, ready to become accessories. Since the only welder on site was a 115 v MIG welder powered by a 50 foot extension cord, my plan was basically to add some little barriers to the existing wedges in the form of strips cut from the new design.

Clocker’s first match was actually a “wedge fight” against the twin over-powered wedgebots of Boom Boom. This match was conceptually easy, but a little frustrating because Clocker kept digging into the wood floor. It was in fact too pokey and I could barely maneuver forwards. So it really wasn’t that good of a test of the Wedges of Limited Liability at all.

My next match was against Botceps, a pretty classic vertical eggbeater style spinner. Built pretty much like a BattleBotsIQ/NRL archetype, it had a very potent 50mm TPPower inrunner on the weapon. This was one of those “if I make a driving mistake, I’m kind of done” matches, so I had to be on my toes. I don’t have a video link of my match here at the moment, but should I discover one, I’ll add it in.

As you can see, I did end up making at least 1 driving error – at the beginning, a few seconds in where I missed a charge and Clocker went halfway across the arena. Luckily, the weapon motor fizzled out barely a minute in, but I did lose the tip off one of the forks and the left pointy-wedge.

The punt that send Clocker up and over also squished one of the outer rail mounting screws clean out of its hole.

At one point, I managed to execute my anti-vertical spinner strategy well, plunging the pointy-wedge straight under the weapon of Botceps – which managed to machine the entire row of rake teeth off it! The match ended with a serendipitous flip which put Botceps on its face (“doing the thing” in robotland) and without weapon torque, it was stuck in that position.

My next match was against Crippling Depression, a pretty innocuous-looking bot (by design) that actually puts a massive amount of power into the undercutter disc – two NTM 50-series motors, which is more motor than Glasgow Kiss. By the way, its builder Robert Cowan has a very detailed video series on a lot of engineering subjects, including full video reports of the build of CD. Basically if I were more keen about doing video versus text, I’d be like that! So go Like and Subscribe™

Clocker didn’t make it out of this match all that well, and I ended up tapping out.

The disc of CD hits very hard, and very low – lower than most of my side rails, so it polished off a lot of the bottom screw heads. It was also positioned just at the right height for Clocker to barely ride up on it, meaning I would actually lose a head to head pushing match. With brushless drive and weapons, we’re now squarely in the era of 30lbers with absolutely no compromises – CD and Clocker have identical drive motors and identical gearing!

The super low level of impact meant a lot of extra stress was put on the wedges here, and it never actually got to ride up to the point of hitting my welded barrier strips. In fact, the first few hits managed to bend down the corners of the left side wedge enough that I had some trouble driving afterwards due to it being hung up on the arena floor on occasion. It also caused the rubber shock mounts to shear off early, leaving me with fewer defensive options as the match wore on, until both the wedges came off and my only real strategy was to try and stay on him. You can see some of the resultant impacts causing a little bit of “Cobalting” on the right side (upper of image), but the middle tie plate kept this very limited and I didn’t even notice while servicing.

Eventually, the disc ended up shearing out a few sprocket teeth and jamming the drive on one side, so I just wiggled my way around until I decided I got the idea and should probably keep it repairable for the loser’s bracket.

This was about the moment I realized that disc weapons (and by extension, shell spinners and other lower-prominence weapons) were going to be a much different story than the archetypal big bar spinner in terms of how to fend them off and deflect their energy, and I decided Overhaul needed a full-span front plow style defense no matter what. Watching Jamison fight CD later on with megatRon confirmed this belief even more. I’m pretty sure if Clocker had a full span wedge, even a connector plate between the two halves, it would have gone differently.

I was losing wubbies at an alarming rate and did not have any more spares, so I had to source them from other builders. Clocker was short spare parts in general – I never got to waterjet additional frame parts or cast new wheels either, since the decision to go compete was made in very short order. This event was actually a great study in how far the design would go on attrition alone.

That was all for Saturday – on Sunday, my first match was against BEAM, a tiny Tombstone. A rookie bot that was still BRUSH-POWERED and with an EV Warrior motor at that! Basically running a classic car in battle, but it had done tremendously well up to this point.

Poor guy was probably being gunned after by every BattleBots competitor there, who saw him as “tombstone practice”. Honestly, so did I, because it could portend the results of #season3.

For this match, I was only able to get 4 wubbies per wedge, and changed out to another used wheel which didn’t have huge chunks gouged out of it. I also cut off the bent tips of the forks so they were a bit shorter, but now much less structural than they already weren’t to begin with. It is what it is, given the lack of preparation.

So how did Overhaul do against Tombstone?

That’s, uhh, not very typical, I’d like to make that point clear. Well, it seems like the front fell off again – much more epicly this time!

With as much handicap as the bot was facing, I sort of ‘drove for broke’ in this match and was determined to see how pure attrition would play out. I think I was actually quite happy with how I was able to deflect Beam repeatedly, even getting it to do The Tombstone Dance a few times.

I felt like I had Beam reasonably until roughly the 2 minute mark when one motor was knocked off internally causing it to lose drive one side. I then just kept pivoting to try and meet it with the wedges.  Also, I again kept the arms up to try and keep them out of the way of the bar, but ultimately they still ended up in Full Dab; these arms for Clocker were built quickly to Sportsmans’ class specs, so any sideways ping is going to bend them.

Near the end, though, one of my welded barriers came off due to lack of penetration from the event welder. At the last possible second, we got in a good head to head charge, and….

Clocker’s frame rails are made from 7075 aluminum, which is really a mistake. I already had the plate when it was being designed, so went for Easy first. 7075 is more brittle than 6061, and will crack instead of bending. You can see that clearly in the arm tower that took the brunt of the last hit. This is why Overhaul’s entire frame including the arm towers are 6061, and only the liftgear and clamp & actuator aluminum housing parts are 7075.

From watching videos in slow motion, Beam was able to climb up the de-barriered wedge much like Glasgow Kiss and firmly planted the bar into the side of the head, which of course shoved everything out the other side. The lift gear was also made of 7075, and I lost a chunk of it near the end – check out how clean the shatter line is. It’s barely bent at all along the rim and still sits quite flat on a table.

The ears were also obviously very suboptimal – they were re-printed, but weren’t bridged or braced and so had the same kind of failure when the bot landed upside-down with an opponent – they simply bent the aluminum clamp sides and rotated, making Clocker adorable and droopy for most of the match.

With the base still working fine after the drive motor was reattached, I entered the Sportsman’s rumble to run around like a dumbass. The P60 motor plates only have two screws, not four, so it was asking a lot to hang the entire drive motor off them. This is actually why Overhaul’s motor assemblies have bracing plates behind the motor endcap. Dumbassery was achieved until Pitter Patter sniped the power link with its sawblade.

So that’s Motorama 2018 for Überclocker. Going 2/2 against three heavy KE weapons with almost no spare parts was certainly more than I expected, which was more along the lines of instant vaporization. I think I confirmed about every fear I had for Overhaul, whose upgrades were almost done at this point otherwise, but now needed revisiting.

but what about the Implication?

I spent the joyride back from Motorama consolidating everything I learned while running Clocker and as well as watching other matches including the final few fights with megatRon, Beam, Cripping Depression, and others. As a fair percentage of the builders at Motorama were also building for #season3, we did some bonding at the event and over pizza dinner Saturday night to consider strategies for the “full size” bots.

In terms of the knowledge gained from testing, it’s a fairly established rule of thumb in the community (if you ask) that “designs don’t scale”. The sentiment is you can’t expect to scale up or down a design 1-for-1 in terms of material sizing and dimensions of weapons and motors and expect the same kind of behavior. It’s a consequence of a whole lot of square-cube laws: motor power scales by volume, kinetic energy stored in a spinning weapon by square of velocity but also implicitly square of weapon dimensions due to moment of inertia changes, material strength both by dimension cubed and linearly by yield strength, etc. It’s why historically speaking, a scaled up or down version of a successful design might perform horribly.

Essentially, the idea is that at the small-bot scale, the energy transferred and dissipated in a hit tends to be much less than the energy needed to permanently deform a material of a given strength and size. This is how 3D printed 1lb and 3lb bots fly around arenas and bounce off walls on a whim and keep going. It’s related to the concept of why you can drop an ant off a building and have it survive the fall, but not an elephant. As robot dimensions increase, the kinetic energies stored in weapons – whether spinners, or transferred in a powerful flipper connect, or in the form of a hammer tip – begin overtaking the ability of the material to elastically deform and dissipate energy, so you end up with a lot more things bent out of shape rather than two heavyweights suddenly reappearing at the other side of the arena.

I mentally call this idea the “robot Reynolds Number” when comparing designs of different sizes: to get the same physical behavior, a bigger robot has to be simultaneously more powerful yet built more rigidly. Consider it characterized conceptually as the ratio of average kinetic energy transferred per hit in a weight class to the material yield stress * volume used in your robot ( KE [J] / σᵧ [Pa] * volume [m³ ). The best example I can think of is probably how Big Ripto and Triggo (30lbers) can both bounce around arenas like beetleweights – but both robots are made of hardened steel when it comes to the bodies in contact, whereas plastics or even aluminum is likely to just deform in the same application; as well, they both cram about 4 to 5 kW of weapon motor power behind them, which is actually more than an order of magnitude from the typical beetleweight (3lbs).

What this means for me is that I have to be careful with interpreting the results of my matches. For instance, it’s highly unlikely that Overhaul will get sent flying end over end from a single spinner hit such as that from Botceps, but rather depending on what gets hit, I’d lose one of the wedges or have a pretty big chunk bitten out of the frame. The best-in-class KE weapons in BattleBots right now run right around 15-20 kW and around 50-100kJ. Going back to Blacksmith vs. Minotaur, you’d expect with strength-invariant scaling that Blacksmith would easily hit the box lights, but rather what happens is the frame/wedge deform and fasteners begin failing.  Also, in my match against Cobalt, my one good deflect was the end result not of being ultra-rigid, but backing the hit with the arena floor by virtue of the rubber-suspended wedges, which is obviously something I want to keep.

While the exact physics won’t carry over through scaling, concepts will. For example, I am fairly confident that

  • Weapons of low prominence, such as discs/drums/shells, are best kept away from you and interacted with lightly since they are less likely to grab entire portions of your bot at once, rather chipping away at it. This is best illustrated by how megatRon was able to keep Crippling Depression at bay with a single low front armor piece, whereas CD had more inroads to damaging Clocker’s separate armor pieces.
  • Weapons of high prominence, like the archetypical spinning bar or single-tooth style weapons, should be more readily deflected if possible, since they have more potential reach (i.e. far more “bite per tooth”) and less predictable reactions. You want them to go away from you as much as possible. I rather enjoyed a lot of the driving against Beam and making it do “the Tombstone dance”, and even the small welded barrier strip made a lot of difference until it failed.

Ultimately, I had to rethink Overhaul’s armor approach and how it interacts with the lifting forks.

  • I came away 1000% convinced it needs a Full frontalplow-like surface up front. It’ll be heavier – I will make weight for it somehow. One downside of separate wedges is only one set of rubber mounts takes all the load of an impact, and while my Cobalt flip was perfect, 99% of hits I take won’t be that perfect. A spanning plow will allow the mounts on both sides to take the load. Besides that, it will obviously decrease the amount of open corners Overhaul (and Clocker!) has. In fact, the plan was for the original Overhaul 1 to have such a thing, but we ran out of weight.
  • I had to retain the ability to lift independently of being able to deflect hits. Did you see me pin CD and Beam against the wall several times, but having to back off to try and get the forks underneath them – which in both cases had bent up beyond the point of usefulness? That’s how they escaped and the match continued. I need the ability to corral spinners against a wall but keep the arms tucked behind the plow, maybe exiting via a small cutout. Overhaul has a set of short arms which remain behind the wedge profile – imagine a cross piece in front of them connecting the two wedges.
  • Beyond just dealing with horizontal KE weapons, Overhaul needs a “Wedge of Limited Liability” of its own, which still supports the bot during a lift but otherwise takes the form of a skinny fork or tine so vertical weapons and things like flipper spatulas have less edges to find. I didn’t get to exercise the WLLs of Clocker this time against a vertical drumlet weapon like Other Disko, Mega Overload, etc. but I did like piking it under Botceps. Depending on the length of the tines, it could be effective on its own by being jammed under vertical spinners, for instance.
  • I need to learn to drive “dirtier”. What this means is foregoing my desire for continous showy and aggressive action – something I am used to in the 30lb Sportsman’s class – and instead maximizing my usage of pin times and arena-outs. Jamison is a more strategic and methodical driver me in this regard – he drives to win, whereas I tend to drive to shitpost. Compare his style when fighting CD (and other heavy weapons) versus mine. In the BB arena, survival is going to be key since we are more in the realm of throwing elephants off skyscrapers than insects. The NERC arena doesn’t really have facilities for arena-outs, but the BattleBots arena does, and I think it will be a key portion of my strategy. Bounce bots up and away, keep them corralled, and try to manipulate them behind the low walls and screw embankments.

Once I got back to the shop, it was time to refactor Overhaul’s design a little. Stay tuned for those updates!

My Life is Ruined Again: BattleBots Season 3 and the Triumphant Return (?) of Overhaul

Feb 21, 2018 in BattleBots 2018, Overhaul 2

> mfw season 3 announcement

The rumors began shortly after July, when Science Channel announced it was going to pick up BattleBots after ABC unceremoniously shat us out in favor of a …. boy band show? Well fuck me sideways with a fracking well, look at how that turned out for you guys! The rumors intensified in November as discussions and negotiations were clearly under way, and reached a crescendo in January, each week leaving us wondering if “next week” was going to be it.

Well, now they announced it. Crap. Now I actually have to finish something!

Overhaul’s upgrades have been in in the works – albeit slowly. After season 2, I had a whole list of changes I wanted to make and “design regrets” resulting from the extremely fast build season and required turnaround time I wanted to address. Really, I (and a lot of other builders) see #season3 as a chance to do Season 2 “correctly”, addressing things that didn’t go the way we want or designs that could have been done better. And frankly, anybody trying to build from scratch for the season now is either a dumbass or more of a man than I

that apparently ain’t hard

So that’s where we are now. The story of Overhaul upgrades actually goes back to right after Season 2 ended, and starts with what is basically the last large mechanical assembly that was designed, the clamp actuator…. meaning it was the most rushed and horrifying.

Ball screws were a bad idea. I was attracted too much to the promise of 90+ percent transmission efficiency, but they ended up being too fragile and also had the nasty habit of backdriving – made most obvious in my match against Beta. During the following 3-bot rumble with Sawblaze and Road Rash, the ball screw stripped out almost completely and began acting like an Acme leadscrew anyway.

Trust me, that hurts me viscerally to look at.

There was also a confounding problem with the actuator design and the clamp arm. In general, the actuator ended up too bulky to hide effectively without making the head ungainly. Because of the positioning of the motor and the bulkiness of the ball screw, I chose to simply add a little ‘horn’ to the clamp arm (the protrusion close to the pivot point) in order to protect the actuator motor from being landed on if Overhaul got flipped over.

In order to get the clamp running again quickly in case Season 3 happened relatively soon (*ahem*) and to explore the large Acme threaded rod market, I actually designed and machined up a retrofit using 7/8″ Acme screws and nuts – the odd size was for the easiest fitment to the existing actuator bearings, since the root of the 7/8″ Acme thread form required minimal machining to fit.

Also, I found the nuts on sale on eBay for like $20. There’s an engineering justification for every spur of the moment purchasing decision.

I wanted to redesign the whole upper half with a new acme screw based actuator to solve this. Furthermore, I wanted to move from a live-screw design to a dead screw one, where the actuator contains the mating nut within large carrier bearings and simply rides up and down a stationary screw, which is the design I’ve historically used for Überclocker.

The premise went from using the higher efficiency transmission option to the more durable and simple one and just overpowering the everloving fuck out of it to get my desired closing forces. As a large portion of combat robots revolves around the latter, it was clearly the way to go.

I cloned the Overhaul 2 CAD model into a new directory so I can start messing with everything. Here we go!

This is the actuator in its current position in the bot.

I wanted to try and see if I could move to a pull-stroke closing like Überclocker has been running. In general, the answer is “not really” due to how far the actuator will stick out into the ‘grabby zone’. In Überclocker, I sacrifice a whole lot of leverage to position the actuator almost vertically so it’s much more out of the way. I wanted to not make that sacrifice for Overhaul unless I had to, or if it were super convenient. functional requirement: be lazy

I also investigated the idea of flipping the thing upside down. In this configuration, if the trunnion tube is made non-offset (inline with the leadscrew) the motor will unfortunately hang down very low into the ‘grabby zone’ and be vulnerable.

All of this position testing though was enough knowledge for me to begin hashing out the next part of the design.

For now, I just imported the model of the P90X gearbox which was never quite implemented. Into the same model, I imported a bearing I bought on McMaster-Carr which I got curious about while specifying new thrust bearings for the this thing.

These are “one piece” ball and tapered-roller thrust bearings, so-called since McMaster usually sells thrust bearings in little kits of 2 washers and a basket of round things. Don’t be fooled, though… the “one piece” part is just a stamped sheet steel shell that holds the two bearing halves vaguely together.

The one on the right is a ball bearing based one, and the left one is a tapered roller bearing which is basically tapered the ‘wrong’ way compared to a normal one. This means it can support almost no radial load but a ton (or approximately 7 tons) of thrust load!

I found the tapered roller bearing one a little janky, though. The full roller complement meant it had quite a lot of drag when rotating, and the packaging was a good 3/4″ thick. There’s also no way I can reasonably use its 14,000 pound rating …. and that’s an industrial rating, mind you, meaning it will happily do that for thousands of hours and not just 3 minutes. So I chose to move along (for now) using the ball version, which only has a …. 7,000 pound dynamic load rating, but was thinner and lighter.


The brown object in the middle is a stock round Acme 1″-4 nut that will either be machined as a gear (quick modeled as teeth here) or have a machined ring gear shoved around it with a thermally-enhanced intereference fit (LN2 the nut, bake the gear, shove them together and run away fast)

You might be wondering what the plan for radial loads is, since ostensibly I have two thrust-only bearings designed into the thing so far. The fake answer is that the leadscrew nut, being bronze, will just ride in the stationary bore of the thrust bearings, since the magnitude of thrust loads will be much higher than potential radial loads on a stationary leadscrew.

The real answer is “yolo”.

Here I am playing with actuator positions again. The “pull-to-close” position in this photo mimics that of Überclocker.  I still felt that the important parts were too exposed here.

Another attempt just flips the actuator upside-down and exposes pretty much only the leadscrew. This was at least tolerable in conception – something being mashed into the leadscrew (which could also be shielded) might still leave me enough travel to get a good grab.

Okay, but what else did I learn from Season 2!? That if you leave something important exposed…. say, a master power switch or similar, and run on the assumption that the chances of something getting into there and causing damage are very low, then it will happen to you 100% of the time.

So I gave up the “pull to close” actuator position in favor of just trying to keep the leadscrew short and fat in order to maximize its column rigidity.  The bonus upside is it woud let me keep the existing center hub between the two arms if need be.

This positioning candidate was actually pretty favorable. I could see how the clamp arm geometry might be changed slightly to better accommodate it, and also permit it to use a relatively short leadscrew

Using the geometric constraints put forth by the toy component placement,  I basically wrapped an aluminum chunk around it. The cavities are for the gears and bearings.

I changed the design to an “embedded P80″ to save length. The clamp motor is being moved to brushless, meaning Overhaul will be completely powered by questionable Chinesium. This time, since the Acme screw will not backdrive, I don’t have to hold the stick to apply pressure to the clamp arm any more, making it more Clocker-like in driving. Furthermore, this also affords me the opportunity to overpower the actuator while keeping a high gear ratio for force application. Überclocker’s current actuator is a regular 36:1 geared 550-class drill motor run at over 2x nominal voltage for moar powar – the short duty cycle of a grab and lift haven’t caused motor burnout problems.

A couple of different brushless motors could fit on this gearbox – right now, the SK3-6374 motor is in for modeling purposes.

Adding more parts and thinking about how to interface to the rest of the bot. The large rod-end is a convenient way to join to the wrist pin in the lift hub.

The design is more or less finished here. Those 4 square holes in the side are actually on a 2.25″ bolt circle, so four 3/8″ screws on each side will fasten the actuator to machined trunnion plates. I may end up making 2 of them dowel pins for shear strength and leaving only 2 as threaded holes.

With the new much more compact design, I was able to get a happy result for placing the actuator within the head. This was a good state to reach – I now have a solution where the trunnion bolt holes line up with the circular arc containing the patterned circular cutouts which Overhaul is known for. As a result, I can just hijack one of those holes (appropriately repositioned) as a trunnion axis, much like it is now.

All of this work occurred in the late December to mid January timeframe. I receive the new actuator billets and custom leadscrew nut back from my Chinese contract manufacturer this week.

In the next episode of Overhaul’s Improbable Overhaul Makeover?, I travel to Motorama 2018 with Überclocker in order to practice driving and strategy – and learn some disturbing new information which might disrupt my #season3 ambitions…




All-Vans Quality of Life Patches for Fall and Winter; Going to the Jalopnik Car Show, and Infectious Vantruck Disease

Dec 12, 2017 in mikuvan, vantruck

Winter is literally coming. As time goes on, my ability to work on vans decreases greatly both for lack of daylight and lack of inside matters. In preparation for overwintering now two  vanbabies, I decided to make some repairs and mods that were becoming more necessary (or more necessary, if you get what I mean…). I like keeping my machinery in good functioning state, and Mikuvan was beginning to feel a little like a daily junker more and more. Meanwhile, Vantruck had some lingering bugs I wanted to address before my hands freeze off holding a wrench.


The most important thing was making sure Mikuvan could still pass its incipient Massachusetts state inspection. You see, since the day I got it running in 2013, the exhaust has been slowly shortening itself piece by piece. Recall that one of the earliest bits of mechanical work I did was to rebuild the catalytic converter flange. A year and some later, a part of the exhaust pipe broke apart, which I had a mechanic repair while it was on a lift already for a brake fluid change and rear drums inspection.

That was 2015. About a year to the day, before Dragon Con 2016, it breaks upstream of that repair. I threw together this patch in my classic weeaboo-redneck-engineer fashion

No beer cans here, only top quality RAMUNE BOTTLES. Three layers of them.

Several months later, that broke off, so I trashed the whole section from the bottle-hack back and replaced it with a 90-degree downwards bend with an exhaust tip on it, hanging on to the remaining muffler stub.

Well guess that, that broke the fuck off earlier this year, likely during the Detroit Maker Faire trip. I didn’t notice. I didn’t even care. It can rest in eternal pain and suffering somewhere on the side of the 401 in Ontario.

I just ran with the stumpy pipe out of the muffler which terminated well under the cabin, sounding vaguely like a ricer fart cannon but offering me nothing except exhaust slowly seeping into the cabin when I was at a stop or accelerated hard.

With the inspection date coming up, I had to do something.

Ah, good old New England Organic Loctite. It occurs naturally, regenerating from any exposed worthwhile metal in its vicinity. In the winter, it feeds off the gazillion tons of salt poured on the road and blossoms each spring.

I’d watched the catalytic converter slowly get smaller and smaller over time – even the new gasket I put on it quickly became one with the material. I actually dumped PB Blaster on this flange connection while it was still hot – that was quite exciting. It then took several seconds of impact wrench before I was able to free the converter bolts. Frankly, I was amazed they were removable at all.

Since Mikuvan is emissions-exempt in Massachusetts, I elected to not buy a new catalytic converter and just latch right onto the downpipe stub.


Time to measure up the exhaust path. I needed to clear the rear axle and end up at roughly the correct length to still put a muffler on. I decided to rear-mount the replacement muffler (which was also already rotted out at the bottom, so it wasn’t doing much muffling for a year or two at least) instead of mounting forward of the rear axle like it was before.


And two trips to Pep Boys later, I have all the ingredients! Several lengths of pipe, a flexible coupler, several rubber-mounting hanging straps, and a bunch of tubing adapters. All that is needed to get the right dimensions is an angle grinder!

What, you thought I was going to weld this shit together? Mandrel bends? Mitered joints? Nah. Clamps and impact wrench all the way.

It’s nighttime in this photo because I ran into issues with the downpipe stub – it was some odd metric size of course, and there was no adapter which fit cleanly either inside or outside. I ended up using a 1-7/8″ OD adapter slit and shoved into the downpipe stub, which had a matching slit to let it expand a little. It was then a dance to get the other end of that adapter (2″) adapted to 2.25″ for the remaining pipe. All of the new pipe is 2.25″.

Yeah, the slit is a built-in exhaust leak. Whatever, it’s past the oxygen sensor. Maybe if I feel enterprising I’ll TIG weld it shut (and ONLY it) later.

This section has a flexible coupling in it since the catalytic converter’s output also did, and I wanted to keep the same constraint architecture. The length of solid pipe from here back is hung at both ends while the flexible coupling goes from the adapter salad to it. Should I be required to reinstall a converter in the future, like moving to an emissions-strict area where they don’t just go by OBD-II diagnostics, I should be able to stuff one back in here.

Compared to the… exhausting… dance up front, doing the up-and-over was quite easy and enjoyable.

I decided to be cheeky and go for a SPORT MUFFLER instead of an OEM style one. What, you wanted to sound like a sports car all these year, Mikuvan. Here’s your chance.

This is a Cherry Bomb “Turbo” multi-chambered muffler, distinct from Cherry Bomb’s usual fiberglass packed ones. I don’t have a turbo. I don’t care. It was $28.99 on sale at Pep Boys, and a little of on-the-spot research told me that glass-packs would definitely bring out the ricer fart cannon sound, but would foul up quickly due to the engine consuming oil. Given that,  I was better off with a chambered type.

Anyways, this first attempt made it hang a little too low, so I had to cut the strap and bolt it in closer to the trailer hitch.

The final position. It’s not actually tilted much in real life, by the way – the perspective of this photo is a little strange, since Mikuvan’s rear lower quarter panels curve upwards and the trailer hitch is actually a little tilted upwards also.

So, how do I like the end result?

i regret everything in my life

Okay, the ricer sound was funny for about 24 hours. Between 1000-1500 rpm and 2500-3000 rpm, it seems to resonate the cabin, resulting in a constant mooing sound, a persistent droning. Guess which RPM bands get used the most during gentle city and highway cruising!?

Mikuvan sounds like it has 75 more horsepower than it actually does, which is a 75% improvement. It DOES have more low-end jumpiness, like the second after mashing it from a stoplight. Additionally, the power available past 3500 RPM improved noticeably – previously, trying to throttle past 4,000 didn’t do me much good, and it felt like the engine just hits a wall, but my gas mileage the week after took a complete dive as I was redlining everywhere all the time.

I think this is less due to a sport muffler than just installing the new system as 2.25″ pipe instead of the stock 1-7/8″ (50mm?) pipe for its majority length. I didn’t bother to check if the 0-60 changed. That’s not the point. The answer is still yes.

Realistically, I might toss an OEM style muffler on there after winter passes.  One Dragon Con and Franklin Institute with the Persistent Moo was fairly sufficient, thank you.

Anyways, let’s move onto the more important part of van maintenance: blinkenlights.  I replaced almost all of the small marker and dashboard lights with LEDs back in 2014. A few of them had begun dying, including somewhat important things like the previously chastised oil pressure warning light. That’s maybe a little important.

inexpensive chinese van lighting 3: the reckoning

I decided that enough time had passed to do a scan of the market again, so I hopped on good ol’ Amazon Prime. The market structure™ is very different now – in 2013 and 2014, a lot more of the LED widget vendors were China based. Nowadays, they (or their underlings) all have Prime fulfillment or US-based shipping.



What I noticed is a rise in these purely PCB-based LED units in small (T10, T5, 194, etc.) sizes. I originally bought several styles which were plastic former incandescent lamp shells containing discrete LEDs with formed leads. Those actually didn’t work very well in the end. The LEDs had no heat sinking and tended to burn out or dim quickly, and the formed leads pretending to be T-series shaped were flimsy.

Also, a lot of the LED clusters were unnecessarily bright, containing 5-10 devices. It’s a marker light, bruh, not a camera flash. The ones I found contained 3 or 4 little LED chips only and seemed to have a lot more PCB copper area relative to their size. An example captured from Amazon is bove.

I was curious about one more thing: Most of these products now claim to have “CANBUS error-free” features. After doing a little sleuthing, I discovered that it’s a New Car Problem (a.k.a I don’t care) of the LED bulbs drawing so little current that the ECU/Body control module will throw an error saying you have a bulb broken.

….so here is how the enterprising Chinese widget makers solve it. They drop a big power resistor across the input. To make it draw more current.

This is utter bullshit. Do not EVER buy a “CANBUS Error Free” LED bulb. If your car is new enough to complain, it’s new enough that you shouldn’t be putting questionable aftermarket glowy things on it anyway.  Get an old shitcan like these were meant for. Preferably a van. I like vans.

Here is what the typically 100-to-200 ohm power resistor does: It heats up.

It heats enough to some times desolder itself.

It’s also right next to the LEDs, so they heat up even more and even faster than if they were over-rated and over-driven. I burned one out on 14.0v after like 3 minutes of it just sitting on my desk. It was drawing 0.2 amps until the end – that’s 3 W of power heating up an object which weighs nothing. I think I know why so many of these products have bad Amazon reviews: sadly, people don’t know better.

I desoldered each and every “CANBUS resistor” on each and every one of the 50 white, miku blue, red, and amber LEDs I got. This did not take long, since I had a reflow cannon, but I was peeved to discover that my worst fears regarding inexpensive Chinese van accessories had come true again.

The white T10 units drew 0.05 amps after I was done. That’s more than enough.

The end result is real pretty though.

I changed the master illumination to the “ice blue” LEDs which is really clever marketing speak for my favorite color, Miku Blue.  I also restored all of the small indicators to pure white units so their original colors were back.

That’s enough for silly lighting. It’s still the case that if you want actually reliable LED units, you should still stick with a retail brand name like Sylvania or Philips. They’re going to be pricier, but unless you also have a reflow heat gun and a night to burn and are at least a little obessive like me, just get them.

More recently, I tackled a more reasonable silly old van problem of a broken sensor wire. While doing the fall-to-winter oil change, I noticed a loose wire.

This used to go to the oil pressure sensor (what is with the oil pressure sensor and light as a recurring theme here…) which is located on the bottom of the engine. Heat and oil had stiffened the old PVC-insulated wire until it just broke off inside the connector.

This wasn’t too epic of a fix. I replaced the original wire with a length of silicone-insulated noodly robot wire, up to where it enters the harness and was still quite flexible. This shows the joint and repaired connector before I sleeved it over with heat-shrink tubing and tucked it back into the wiring loom.

Back in place we go!  Excuse the grunge. That is Mikuvan leaking the correct amount of oil my self-applying undercoating system.

I additionally performed some mercy maintenance on the left side. My original body repair on the left rocker panel corners fell off earlier this year. I was kind of expecting this, since I was never able to get the holes in the front (behind the front wheel) fixed and so that repair only trapped water, causing it to fail eventually.

I decided it was better to just leave the lower panel holes open but seal-coat them inside and out. This strategy had been working (and continues to work) for the two holes forward of each wheel, which I coated in Eastwood Goo back in 2014 thoroughly.

So out comes the wirebrushes, in wheel and tooth form. I wire-brushed off all outstanding surface rust first, and reached into the panel holes to manually wire brush off the loose rust inside. Additionally, while I had it up on ramps, I used my slide hammer to try and pull down the damaged lower rocker panel and pinch weld. If you buy a derpy Japanese van, chances are someone’s tried to jack it up by the pinch welds and completely fucked over the metal in the area, I guarantee it.  I only take Mikuvan to mechanics I have talked to and trust for this reason: I don’t trust anyone to know it can only be jacked by the frame. This area came rusty and bent upwards, and had only been deteriorating more. I couldn’t get it completely flat again, but it at least looks better than it was.

Prior to the application of Eastwood Goo, I touched up the paintwork right next to the fuel filler door and immediately in front of the rear wheel. The former had been slowly dissolving due to gasoline fumes and accidental overflows, and was turning the whole area dark and ratty looking as well as causing some of my original bodywork to start chipping off. If I had to point to one thing which crossed my “daily junker” threshold, it was this. I haven’t found a rattlecan product which can completely resist gasoline, so this area will only become ratty again until Mikuvan gets a real paint job.

After the color and clear coat were vaguely dry – as dry as they could get in 40-something degrees, I drew a big fat line with the Eastwood Goo both on the outside here as well as the opposite side, using the extendo-straw to go well into the interstitial space of the panels on both sides.

Essentially I’m just preserving this area from further deterioration. Should I decide that dropping several thousand dollars on a full restoration and repaint is worth it in the future, I will source this body panel either domestically from the southwest/California, or internationally since this generation of Mitsubishi van is still (somehow) in production in various developing countries. Otherwise, an experienced body shop would just strip it all to bare metal anyway. Should I embark on an electrification project, I’ll likely start anew with a donor van in better condition from the same areas (since I assume that if I’m going ahead with cutting up Teslas and Nissan Leafs, that I’m well off enough to have my own garage and lift!)

So that’s Mikuvan’s history for the past 2-3 months. Interspersed with all of this was of course the comparative 800-lb gorilla and relatively white elephant of….


Oh god why do I still own this device. It’s been a year, yet it still feels new and interesting.

As I had sampled a pile of LEDs again, one of the things I did immediately was to retrofit Vantruck too. The incandescent bulbs it came with have long darkened and were sort of miserable looking. The dashboard was so dim it was almost impossible to see even at night.


Well that’s no way to live! Luckily, it uses type 194 bulbs EVERYWHERE. Even the idiot lights. I had to buy another pack of T10/194 type LEDs to satisfy it. (Vantruck is the undisputed king of the phrase “I had to ____ another ____ to satisfy it”)

Naturally, all of the dash illumination went Miku Blue. This was also taken before its 77777th mile party, celebrated by Dane on the road to a Power Racing Series race. Without him realizing it. Hurray, Dane!

By the way, my friends have put more miles on this thing than I have. Since the fuel injection retrofit, it has somehow registered no less than three trips to the New York / New Jersey area and one to southwestern Massachusetts, plus the odd DUDE BRO CAN I BORROW YOUR TRUCK BRO moving trip around town.

I don’t feel bad at all. Buying gas is punishment enough for them.

Along with the interior lights, I also redid the running board lights and forward exterior marker lamps. They were….. you guessed it. 194 type bulbs. I changed the “I am a van” lights by the door handles to Miku Blue since I’m Mr. Vain. It turns out that the bed marker lights are a sealed non-replaceable type, but I can get new ones which are all LED. I haven’t done that yet. I didn’t do the roof lights either – they are fastened from underneath, meaning I’d have to take off the roof liner to access them, which I was not inclined to do.

Notice something else cool? Vantruck now also has LED headlights. They are the same type of unit I got for Mikuvan, except in the H6054 size. They are available in all manners of Chinesium – here’s one example. Just search H6054 LED and don’t buy the 15,000-LED cluster bombss or the fake projector types.

After the LED switchover, I noticed a particularly Vantrucky bug becoming much worse – the lights were flickering hard. LEDs have no thermal mass unlike incandescent filaments. Something was causing all of my lights to flicker, including the dashboard. When this kind of thing happens, there is generally one culprit: a bad ground connection. I dunno whose amazing idea it was to chassis-ground automotive electrical systems, but it’s horrible.

In conducting a test to verify the problem, I connected one end of a voltmeter to the negative battery terminal, and through an alligator clip of sufficient length, to various “grounds” of the electrical system, such as the negative pole of a headlight, the body metal right next to the dashboard where a bunch of grounds for switches and knobs come together, and right next to the battery on the alternator. With the engine running, I captured an incredible 1.2 to 1.5v between battery ground and most things. The worst was, as expected, to the dashboard and interfacing with the body lighting harness in that area. (The correct expectation range I found is usually no more than 50-100 millivolts, and the lower the better just from my electrical engineering intuitions)

Holy crap. Well that explains why the FiTech ECU screen always tended to read my battery voltage as 12.something or 13.something. I verified that from the alternator output to itself I was getting a pretty consistent 14 volts.

The culprit was right behind the alternator – that’s the engine block to battery negative ring lug. I don’t have before photos, but let’s call it “rather pitted and sad looking” and its attachment bolt entirely coated in rust.

My solution was just to epicly wire brush the bolt and the attachment face until they were shiny, and crimp a new terminal onto the 4-gauge cable which was still otherwise in reasonable shape. After retightening, I smeared dielectric grease around the entire setup.

I decided at this point to also give the thing new battery terminals which I had purchased a while back but not installed. I furthermore gave the body a dedicated 10-gauge wire running from the attachment point where (as far as I can tell) the headlight and turn signal harness is grounded.

So I’m not sure if this is an Old Van Problem or is still present in newer vehicles, but it seems strange to me to ground everything to the body and frame yet only give the battery a cable to the engine block. Is the return current supposed to find its way back through to the engine block, jumping through things like bolts and bearings and chains and driveshafts? That just seems extra bad.

I mean, it’s clear there is enough metal contact for it to work for most everyone. Even Mikuvan only has 1 epic ground wire going to the battery from an anchor point on the engine block and nothing else that I can see. Unless I’m missing something, it seems like a dedicated ground wire for the body is really beneficial. It could be that in both cases, there is an actual connection somewhere else on the block to the body, but it’s buried so far in there I have not been able to find it.

Anyways, the moral of this story is wow, I didn’t know all of these lights could be so bright. The ECU display now reads very steady and the correct voltage – 14.4v right after starting and 13.6-13.8v idling when warm. The dashboard is almost comically bright and I had to turn it down with the dimmer for once. Cranking is much faster and less arduous. I should probably go inspect the status of the ground lug on Mikuvan at some point.

By the way, after resolving this issue, I completely reset the FiTech ECU and had it ‘relearn’ the fuel maps by driving around a bunch in mixed regimes. The stable and higher voltage power supply probably helps with a lot of things, so I gave it a chance to re-adapt. Regardless of any other changes behind the scenes, it definitely idles more stably now, so I experimented with leaning out all of the air-fuel ratio targets so it wouldn’t chug gasoline as hard – maybe a few percent less.

Well, over a long distance, that sure matters, because I’m going to a CAR SHOW!!!

the Jalopnik Car Show for Great Justice or Whatever

Delayed once due to being rained out and with the full force of Internet irony behind it, the Jalopnik car show was held the Sunday after Thanksgiving. This would be the first road trip that I myself will get to take in my own vantruck. It would also be the first car show that I actually signed up for. I’ve been to others, including smaller local ones. Everyone has to remember that I am not actually a “car guy”, just a “this one particular silly van” guy.

It was going to be 4 hours on a Sunday in Newark (uhh), which alone is too short of a stay for me to want to drive 55mph the whole way there and back. So I turned the weekend into a general New York City excursion.

With this thing.

If there is some poetry in having a big-block V8-having 9-miles-a-gallon-getting emissions-exempted 21-foot long 65-tons of American Pride occupying a Tesla supercharger spot, I missed it for the funny photo opportunity.

The two Tesla drivers who came in and out while I was hanging around uploading this photo for peoples’ amusement didn’t say anything. Not to me, not out loud. They didn’t dare defy the embodiment of all that is America.

And here I am poking out of a parking spot in Flushing! I’m backed all the way up to the green wall. Actually, it’s pushing the green wall back a good 3 or 4 inches. I felt the contact, and kept shoving a little. Sorry, wall. Sorry, whatever was behind the wall.

So before getting here, I actually drove it straight into lower Manhattan and the Financial District/Battery Park area to try to find…. a location where I could take a photo of it with the Statue of Liberty in the background.


Sadly, that part of Manhattan is too busy and blocked off for any of that to happen. Through friends, I was told that I’d have better luck in Jersey City or parts of Brooklyn. I decided that was out of scope for the day and retreated to Flushing to gorge myself on noodle products by performing a rolling Denial of Service attack on the Brooklyn-Queens Expressway.

Bright and early the next day! They said to show up early. I assumed people were going to start lining up an hour before it starts, so I hustled out of Queens and got into Newark around 10:30. It turns out the organizers had barely even gotten there, so whew.

Well, at least now I have one clean and recent photo of Vantruck before everyone else showed up.


Hey! I brought Chibi-Mikuvan along for the ride, and it was extremely popular. I did some promotion of Power Racing Series, but only when asked.

Originally, I wanted to trailer Mikuvan down, but decided it was simply too much of a production for a 1 day event, and dealing with a now 40-foot long assemblage of vehicles in New York City was a little excessive. If Jalopnik chooses to do a weekend festival of shitboxes or something, I’d happily organize a carrier battle group rollout.

The closer we got to noon, the more interesting things became.


It’s 12:30 now, and we’re starting to have serious traffic problems. Got it – so that’s what “show up early” means!


A Mitsubishi Pajero appears! This thing with a Mikuvan bolted to it is the international 4×4 Delica Star Wagon. They share a powertrain and running gear, whereas the 2WD Delicas (Mikuvan included) share more parts with the 2WD pickup truck.


This Pajero was indeed the turbo-diesel version, and a recent Japanese import. jdm


itp: hipsters

By 1PM, they had to commandeer the neighboring parking lot for all the Chad-come-latelys.

i have money watch me spend it

Okay, okay… that’s harsh. I am sure the owner of this McLaren 720S is a swell fellow. I think what I found endearing about the Jalopnik Car Show overall is that the variety was so not car show, by design.

I’ve been to ones which were Camaro-Mustang-Corvette-Lambo-Stance-Stance-Revolution where every entry was meticulously detailed and shiny and hardly looks like they’re driven. I don’t believe in trailer queens personally – despite keeping my machinery in good running order and generally sound cosmetic shape, they’re not perfect because I use them daily.

In the same vein, I’ll repair and upgrade but never restore Bridgett or Taki-chan, because you create a machine too clean and shiny to be used. Someone else can do that. I don’t want to be hit with regret every time I drop a piece of stock, much like I prefer to be fearless with Mikuvan and occasionally push dumpsters a few feet when the hauling company can’t be buggered to place it back such that it doesn’t block the loading dock.

tl;dr don’t hand me a nice thing

Hey, another truck-like thing! This Ford Bronco is of the first body generation, prior to Vantruck’s year range.  It was one out of only 4 or 5 SUVs/jeep-shaped objects, counting the Mitsubishi Pajero.

And another pickup truck, what a relief to see.

Overall, Vantruck was the only van/conversion van of any type (not counting CMV, of course) and one of only three trucks present, and literally only van-truck of course. Counting CMV, I also had the only van, only cab-over van, and only electric van.

Hell, it was the ONLY electric ANYTHING. If I had one thing to be disappointed about this show, it was the lack of electricity. Surely someone thought about coming with a Model S or a Chevy Bolt or something? Nope.


My other takeaway from this show besides my aforementioned desire to never own a nice thing is that even show cars aren’t all perfect. Again, my only experience with car shows prior to this is ones where everything has an aura of perfection and polishedness along with a nose-in-the-air presentation vibe. So I had a skewed perception of “car people gatherings” as a bunch of perfectionist snobs. I had never wanted to bring Mikuvan to a show since it’s full of my mechanical cockups and bodges.

I think overall going to a show like this was a good confidence booster. Hell, even the Monkees replica car (1st photo, behind Vantruck) had clearly patched and painted over spots where the bodywork had cracked or deteriorated, and a lot of the more nicely finished modified/tuner cars had stuff just hanging off them and random dents and paint chips. However, again, that to me is more honest than a perfect display piece and ‘matching numbers’.

malignant vantruck syndrome

About two weeks ago, I was making my usual patrol rounds using my pre-generated Craigslist searches…

Yes, I have a couple of those in places I often go. Vantrucks show up not that uncommonly – I’d say once or twice a month. But generally they’re either extremely beat up & have sat outside for 20 years, or pretty severely overprived for the condition they are in (e.g. literally over $9000).

This one popped up, though, and it was in a near perfect combo of condition according to the seller whom I talked to on the phone, initial price, and closeness.

So naturally, I had to go and check it out. Portsmouth is but 50 minutes away, or an hour and 15 minutes in Vantruck speeds.



DOUBLE VANTRUCKS! The cause of global warming is right in front of you, ladies and gentlemen.

This one is indeed in very respectable condition. The owner is a retired engineer who has had four of these things throughout recent history. How do I know this? He had a dedicated photo album, each photo laminated and in a pocket, of all of the repairs and modifications he’s done to all of them. I want to say “wow, this guy is like me but with real life pictures” but the magnitude of things is so different I can’t begin to use myself as the reference point. It was, though, very inspirational to see how excited he was about all the ones he’s owned and the customization work he’s done.

Anyways, the best crusty old vehicles are usually owned by dedicated retired owners. This one had a slew of mechanical work and replaced components in the past 25,000 miles that I won’t bother listing here. I did some sleuthing underneath to determine the state of the frame and other bodywork. The interior is immaculate and all of the coachwork is original.

So you might think that I went ahead and expanded my aircraft carrier fleet. Well, kind of….

The trip was actually a scouting mission for a robot buddy, Alex of Wedge Industries, a long time northeastern robot competitor. In fact, me versus him was the Franklin Institute finals in the 30lb class. So now this thing is in the robot family… and Motorama 2018 is going to be certified dank.

I was going to Double Vantruck to go meet him for pickup, but the heavy snowfall on that day caused me to rethink that plan and I instead headed out with someone else who had 4×4 and a not 70/30 weight distribution. Here, Alex stops over in the shop after getting a trailer to tow his own car back with. Have fun with your 9 miles a gallon all the back to Pennsylvania :p

I now leave you with this.


The Return of a Legend: ChibiKart Reunion Tour feat. Brushless Rage

Jun 20, 2017 in D.P.R. Chibikart, Motor Controllers

Brushless Rage is moving along quickly! I’m really hoping now to do a limited release (to people with known loads and needs) in time for Detroit Maker Faire. I’ve been working on it more sporadically in the past month due to other… obligations? but now I see the tunnel’s end. Here’s what’s been going on with it in the past few weeks when I haven’t been hiding under a van.

So that 2-way optocoupler salad was good in concept, but it ended up being incompatible with its purpose in life: to communicate bidirectionally so I could use the servo cable as a programming cable for SimonK/BLHeli enabled bootloaders.

It seems that the protocol requires the ability to tri-state, or at least assert both high and low logic levels. The circuit I modified can only drive high (or low) and otherwise has to rely on a pullup resistor, and that might not be playing nicely with the needs of the protocol. That is something I haven’t studied in depth due to its poor documentation, so if you know the specifications for the protocol, chime in!

Either way, it was looking like the final board revision would just use a single unidirectional optocoupler for the R/C signal input, and another galvanic-coupled pin on the same line but on the microcontroller’s side of the optocoupler as a programming header.

When the optos were bypassed (….again…. sigh) I was able to use the AfroESC programming dongle to re-upload firmware and change settings at will. The first step in this process is to flash the ATMega microcontroller with a socket and use the Enable Bootloader setting in KKMulticopter. Then I can just use the USB dongle instead of breaking out the socket every time.

I prepared two units this way, and also had heat sink plates cut. These heat sinks were designed a while ago for the Half-Rage that doesn’t exist yet – it has exactly half of the spacing of the mounting holes of RageBridge! So it was a good pick for the 6-FET power board for Brushless Rage. I cut out a square of silicone pad to fit underneath. In the ‘production’ version they’d obviously be die-cut to shape.

So now I have two mini Brushless Rages. What would I ever test them on!?


It’s back! I reclaimed the D.P.R Chibikart from the MIT shop not long ago, since they were refreshing a lot of the space displays and I’ve been gone a full year and a half now (…). This thing was kind of the pinnacle of my design class years, it having won an Instructables contest and all, and serving as a foundation for not only my next few years of students but for about a dozen or so builds worldwide (possibly more – those are just ones who wrote home).  A lot of tricks and hacks were used on it to make things easy to build for people without machine shop access. It’s also just stupidly fun to drive, and before the MIT IDC became populated extensively, we had stupid indoor go-kart races in it.

Over the intervening 2-ish years after my EV building class finished its run and now, it had been on display in various forms, so it wasn’t operational. The batteries had been removed and the motors’ sensor boards (which were partially designed for vehicle projects like this!) were broken off.  So I was just going to rewire it from scratch to accept two Brushless mini-Rages!

I focused on mechanical restoration first, like retightening some bolts. I had to add a new chain on the right side since the old one fell off (with the sprocket) a long time ago.

The sprocket itself is also quite well used, and the set screws are no longer very tightenable without stripping. I’ll likely have to drill these out to rethread them later due to the much higher potential torque going through them now with Brushless Rage.

Battery-wise, I decided to look for a 36V solution to make sure they can operate at 10S/36V reliably. I had some older 10Ah e-bike packs which were given to me with broken BMS cards. So I just removed them and soldered output wires in place. Classic!

The output wires terminate in XT-90 connectors, which were also retrofit to the existing wiring harness.

The Brushless Rage units are mounted with not much more than some Dual-Lock patches, and.

I had to pick through two boxes of random electronics to find my last working servo tester unit. In a pinch, these can be chopped up to accept Hall Effect throttles in place of their potentiometers. The Hall throttles typically put out between 1 and 4 volts instead of a full 0 to 5 volts, so the motor controller would need a calibration ability to get the full range out of it.

As expected, the Hall throttle’s 1 to 4 (well, about 0.9 to about 4.2) volt swing puts out somewhere around 1.13 to 1.85 millisecond servo pulse lengths. I set the Brushless Rages to accept 1.2 to 1.8ms as a result.

Everything’s bundled back up now!

Riding this thing has now become very interesting. Due to its very low gearing to the ground (only 3:1), it does have a hard start, but will always do so after a cog or two. This was actually a good test of how tuned out the SimonK firmware is; the mass-to-force ratio of an EV is usually much higher than that of a robot, even the 240b Sadbot, so it’s a tougher load to get going. The power is not unlike what BurnoutChibi ended up having, but more muted; BurnoutChibi had the advantage of being able to spin the motors much faster to get some ‘free power’.

I immediately ran into the problem of blowing the set screws right off the small filed flats on the motor shafts. This thing was originally designed for maybe 500-750W of power using the e-bike controllers, not an unlimited-current dump.

Either way, some replacement set screws and Loctite enabled some “road testing”. Here’s a highlight:

Results: My Starting-and-reversing optimized SimonK is okay in an EV application but only under some circumstances.

Specifically, you need to either turn down all the braking ramp speeds and magnitudes, or remove motor braking completely. In a robot drive application, the motor braking very closely following the command input helps decelerate the load and therefore reduce the momentum the motor has to start against the other way. In an EV application, that just means you decelerate as hard as you accelerate. It COULD be okay for some things, of course. I found that Chibikart drove well if I had the BRAKE_POWER setting cranked down to 1/8th of MAX_POWER, as well as the BRAKE_SPEED (ramp-down rate of the output PWM, basically) reduced to 3.

With these settings, I could modulate the throttle pedal to give a predictable regenertive braking effect. Too fast BRAKE_SPEED or too high BRAKE_POWER and you just end up impaling yourself on the handlebar here. I could see this on a tight Power Racing Series just thundering around never touching the brake pedal/handle, but it would still be a little annoying for a scooter or electric [skate,long,mountain...]board where you’d rather coast. In that circumstance, I’d just turn MOTOR_BRAKE off and forget about regeneration anyway.

For comparision, I found that Sadbot drove the best with BRAKE_POWER = MAX_POWER and BRAKE_SPEED at 4 (BRAKE_SPEED maxed out at 8 actually tried to slow the motor so fast it tended to either lock up wheels or slip motor poles on deceleration).


And with that, I sat down and pounded out board rev 5:

The main difference is removing the bidirectional optocoupler, as discussd, for a normal one. That’s still a 2-channel opto; I have yet to find a single channel (4-pin) opto in a package I like, but it does make more sense to use one here. Besides that, in rerouting some of the optocoupler traces, I got suckered into giving it better analog and digital signal separation (oh, boo-hoo…).

I also finally implemented the damned LEDs. SimonK actually has LED support, for signals that indicate throttle state and motor state. About time I figure out what this thing is doing!

Overall, I think Brushless Rage is ready to be fitted on something for Detroit Maker Faire. I’m not sure right now if I’m racing anything, or going to marshal and tech-safety-Stalin. I may choose to temporaily rebody Chibi-Mikuvan for funsies, since I want to keep the CMV shell in good shape after retirement.

Well, those are just thoughts anyway. There are also other thoughts: