The Long, Cold Winter of No Posting Ends: Awaken the sadbot2019

Is this thing still running?

Needless to say I am a little less than proud of the longest post drought this site has ever seen. I’m still alive! Just preoccupied largely with getting the company product technology to a stage where I can at least be assured the tunnel will end eventually, even if I haven’t seen the light yet. Along with this gradual better-scoping of product tasks, it’s taken me a while to get used to not just working on dumb projects all the time in big blocks of work, but learning how to divide up tasks and think about their dependencies more, such that I can pick stuff up and drop it back off easily. I brought this up a bit in the preamble of the Great Mikuvan Engine Rebuild Scandal of yesteryear. Whatever, it just means hopefully I can get back to working on stuff, but until the day I exit the company and become a full time bad idea investor, the pace will inevitably be slower.

I believe the real world calls this “Adulting”.

Anyhow, adulting is dumb and robots are cool. Let’s revisit Sadbot, which was left kind of functioning last year some time before I sold the pokey dingle to a west coast team. Then, incrementally more parts started coming out of it until there wasn’t much left but Overhaul 1’s drive system and a steel box. With the coming of BattleBots season 4, I figured I had to at least work on something, and I should probably consider repairing my wreckage instead of creating new wreckage.

Oh, if you notice the timing of this post – I’m clearly not competing in Season 4 with Overhaul.


I mean, if I didn’t even clean the barbeque out of it yet….

Ultimately I didn’t think I could muster up the time and resources to do the bot justice, and enough of the team has split off to get real jobs (among other things, I mean) that i would have had to rally up a new crew. So, perhaps next year, and maybe it’s a blessing in disguise because now I’ll work on Sadbot out of FOMO and remorse, perhaps discovering something new about Overhaul in the mean time!

Next season, though. I promise I’ll be Bach.

So I set out to change up Sadbot in a fashion that would reflect the mods I want to make to Overhaul for next season. That in detail is itself an entire blog post for when Overhaul is modified for #Season5 one day, but in short…

  • Change the drivetrain to the 80mm “melon” motors – Sadbot being a single motor per side, it will 100% reflect the drivetrain setup I want Overhaul to have in the future, as in my post-season assessment the dual motor setup has not been as reliable as I wanted.
  • Using this opportunity to make sure the 12FET Brushless Rages weren’t actually trash, but were not utilized right the first time out. I had some more testing and changes I wanted to make after Season 3 that I hadn’t gotten around t
  • Finally doing the tractor pull contest between Sadbot and Overhaul which never happened. While the bot had “more” traction than Season 2, it wasn’t that much more, just more linear and predictable. I want more, something which I suspect is beyond the capabilities of my current bot architecture.

The first step is putting the damn thing together again. To do that, as usual, it has to come apart more first!

To retrofit the 80mm outrunners, I had to re-introduce Overhaul 1’s intermediate drive gear. The previous motors in Sadbot were 59mm SK3 outrunners running into Banebots P80 gearboxes, so the motors were already geared down and only needed the center sprocket.

To get the ratio I needed using a gearbox would have made the assembly too long to fit into the frame, so I needed to directly attach the motors to the face, needing the extra ~3:1 the intermediate gear provides.

I machined these gear-sprocket combos late last year. They’re waterjet-cut 12DP gears that are pinned into the sprocket face such that the assembly rotates on a dead shaft (pictured mounted in its former home). These were virtually identical to Overhaul 1’s (which were long disassembled or I’d have used them again!) but a different ratio.

New socket cap screw holes sunk into the drive plates – now featuring THREE bolt patterns! P80, 3″ Ampflow/Magmotor, and 80mm C-series outrunner.

Modules taken apart, cleaned, regreased/re-threadlocked, and reassembled.

At this point, the ‘skateboard’ of the bot weighed 163 pounds without batteries or the controller housing. This began my contemplation for bringing back the pokey dingle. I’d sold it because I didn’t like the design any more and someone else was going to incorporate it into another bot, so what better way to force me to start over?

Originally, Sadbot was going to be 220lb (nationwide Heavyweight class) without the pokey dingle, functioning only as a pusher/brick bot, and 250 pounds with it in order to simulate a BattleBots practice opponent. I was, after this weigh-in, now convinced I could make it close to 220lb even with a weapon, which would let it compete in the rising amount of “Heavyweight Sportsman’s” events around the country. Or at least I’d get it close enough that the laid-back nature of these events would make them take it anyway!

Next up, assembling the control deck. These were parts also designed and cut out last year – a basic polycarbonate tabbed box and nutstrips to make a second floor. Batteries will be mounted on the first level, and the aluminum plate mezzanine will house the brushless Ragebridges.

Originally, I wanted to use the Overhaul 1 batteries after being decomposed from the modules. However, they aged poorly and the cells have high internal resistance now, so I put together some of the Overhaul Season 3 batteries (Hobbyking Graphene bricks) instead. I greatly enjoyed how these worked in Overhaul, and while it packed four, I think two would have been plenty. Well, it’s time to test that!

The batteries are secured on the bottom with a few acres of Dual-Lock – think gender neutral Velcro for the woke Millennial – and then pressure-retained downwards by the 1/8″ aluminum plate with the adhesive-back foam rubber pad.

I forgot how easy wiring a 3-motor robot was. That’s it! This is the whole wiring harness, minus the battery-side mating connectors.

The rest of the wiring was built up over basically one evening. I had two leftover Whyachi Switches, one which I’ll set aside for the New Pokey Dingle weapon and the other for the drive. This keeps the activation process similar to what’s expected for BattleBots and also just allows me to test one system or the other.

And that’s all, really! This build was quite short and pleasant, occurring over several nights in about a week and some. Sadbot at this point weighed just under 180 pounds, which only left me around 40 for the entirety of the New Pokey Dingle. Difficult, and I decided at this point that the 220lb max goal was probably not that important, but I’ll give it a try.

 

I knew the parts I wanted to use already – an Overhaul lift gearbox (Banebots BB220 16:1 and SK3 59mm 149kv motor), enough reduction to get to about 180:1 which is the same ratio as Overhaul, and that the end effector should still be the Harbor Freight “manual slide log splitter” / toe destroyer. Furthermore, last time I permanently welded the tool to the output shaft, but this time I intended to make the output a socket to potentially make interchangeable ends.

I started flowing plates around some initial component placements. The output sprocket was only going to be able to get so big, so I fixed that first (48 tooth) and gave it a position that had some clearance to the electronics box, some clearance to the ground, but high enough to allow the motor to tuck underneath with some semblance of an intermediate stage. The width was fixed by choice at “Between the Melons” – one of the things I didn’t like about the last Pokey Dingle was how wide it was for what it did, and I had some ideas which involved moving the sprockets around to make the while thing narrower.

Here’s roughly what that looks like. The chain stages are very short and all overlap. Technically, I could have made this just open gearing, but chain drive is more available and serviceable – all of these sprockets, save for the output, are off the shelf parts.

I was contemplating how to make interchangeable “manipulator” sockets compact enough to fit in the confines of the side plates. I played with a couple of ideas including welded machined parts, making it a live shaft again (but with a socket tube welded to the shaft, and so on. All of these ideas turned out to be either too wide or, after a moment of …. brilliance? too tacky and complex.

Why not just weld the damn square tube to the plate sprocket, using the bushing as a locating feature!? The  wall thickness of the tube certainly permitted any inserted attachments to not come in contact with the drive chain, so that was really all.

I also utilized a chain tensioning approach which I remembered, but couldn’t place where I learned it from. Typically with a slotted mounting system you’d place the slots parallel to the direction of tension needed, e.g. slide the motor away from the shaft perpendicularly if you need to adjust the chain/belt spacing.

However, this arrangement doesn’t resist the normal forces that chain tension plus torque puts on the sprockets, which tends to force them together and loosen things up. At least, not all that well unless you had massive fasteners. Instead, I angled the bearing mounting slots at 30 degrees from the perpendicular to the axis joining the sprocket centers. This means I move the sprocket more “up and down” relative to the other one, but there’s that 0.5x component introduced by the 30 degree inclination which adjust the actual tension, and also reduces the effect of the tension “attractive force” immensely.

The downside? I have to have two chain pitches (well, sqrt(3) / 2, so basically 2) of vertical free movement in the chain to gain that horizontal spacing. That’s not all that much when considered, so only a little geometric squishing was needed to get space for everything.

Initial positioning in the bot  yielded some more Geometric Squishing to get the parts to all clear. I went out of my way to make things actually parametric and geometrically related instead of hard-coding dimensions, enabling some click-and-drag placement.

The final assembly by itself modeled with shaft and bearings for a realistic weight. It looks like this thing will weigh a little over 40 pounds after all. I added some cutouts to remove material where it wasn’t really needed, but the final bot will still end up around 225lb.

About a week later, this showed up! I sent these out to a local shop which the company has built up some rapport with, and they laser cut it from 1/4″ regular-ass P&O steel. Nothing too high tech going on here, though I recommended they stock up on ARx00 steel for future robot seasons…

Because I wasn’t in control of the machine any more, I very liberally oversized all the slots and shrunk the tabs – I went 0.015″ oversize in X and Y for planar square slots, principally. This turned out to be a near perfect, slightly jiggly fit. Laser cutting generally has a less clean finish than waterjetting, plus I couldn’t order them to “move the laser inwards 5 thou” like I typically fudge nozzle offset distances when waterjetting personally.

 

The machined parts needed for this new Pokey Dingle was really just the output dead shaft/epic standoff. I used some 1″ precision ground shaft leftovers and end-tapped both sides.

It was then used as a welding fixture. The P&O (pickled and oiled hot roll) steel was very clean from the get-go, not needing the intense sanding/brushing typical of A36/A514 hot roll or tubing products.  So it was literally just 10 minutes of MIG blasting here.

I modified the Harbor Freight Robot Tallywacker this time by cutting off the heavy punch weight at the back – it’s just a 2″ steel billet chunk. Previously I drilled a 1.25″ hole into it and welded it to the output shaft. This time, the 1.5″ main body section is more interesting, since it will be cross-drilled and bolted to the tube socket.

 

I then proceeded to get carried away.

It started out innocently enough wanting to paint the frame of the New Pokey Dingle my signature Overhaul Miku Blue – and then I discovered I had a Miku Magenta can I never used!

Yeah, well, this is what you’re getting now. aestheticbot9001

I was also tempted to paint the frame itself a light pastelly purple, but by this point didn’t want to disassemble the thing again.

Marking and drilling the front mounting holes for the NPD which brace the thing against the massive C channel section of the front of the bot. Another reason I didn’t like the old NPD – it only was bolted to the bottom of the the bot, meaning a hard enough hit and it will probably just bend. The New Pokey Dingle acts as a truss structure to past the center of the bottom plate – hopefully this will yield substantially more rigidity.

Drilling the holes themselves was an adventure. It was too heavy to put on a drill press, and too tall for the Bridgeport. So hand marking and drilling we go!

I’m rather fond of step drills. For hand drilling in steel, twist drills are almost inevitably too aggressive and tend to either spin in the drill or catch and throw you halfway across the room. Step drills feed more controllably and never dig and then slam the drill into your kidneys.

I machined a small donut piece to bridge the gap between the shaft bushing and the plate sprocket’s 2″ bore.

The same annular cutter I used for the original Pokey Dingle (as well as Overhaul’s gear holes and a lot of other parts) was used to put in the 1.25″ bushing hole. Really that was the only operation needed here! Just a single, albeit massive, cross hole.

Then you clean the parts and MIG-smash them together!

Here’s how it is going together. Since I don’t really care about the precise alignment of the socket tube, all the bushing has to do is center it. I put a few tack welds around that end first and more solidly welded the other side (carrying most of the load) – welding too much around these bushings would deform them due to the lower melting point of pressed-together bronze particles (not to mention sweat oil everywhere!)

Installation was simple, with just a few shim washers needed to space everything out.

Well, I ended up taking the whole bot apart again anyway, so maybe I should have painted the frame purple. But here’s how the New Pokey Dingle elevator machinery bolts in – a line of nuts on the bottom, and the four big bolts on the front.

I broke into the electronics enclosure again to add the 3rd Brushless Rage. I ran out of production-spec boards at this point, though, having packed the majority of them already for product shipment and BattleBots, so I pulled out one of the previous version power boards (The signal board is the same production-spec one though, just an older power end revision).

The 1″ UHMW top lid had to be modified a little to clear the new sprocket placement, which was a simple jigsaw job with drilled holes at the vertices to turn the saw around.

And here it is put together! The final weight? 226.5 pounds (with the top plate, which isn’t on here). I could probably knock 6 pounds out of it somewhere, but that’s not really the point.

So that’s where Sadbot sits now! It drives great withe dual 80mm “melon” drive, but I haven’t fought it against Overhaul yet or otherwise substantially battle-loaded it. In between the first photo and now, we actually moved shops again, and Sadbot is still hiding behind a few pallets. So, watch for both a post about the #NewVapeShop(tm) as well as some more testing and driving videos.

The Overhaul Design and Build Series, Part 5: “Don’t you have to ship this on Wednesday, dude?”

Yes, yes I do. STOP REMINDING ME OR YOU’RE HELPING MACHINE THINGS.

Well here we are, after the airing of the Overhaul vs. Sawblaze fight which will be on Science Channel’s website and other streaming service soon! I have a full writeup I need to do on the lead-in and post-match analysis for that one! That will come after the conclusion of the build series in this post.

We rejoin our heroe…. dumbasses in the 2nd-into-3rd week of March. March 21st was the latest ship date available for east coast teams in order for everything to make it there on time (or so we were told!?). Luckily, Overhaul was actually not in a bad position, at least compared to Season 2 when the extra long days really started kicking in. All I really had to do at this point was a final assembly, then work on remaining spare assemblies.

After the Week of #WeldingGoneWild, it was actually very easy to do a fitting of the whole front of the bot.

That’s about it. The only thing which wasn’t added in this photo was the clamp actuator itself.

I’m much more a fan of this design already. Once the whole thing is loosely assembled, there is a degree of “elastic averaging” *ahem* that goes on as all the bolts get tightened down, but after that, the arms are rock solid.

The drivetrain is being assembled more now. Check out the Markforged nylon engine timing chain style guides! The front chain was still a bit loose after this so (at the event) I ended up making a different set, to be shown.

The one on the right between the two motors is a little ridiculous. We were running so tight on time that I wasn’t going to get the #35 half-links of chain in on time, or at least too close to risk not being able to drive test. So I invented the stupidest possible chain -pincher for the intermediate drive chain – it was gonna wear out very fast with its profile, but would at least let me get some test driving in.

Closing up the other drive side. The design for serviceability that I did 2 years ago is really coming back to help me here. Remember, my team this year is scattered – Paige is working a real job across the country, Cynthia is occupied full-time and could only help on a limited basis with set-up operations, and I only had Allen’s help briefly with welding too. Most of the photos taken in this build series was work done by myself solely.

Here’s the first test-fitment of the entire bot with all hardware installed. I’m really liking all the design changes to the steel parts. In person, the new clamp and forks look better proportioned to the bot. At least to me, way better than OH2 for 2016.

(Fun game: See how many dumb project artifacts you can spot in the background of this and other photos. Chibi-Mikuvan currently resides under my desk.)

 

I spent an evening just pounding out spare parts for the incipient shipment. For one, I was short on drive motors now, but with a shipment of new HobbyKing Sk3 6374-192s waiting, I needed to key the shafts and secure the hardware. It was easier to pop the shafts out en masse and set up the mill properly.

This and more! I went through….. zero 2mm endmills, somehow. Still a harrowing operation.

It was now the weekend before, and I realized that I wasn’t going to be able to get my last round of waterjet-cut parts in time. These days, I get to be in the back of the line for MIT shop waterjetting – which I think is a very reasonable voluntary position to be in, as I have no official involvement any more with the institute. But dangit Sawblaze, you guys still do!

The electrical deck could conceivably just be drilled from a plate of aluminum, so that is what I ended up having to do. Out comes the TERRORISM. I just cut a chunk of 1/4″ aluminum plate out and started marking holes like high school Charles would have done, and he is always right.

Please do not ever, ever, ever do this. This is how you die. This photo is for illustrative purposes only and should never be attempted, building a robot is dangerous, etc. If you do, use the finest tooth blade you can get and have someone else pump WD-40 or cutting fluid constantly. Or you will die.

 

In the middle of the process. All the small holes are #4-40 tapped for Brushless Rages. Notice how I put six holes in some positions? This will be important later.

 

I’m loading up the bot with electrical deck hardware and wiring now. The shock mounts are in (and secured from the bottom) and some of the battery harness is visible.

The wiring for Overhaul this time was…… ad hoc, to say the least. I said I would dispense with the carefully cultivated greenhouse of busbars. The fanout occurs at the master switch terminals this time, with 8 gauge main leads splitting into multiple 12-gauge intermediate cables to the Brushless Rages. Single 6-AWG conductors handle the “fan-in” from the 4 batteries to the master switches.

The octopus taking shape, with ESCs installed.

Remember those 6-hole patterns? I had to temporarily use the 6-FET models for drive, keeping the 12-FETs for lifting and clamping.

What, are they magic or something? In actuality far underrated from their published specs?!

No, but I accidentally sold my entire product line – which is great – leaving ME with no remaining 12-FET units – which isn’t great. I had to dig into all my pre-production units here to even complete Overhaul at all. What is stock tracking even??

I wasn’t running off a cliff without a drone backpack, by the way. A month earlier, I had placed the assembly order for a new batch of Brushless Rages, but they wouldn’t get in until the Monday of the ship week and I was not taking any chances.

With the octopus wired in, the drive base is live for the first time. Check out the motor-on-a-stick I used to simulate having a clamp motor.

 

Overhaul prior to its first indoor test drive! At this point, I’d fight in 20 minutes if I had to. I think I was actually ahead of the curve here by a distressing amount.

 

I cleaned up and routed more wires into loom in order to un-nest the wiring some. This is inconceivably ugly to me, like a steaming fresh pile of partly-regurgitated dog squeeze smeared across a sidewalk by multiple unassuming passerby. I would never ship this in a consulting project. Yet some of y’all at the event said this was the cleanest wiring in a bot ever? What the actual hell is wrong with you?

(After seeing inside everyone’s bots, I’m not inclined to disagree. Sorry not sorry.)

 

Monday came, and hey! Look what’s here! More Brushless RageBridge units for all! I’d end up assembling 25 12-FET units and packing them with me to the event (not including my own spares, which were 4x for drive plus another 4x for spare overhead)

One of the put-off things was welding the wedges together beyond just tacking, so I spent much of Tuesday doing this. The plan was to take the tacked units to MIT to use one of the MIG welders I have access to and really smash them out quickly.  Using the bot itself as a welding jig made for expedient alignment of all my wooden dinosaur puzzle pieces.

DETHPLOW ™ was tacked together in the same fashion, by actually mounting all the pieces to the bot and locally squeezing with clamps.

For completion purposes, I fully TIG welded one set of pointy-wedges and mounted them on the bot in their final positions. TIG welding is truly the wrong technique to use for bitey pointy robot parts, in my opinion, since it takes so much time compared to MIG in an application where the sensitivity is not really reflected in the end product.

As I mentioned previously, it ended up pissing us off so much we immediately bought a MIG welder after we all got back from the event.

This, for instance, is DETHPLOW all MIG-welded together, a process which took only 15 minutes or so once it was jigged up. I designed all these pieces to be MIG-filled anyway. Here I am doing some TIG touchup on areas which I fell a little short with the wirefeed or missed, or had a gap that I couldn’t bridge as the fitup wasn’t 100% perfect. This is a fine state of affairs for me – blitzing and then fine tuning if needed.

One thing which occurred over the weekend was crate setup. I decided to just spent money this time to get an elegant and reusable SINGLE. PALLET. solution. As someone who’s had two double-pallet crates wrecked over two BattleBots seasons for reasons unknown, I decided I was much better off with a tall single pallet. U-Line came to the rescue with this 4′ x 4′ x 6′ tall snap-together crate,which I modified by adding some removable side-in shelving levels. The bot with its lift table and large tool chests/boxes was to fill the bottom floor, and more containerized accessory suitcases in the middle, along with the pictured Markforged gear – Markforged went ahead and lent me a 2nd printer for the event.

The top level would contain the loose large parts such as the frame rails and spare welded assemblies.

You know all those spare-everythings I was cutting and machining? They ended up in a tote which contained all the important mechanical bits of the bot. I’d prepped a full set of drive and lift motor spares, along with a few mor prepared motors. There was also enough cut tube sections to weld up a new clamp at the event if it came to be.

 

And here it is, Overhaul and all of its support equipment and tools plus spare parts, all ready to load up into….

 

Vantruck??????????????

 

THE PLOT THICKENS! During the week prior to shipment, a few of us NE builders came together to ally ourselves against the forces of time.

You see, Team Forge & Farm was planning to road trip across the country with their bot in tow. For a nominal fee of a few spare RageBridges, they were willing to also bring Overhaul along.

This effectively bought me an extra 4 days to work on the bot – in fact, the electrical work and spare welding photos you see were done after the 21st. On the Monday following shipdate, I picked up HUGE along the way and ended up in southern New Jersey.

 

…where, under the cover of darkness, we packed Earl’s truck up with our bots and his alike.

So… what’s in the crate? Well, there was still the lift cart and all of the already-completed spares, the printer, the mechanical tote, and other support equipment like the battery chargers and power supplies, and a few doen Ragebridges of various flavors. I handed off the radio suitcase and both of my event toolboxes off with the robot.

 

There were still a few kibbles I had to take care of after the bot went out but before I did. So, how do you jig up pieces for welding with you don’t have a robot???

You 3D print an imitation of the robot! This is an Onyx print with the same hole spacings and offsets as an Overhaul front frame rail. I used it to tack the pieces together quickly (as to not melt anything) before removing them.

And with that, the build of Overhaul 2 for the new BattleBots concludes. To be entirely honest, I found this build season pretty stress-free, largely because I didn’t have to build a new bot from scratch and was making only well-scoped changes I had anticipated in advance. In the position I am in now working on a new company with my friends, I don’t think I could have pulled off the record build of OH2 for Season 2 in 2016. My (and my friends’) experiences in this build and competition season of BB – without going into NDA details of the show – has really shown me that I have to move back to a “When it’s ready” format like I had to do during my busiest times at MIT trying to vaguely graduate on time. I have a lot of thoughts on the show as a whole and the direction I’d like to move in (and the show should/shouldn’t move in) that are much better reserved until after the entire show airs.

But for now, hang out here a bit for the event report and a SawBlaze vs. Overhaul post-match!

And now a word from our sponsors!

HobbyKing – Somehow still loves me and enthusiastically supportive of my efforts to abuse R/C model parts for unintended applications! I’m running a HobbyKing radio (9XR Pro) and batteries (Graphene 6000 65C 6S packs, times 4), motors (SK3 63-series), BECs, and a whole lot of wiring and connectors. Not to mention the Reaktor battery chargers and who knows what other HK kibbles have made it into my tools and accessories. I like to think that I had a large role to play in the commoditization of silly electric vehicles using R/C parts also.

MarkForged – from the days when I knew 50% of the company to today when half the new marketing and sales staff go “Who is…. Professor Charles?”, they’ve provided me with high-strength printed parts for a lot of different projects, both on this site and off. Introducing them to the robot fighting community via Jamison’s and my efforts pretty much made MarkForged printed-unibodies the competitive standard in the 1 and 3lb classes, and trying to find new niches in the bigger weight classes is one of my goals. This time, Overhaul’s drive wheel hubs and casting molds are printed from Onyx, and there are also plenty of smaller chain glides and tensioners and accessory parts.

SSAB – I find it interesting that the company’s full name is SSAB AB – Svenskt Stål AB AB, or Swedish Steel Company Company, but these days the lettering is the whole company name so that’s actually not true. This year, I’m working with one of their North America regional distributors and all of the armor steel on Overhaul – including the entire clamp arm, top plates, and new wedgelets and DETHPLOW™ are Hardox 450. Hardox is the easily-obtained ARx00 of Europe and other regions worldwide, and bots overseas have used it for years, but it’s not really had a foothold here in our scene compared to the number of AR-spec steel products in the US. So hopefully I can help with advancing that brand too!

BaneBots – I was called an edgelord for even thinking about using P80 gearboxes in a modern Battlebot. I always thought they were under-loved after the FIRST Robotics Competition quality issues of the late noughties, and had used them otherwise in several projects including consulting projects before shoving shorty Ampflow motors into them for Overhaul 1 in 2015. And you know what!? They’re great! Overhaul 2 ran them exclusively for Season 2, and now for Season 3, OH is sporting the new BB220 series with much stronger planetary output stages for the lifter.

Equals Zero Designs – Yeah, umm, I don’t know much about those assclowns.