Category: your waifu is trash

Your Waifu is Back…. and Still Trash! Rebooting the Dumpster Miku: The Designing

chorl
what a title. we live in such a society

Poor Dumpsterbot. After the initial build in 2019, it might be the highest mileage robot I’ve ever built – literally, not figuratively. It’s driven through probably five conventions at this point, to great fanfare…but the shitty Miku Blue spraypaint I put over the shitty gold spraypaint NHRL put over the original green paint just flaked off in large pieces over time.

It was already showing its cracks in Houston in 2022. I decided to retire it for Momocon 2023 because it was just too decrepit to be funny:

There was like… no more miku left because all the stickers gradually fell off. And so, YWIT just sort of lived on that workbench being shuffled around until I tossed it in *somewhere* to move to the New Robot Trap House. It then lived on a hangar shelf for the better part of the year as I attended to the everything else that needed me screaming and running in a circle.

But with the NRTH having reached some kind of ignition/criticality state earlier this year as I managed to birth Stance Stance Revolution, I decided to revive it for Momocon 2024. Well that came and went…

Some time in early June, I pulled it out and decided to see if I could get a local auto body place to strip the paint off and start over. What I learned during this process is that America has pretty much decided it’s done making or building anything, for a variety of reasons.

I mentally had a “Probably around $250” estimate for stripping or at minimum roughing up the original coat, then blasting it in obnoxious candy-Miku-Blue. I had a bunch of places, even more Bubba-n’-Sons looking corner auto shops, decline this (admittedly rather small) work and give me farfetched tales of costing hundreds of dollars to boot up their sandblaster. It might just be the area and I didn’t go exploring far enough out or deep enough into the hood, but all of these shops seem to be hooked on big-dollar insurance claim jobs.

I get it, labor and input costs are higher than ever, making fewer and fewer products worthwhile to repair in our society. Just throw it away and make it China’s problem. That’ll never come back to bite us in the ass or anything, right? But then I got a reference to a nice hole-in-the-warehouse shop whose owner would allegedly paint anything.

This was the state I delivered YWIT’s emptied shell to him, and just two days later, for only $150, I received it back like so!

Fuck it, I just left the whole $200 there as a tip for not giving me a runaround fairy tale about the role of the President in deciding the inflation rate. Absolutely galactic, brilliant glossy single-stage Miku Blue!

So much of my value add to both my own projects as well as organizations I am a part of is simply knowing where to get things made/done. This is a critical part of keeping costs and lead times (and frustration) down. I had a bunch of favorite service providers back up north, but have been slacking establishing those same relationships here because the industry is a lot more dispersed here, and I have mostly been a fully DIY, self-contained universe for the past handful of years.

Plus, it’s not like this was just some old dude in the back of a shed – they had some very good looking work going on, and I wouldn’t hesitate to bring Vantruck here for some severe “paint correction” work (a.k.a please rescue my orange peel and runs because it’s not worth putting anime girl wraps on)

Okay…. with the Dumpsty chassis looking THIS GOOD I had no choice but to press onwards and revive it!

The Wishlist for a New Miku Dumpster

As usual, the inspiration for this whole affair started with a random motor.

Well, back up a bit. The random motor bit is nice and all, but I did have a list of wishes for YWIT ever since I fired it up the first time.

  • Drivetrain-wise, it used four Pololu 25mm gearmotors that I had left over from who knows whatever robot years ago. I made the choice back then because I was actually targeting it for the 12lb weight class. The dumpster alone was about 8 pounds, so I was basically making a beetleweight with a big hat. This meant the bot was limited to around a slow walking pace, turning on concrete/smooth indoor floors was a challenge, and I was losing a motor per two events or so (three in total) just from wearing them out under heavy loads! Like I said, highest mileage robot.
  • It likewise had the issue of very low ground clearance because of the small size of the motors and small wheels. It had to be helped over most obstacles including cable protectors and any higher than about 1/4″ gaps in sidewalks. So really, it was operated in open areas with low foot traffic indoors (like at Comicpalooza) or in limited smooth patches outside in full c l o u d z mode. Commuting with it was quite difficult.
  • The battery cavity was small, in order to work with packs I already owned at the time, 4S 1.8Ah ones. So I carried up to 2 or 3 spare batteries. Each one was good for about a half hour at most before it got too sluggish. I lugged around a shoulder bag full of mixed lithium batteries and pointy hand tools all the time – how positively quaint.
  • As for the smoke generators themselves, I was donated two CosClouds early prototypes by Spoon Makes which I basically drove into the ground and put through immense abuse… and they took it! Only real downside was that their capacity was limited… because they’re supposed to be worn and hidden on costumes, whereas I am basically putting wheels on a fog machine. This meant I stopped to refill it every 15 minutes or so.

So basically, I went into the new build with the following points of functionality in mind:

  • I’m going to size the drive motors for continuous operation with no thought given to a weight class, and likewise, the power parts will grow to accommodate it.
    • That said, check the giant bag of Banebots 1.625″ wheels in the photo. I wanted to keep using these wheels because I actually have about twice as many as shown in that bag (this has gotten out of hand… now there are two of them) and really only wore out a few.
  • In retaining the wheels, I’ll need to drop the axle height relative to the drive motors (a portal gear system, in a way). This will necessitate a little “lump” by the wheel that’s still relatively low clearance, but as long as the wheels are what roll over the skinny extension cord first and not the body plowing into it, we’re good.
  • The wiring for the thing will be actually planned, with places to mount things properly. The existing wiring was very “thrown in there” and I lost at least one of the CosClouds units once just by breaking the ass of it off at the connector interface, and the battery probably landed on it or something.
    • Specifically, I’ll spec a larger battery encasement than I’ll probably run – say, for a 3300 or even bigger battery, if I choose to do so in the future. Otherwise, I can still just run the packs I own.
  • I’m going to be adventurous and explore the drone enthusiast vaping rig ecosystem myself and see if I can put some heavier duty parts on it. The current generation CosCloud system looks great, but I’m seeking a lower profile blob or brick form factor. Some initial eyeballing at local vape shops (literal vape shops…. not my kind of meme vape shop) turned up some candidates I’ll go revisit.

Other than keeping these guidelines in mind, I was just out to build a damn robot again, man. Remember the quote from the end of Operation IDIocracy 1.0:

It’s time to just shit out some beetleweights. 

- me, at some point recently

Alright, so about those motors.

I have a lot of motors. Not only from my own collection, but back in 2021, I also “inherited” in a way much of the EE lab and machine shop stash of the late Dale Heatherington, my O.G. robot mentor.

I helped the family clean up the workspaces and the house, and identify things of historical or cultural value. Many of those items, such as early hand-wired PCB prototypes of Hayes Microcomputer products, went to the Computer Museum of America, a local institution. And some of them I personally delivered to the Smithsonian on a commissioned run to Washington, D.C. last year. There was absolutely no way you could convince me to trust UPS with any of it.

There were also plenty of things of no historical or cultural value. One guess as to what I ended up with! This photo is probably 2/3rds of the motors I have, by number, but less than 10% by mass. To the left is another 7 foot tall rack that’s full of….. much larger motors. Way too big for this project.

(BTW, organizing The Stash like this took weeks. I was basically combining three standards – mine, Dale’s, and my old chaotic ADHD pile. 97% of the house-related efforts of the past year have just been making my shop. I pretty much otherwise only mow the lawn…)

I dug around for a few choice candidates. I was looking either for two motors that were powerful enough on their own to be plausible, or four smaller ones that were still larger than the Pololu motors. Dale’s collection is a lot more vintage than mine (literally… than me in general) and contained quite a few old hobby robot legends like Pittman, Globe, Buehler, Maxon, Escap, etc.

Names that middle/high school me would have coveted from the AllElectronics (RIP) or MECI (RIP) catalog. Now all superceded by shitty Chinese outrunners. shakes cane and creaks

The motor on the silver bracket in the center of this lineup ended up being the most interesting. It was some kind of servomotor used for motion control, as it had a very fine encoder on the back and a MXL timing pulley already attached to the output. Dale had helpfully scribbled using a marker the Important Motor Numbers: 172 RPM/v, and 5.2 ohms line to line.

That’s really all I needed to know, because combined with the sheer size of the thing, I knew I could use two of them only and drive it on 6S (22V). And then it’ll still be spinning slow enough (3900 RPM or so), that I’d only need a roughly 2.5:1 or so reduction to achieve a healthy 7mph speed. That’s reasonable jog territory, and can be accomplished in one gear or belt stage.

They were torque beasts – 172 RPM/v becomes a (theoretical) 7.8 oz-in/A torque. I measured the stall current to verify the resistance, and so concluded the system should be capable of around 100oz-force of push per side through that 2.5:1 gearing with the 1.625″ wheels.

Probably not something I’d choose for a combat drivetrain these days, but this drivetrain will be incredibly smooth and quiet!

(I am so, so not sorry to anyone who has to deal with this private electronics surplus store in my basement if I accidentally and irreversibly drive Mikuvan off the cliff instead of perilously close to it)

With my Random Motor Inspiration, it was time to start throwing down lines.

A few critical elements define the initial sketch. First, the overall length was constrained by the fact that I have to lower this thing into the Dumpsty chassis from the top. The wheelbase was fixed as well, given the position of the existing wheels. Next, I set the ground clearance I wanted, referenced from the bottom of the chassis.

That really meant this drive pod housing could be one shape. Man, it’s nice working within existing systems, even if you’re the utter fuckup that designed the existing system in the first place! I keep coming back to the thing I told the former 2.00Gokart students – some times you have to design and make the first one shittily to know what to design and make better next time.

Even at this early stage, I was already looking at a portal gear drive from the wheel axis up to the drive pod body, where the two wheels per side will be linked by a timing belt linked to the intermediate gears. The motor is probably sitting at one end and driving another timing belt to reach down to the intermediate gear shaft.

I decided to play around a bit with the whole “smooth and quiet” part and generate myself some helical gears. I haven’t used this feature of Autodesk Inventor in forever. The helical gears should hopefully mitigate some of the inevitable whining and whirring of 3D printed gearboxes.

Inventor makes the helical gears using an extrude-along-a-helical path where the helix is already calculated for you, based on your inputs to the Design Accelerator tool. I was planning on just whipping these gears out using nylon filament, so I was after their basic geometry, to which I can append other features.

I do a thing where I pre-emptively shrink the gear tooth profile inwards using said Design Accelerator tool, because 3D prints almost always end a bit fat – with extra material, or simply surface irregularities acting like it. In the tool, you can enter a total backlash (slop) desired and center distance compensation, which effectively makes the tooth skinnier in the model.

I extruded the original sketch and hollowed it out to make a basic gearbox casing. Bearings are specified based on what I already own – drive axles are type 6801 12mm bearings, and the intermediate shafts are MR148 8mm thin-section bearings. I added a “registration feature” to the bottom of the drive pod so it can align with the existing cutouts I made to the bottom plate of the chassis.

All of these dimensions were kind of in flux and were changing as I made new geometry. For instance, the casing had to get a lot deeper once I added…

…the timing pulley to the intermediate gear. I simply stole a model off SDP-SI, cut off everything but the tooth profile (performing my pre-emptive profile shrink too), and imported/inserted a body into the gear part, then bridged the two to fuse the bodies. This robot is definitely an act of 3D piracy across multiple vendors.

I’m not sorry.

The belt in question is the 3mm HTD profile that I mentioned is my go-to for manufacturing on my tiny nozzle equipped Ender 3. I elected to use a 9mm wide belt to handle the torque of the drive motors.

The belt length I needed was discovered by a bit of back-and-forth using SDP-SI’s Center Distance modeling tool. I started with the dimensions I was hoping for, and it tells me how wrong I am. I then feed its recommendations back into the CAD model and see if I can make it work. Usually, the answer is “Yeah, fine, you’re right. Have it your way” and I then have my retribution by buying the timing belt from a random Amazon store instead of the company.

A 105mm center distance with 20-tooth pulleys actually works quite well, so excuse the glibness. I added a very tiny amount of overhead (the 0.05mm extra) as a hedge against ever needing a tensioner.

Here’s one of the intermediate drive gear + pulley combos. It will be riding on a live 8mm shaft, so the bore will be sized for a moderate press fit. I’m simply going to cross-pin the gear to the shaft to keep it in place.

Continuing on, I made placeholder parts for the lid of the gearbox and the input and output shafts. The same 1/2″ hex shafts with retaining rings to hold the wheels will be used.

After importing the wheels, I was able to size the shaft length and snap ring groove offset positions much better.

At this point, I dragged everything into the main assembly for a visual test fit. I like it so far! This test fit allows me to determine a way to stand it off from the inner side walls and also where to put mounting holes. A little bit of dimensional tuning later, and now it’s time to figure out where to put the motors.

I might have already intimated that these motors are quite large for what they’re going to be mounted in. At least, this drivetrain is going to take up far more space than the old Pololu gearmotors! I had to keep in mind where the lid servo (remaining unchanged) was going to swing, as well as make sure I leave space for the future electronics package and the still yet to be fathomed vape cannon.

I decided to keep the motors in the front for center of gravity reasons. I’d rather this thing attack obstacles on the ground with the heavier end instead of potentially kicking back and tipping over backwards.

To mount these things, I decided on simply ring clamping them, versus doing a face mount using screws. It kept the gearbox housing as a single 2.5D shape with a flat face to put down on the build plate, and it was deep enough (35mm) to give plenty of circular area to hug the motor.

Notice the small design change to the center rib with the two holes in bosses on the ends. I changed it from a straight-across rib to a curved one. The curve acts like those fences made of zig-zagged pieces of wood and adds a little more rigidity to the result, plus helps disperse stresses from the print process and adds some stability during it too.

The motor will sit like so, in the loving embrace of a circle of fiber-filled nylon.

Clamping ears for the motor are now modeled as well.

I change my view preference depending on the time of day – at night, I’ve long preferred this quasi-wireframe mode with the softly hidden edges. It lets me “select everything” as I need, is not as irritating to look at as a straight wireframe, and also works better with the generally darker night skins of the software. It’s also very aEsThETiC.

One random trick I’m adding, largely because I am non-committal as hell about the eventual belt drive ratio, is a way to easily adjust the motor up and down a bit and hence the tension of the belt. This is accomplished using a split clamping sleeve that has an offset hole in it. In this case, the offset is 1.5 mm, enough to let me swivel the motor around a central axis a total of 3mm. It should be able to accommodate any random pulley setup I can think of!

All I’d need to do is partially tighten everything with the motor belt installed, then just flick the lever, rotating the motor around until the belt is tensioned.

I’ve moved on to sizing the input timing belt and pulley. Back to the Center Distance tool we go!

A small amount of dimensional massaging (adjusting the vertical position of the motor axis using the big clamp feature sketch) and I have a pretty good solution. This is actually a 2.66:1 ratio, which is a fine conservative increase from the 2.5:1 I had in mind. With the motor in a center position in its tensioning ring, I can use a very common 60-tooth timing belt.

Here’s a full view of the drive pod from the “outside” of the bot with the casings hidden. This was pretty much ready for showtime.

The pod will be mounted on each side using four holes. Two will be in the main gearbox casing, and two will be on a Big Combo Spacer.

So… the part I care about is basically done. I generated an in-assembly mirrored model to check everything out from a space claiming perspective.

After test-printing a gearbox casing, I went back to make a design change to require no support lattices. It was kind of annoying picking the support droppings out of everything, when really only one area (the motor clamp ears) needed it. This version uses a single clamp bolt and some crafty angles that I just tell the slicer software to shut up and deal with.

In another sign of noncommitment, I added a series of breadboard standoffs (the four vertical pillars here) each of which contains three holes spaced 0.5″ apart, making for a small 3×4 grid of holes. This means…

…I can add WHATEVER BATTERY AND ELECTRONICS TRAY I WANT. At any point I want. Like this battery bucket, which is sized for a 2500-2600mAh 6S lithium pack. I already own a 6S 2200, which will be a bit loose in here, but whatever. That’s what not falling down the stairs is for (….oops)

On the other side, another Metabracket supports the RageBridge2 I used for the drive controller, as well as the…. I’m sorry, did I say earlier that all of the wiring should be planned and the electronics should have homes and stuff?

Well, consider this some kind of cash-under-the-table flophouse or a “free rent with terms and condtiions”. I just made a bunch of slots and holes to run zip ties through. The vape cannon power supply, Pololu R/C relay, receiver, BEC…. they’ll sure mounted to something alright.

This concludes the design of the important bit of the new Dumpster Miku! Next up, the fun that is fabricating and assembling it. You’d notice a scarcity of vape cannon information, which is forthcoming – at this point, I was still unsure what form it would take.

Late Stage #PostmodernRobotics: Welcome to Your Waifu is Trash, the Robot Dumpster Fire

chorl

Now that there is an #BrandNewSentence.

Ladies and gentlehams, this is a Golden Dumpster.

It’s several things. First, it’s the prize if you win the Norwalk Havoc event held roughly monthly at the 50Day Makerspace in Norwalk, CT.

Second, it’s a repainted Dumpsty desktop dumpster. No, this is not a joke; someone’s real life job in a shop and assembly room somewhere is manufacturing tiny dumpsters for consumption by edgy postmodernists everywhere.

Third, it’s about to be digitized and turned into a robot for Dragon Con.

Yes, this is truly the state of combat robotics today. This is what we’ve become, when all of the Ampflow brushes have melted away and all of the poorly-restrained Li-poly batteries have vaporized, the essential oil of the Harbor Freight drill gearbox. Welcome to #PostmodernRobotics.

So alright, what’s the actual story here? This is starting to feel a little bit like Stance Stance Revolution or Colsonbot.

In a way, it kind of is in the vein of those two. They’re all supposed to take the super cereal sport and turn some element of it upside down for entertainment. As you know, I tend to complain about how overly tryhard the sport has become especially with folks getting into the sport only building spinners (or worse…. only operating the same spinner for the better part of a decade) and focusing on win counts at rigid tournaments. Not that I have a strong moral basis to stand on here, since I obviously had my own era of “tryharding”. But regardless, in the position I am, I might as well keep at injecting whimsy wherever I can!

This dumpster is actually Jamison’s – not mine. I only took Roll Cake to one Norwalk Havoc event for kicks (where it did disturbingly well in a rumble, more than I expected by far!). It was bequeathed to me while Sawblaze was in temporary storage at Big Chuck’s Auto Body – you can kind of see it peek out in the background of a few photos.

Basically, we both wanted to see an assbot made for Dragon Con. With the time horizon for rebuilding Überclocker/30-haul vanishing, I decided to punt on the project instead of rush it (see, I’ve learned SOMETHING about project and scope management) and focus on putting on a show for the convention.

Obviously it was going to be almost completely ineffective as an entry – that was kind of the point. It’s too small to put any really worthwhile drive power in, at least not without actual engineering effort, and the dumpster itself actually weighs close to 8 pounds – it’s not a plastic shell, it’s actually stamped and welded steel sheet (speaking of engineering effort for questionable ends…).

My “goals” (???) for this build condensed as I was buttoning up Vantruck lighting. It was going to have powered flipping lids actuated by servos, and an internal vape system to emit a lot of thick smoke. I was just going to drive it around the convention as a neutral political statement – you can project any insecurity you want upon it! It doesn’t mean anything and nothing matters! True #Postmodernism!

 

I made it a curiosity goal to see if I could get it to “self right” with a linkage that allowed the lid to extend far enough. Granted, it can only ever self-right from one face only, so how much would it really matter? Does anything actually matter?

After a couple of iterations, I found a good enough solution where a 120 degree servo travel was enough to get about 200 degrees of output travel. The linkage consisting of the servo crank arm, the lid, and the coupler link in between basically was almost toggling at both ends (very folded one way or another), but was much more squared up when the lid was vertical, or when it would have the apply the most force to “get up”.

I was going to need a hefty servo – probably a 1/4-scale size – to do this motion anyway, so I was browsing for candidates as this process was happening. The short turnaround time before I was leaving kind of mandated a Jeff Bezos Special, and I ended up just going with one of these generic “60 kg” servos. No, they don’t weigh 60 kilograms, but allegedly put out up to 60 kgf-cm of torque. Do they actually? Who knows?!

They come in 180 and 270 degree travel varieties for different applications, with the latter being more rare, so I constrained the design to use the 180 degree type only just in case I had to find a spare. Always be prepared to quickly service your miniature flaming dumpster?

I was going to keep the drivetrain simple, using some 25mm diameter Pololu gearmotors I’ve had since forever. The “25mm diameter gearmotor with a centered shaft” seems to be a very commoditized motor form factor, and many generics exist and are popular with beetleweight (3lb) builders. Pololu sells a “HP” line which uses a hotter-wound motor that pulls more power, whereas the common ones on eBay and Amazon are very tame.

The drive layout was going to be 4 individual modules, with 4WD. Instead of just slinging a robot frame under the dumpster (sorry, I can’t take myself seriously typing that), which would pretty much guarantee it tipping over all the time, I wanted to hide the drivetrain within the bottom by making flange-mounted modules that attach to cutouts.

The drive bits itself are just four machined aluminum hex hubs for 1.625″ Banebots wheels I had on hand – dating all the way back to #CandyPaintAndGoldTeeth (which is a bot that REALLY could use a reboot!), and the motors are clamped in place with a U-shaped block.

I made sure to add an external support bearing for the wheels, since these motors do not have all that much output shaft support; one of the ways they get broken easily in battle is just by applying too much force to a directly-coupled wheel. They really should get external support, even in beetleweights (some get away with it using very soft and foamy wheels like Lite Flites), and definitely in anything heavier.

Here’s what it’ll look like from the bottom. The wheels are basically pushed out as far as they can go into the corners for …. stability. Yeah, stability!

I made the rest of the linkages based off the skeleton sketch model. By this point, I actually got the “60kg” servo in house, so I measured up its stock arm and designed a linkage to be sunken into it (for maximum torque transfer if this thing really can push 60 kgf-cm, I didn’t want to just rely on 2 or 3 screws)

And here we have the more-or-less final design, minus a few electronics detailed which were mostly freelanced as the build progressed.  Let fabrication commence!

Making the cutouts in the bottom of the …. dumpster … was actually more strenuous than I thought. This thing is straight up made of 18 gauge steel all around. Thicker than most Mikuvan body panels, as terrifying as that sounds. I used a Dremel with an abrasive wheel to make the cuts, then used a 3D printed prototype of the motor module as a drilling template.

These components are 3D printed in Markforged Onyx, a part of this balanced breakfast. The wheel outer bearings are thin-section Inch bearings I got a long time ago for Roll Cake – they’re type R1212, 1/2″ bore 3/4″ OD and 5/32″ thick. I rather like them for doing little inch-centric quick projects, such as this one!

Four of the motor modules assembled together. This is a very formidable beetleweight drivetrain, all things considered.

I went just maybe a little nuts on engineering a flaming dumpster by making the servo linkages all use ball bearing pivots. Seriously?! Yes – when you build a complex bot like Roll Cake, you end up with a million little bearings you bought but don’t end up using, and can peruse through later.

These bearings are type SMF83 – 3mm bore, 6mm OD, and flanged. They’ll ride on 3mm shoulder screws (also left over from Roll Cake iterations). Ball bearing pivots reduce the friction of linkage joints by a good order of magnitude or more, and reduce slop since they don’t need to have inherently loose fits – for a linkage as edgily-designed on the verge of toggling as the lid, I figured it would be a nice touch.

There are a couple of little 3D printing tricks that I’d like to point out. They’re not too visible in these photos, so it’s easier to show a CAD screenshot:

Basically every part on this bot is printed at an angle with the part not oriented “flat” like what’s common. I started doing this years ago, probably by accident while trying to print an awkward part, and realized it confers a lot of advantages.

First, you can distribute a load on a 3D printed parts in all 3 planes by tilting the part – at least going from one plane of loading to two.

In a part that has multiple orthoginal axes of pin joints or screw holes, instead of extremely compromising on the strength of one axis by forcing it to be loaded “between Z layers” where deposition modeling is the weakest, you effectively slice the part to include them in a skewed fashion.

That way, for example, my pin joints and bolt holes are all grabbing material on the 45 degree plane, each capturing many layers of material – clever part design with this in mind can even result in fasteners that capture the entire width and length of parts. This approach also makes 3D printing embedded trusses and other lightweighting shapes more effective for the same reason: the geometry permits material stress transfer into more planes and axes at once.

Second, by tilting some features, you can avoid support lattices in hard to reach places or which would cause poor finishes in a precision-requiring area. A lot of the parts were printed without support, only a “brim” on the lower 3-5 layers to promote adhesion.

The downside is your printer has to be in good tune to do this, since it involves printing almost exclusively in 45-ish degree overhangs and support lattices generated may be much smaller. I trust the Markforged machines to do it – a tunable RepRap-family printer might need some intensive training beforehand.

The electrical deck was freehanded from a sheet of 1/8″ Garolite, and the battery cradles and clamp designed up and thrown on print also. I designed around a set of 4S, 1.8Ah packs I already had, but made sure the cradle was big enough to support the next size up like a 2.2Ah or even 3.0Ah for longer run time. Remember, I was out to drive this thing around for the better part of a day, so it’s carrying about 4x the amount I would ever expect a beetleweight drivetrain to use in one match already.

While this was all happening, I was also waiting on paint to dry. Because of course I will make it Miku-colored!

I ran into some serious issues with painting it, though – the Golden Dumpster is in fact made from painting gold over a regular green dumpster, which is green paint over the bare steel. So I was adding shitty spraypaint on shitty spraypaint on more shitty spraypaint. Some of it started crinking up and detaching in chunks, which was unfortunate. I decided to just forget about appearances – a weathered dumpster with chipping paint and dents just sells the story!

Anyways, here’s the motor pods and electroncis deck dropped in and screwed shut.

The electrical system was actually a little bit of a conundrum. The big servo preferred 2S lithium voltages (7-9v or so), the drive motors were 12V rated so it’s best to run them at that or higher (hence the 4S battery chosen – 14.8V nominal). Most vapes run at only 1S (3.7V) or in that neighborhood. The receiver could take up to 12V in, but the Pololu R/C relay should stay around 5V for best reliability.

I ended up digging out an old adjustable BEC (R/C power supply) from the pile of robot electronics, which could be tuned to 7.2V, an old 6-cell nickel battery voltage. This would feed the vape power (and a mini air pump that is needed to actually push the vapor out with force) and the receiver. I tested the Pololu relay and inspected its diagram to see if there was anything that bad about running it on 7.2V instead of 5 – nothing ostensibly besides possibly overpowering the relay coil itself. Overall, 7.2V was a reasonable compromise for everything on board.

The BEC unit and Pololu R/C relay get globbed onto the electrical deck, next to the most overkill possible controller for this application… a spare RageBridge. Talk about being under-employed!

The smoke effects actually got super interesting. What you see here is two CosCloud units provided to me by their creator Alina. They actually figured all of this conundrum out for me (and you too! and many others!). I was otherwise going to a literal vape shop to buy literal vaping gear to make my own.

I’ll be running two of these “smoke generators” in series to line up with 7.2V pretty well, and the air pump in parallel with that group. This whole assemblage drew around 3 amps at Full Vape, which was quite tolerable with the 10A-rated BEC.

The final packaging and zip-tying now commences.

And here’s what it looks like from the outside! The final weight – 11 pounds 8.8 ounces. Hey, there’s room for a… weapon?

The dwarf dumpster fire, in its natural habitat.

And what it looks like in action at the convention. The secret is to turn on the Vape Turbopumps for a few seconds, then open the lid and watch it all waft out. This thing was an absolute riot everywhere it went, and it fulfilled the “neutral projection target” social experiment perfectly. I heard quite the range of things referred to while pointing fingers at it, from the usual remarks about Congress and the presidency, to the Marvel Cinematic Universe, Disney, Game of Thrones, Star Wars continuations, game companies…. there are apparently many ongoing uncontained dumpster fires in society as we know it.

Some of them even move around and come to you.

You can see #YourWaifuIsTrash in action in this Dragon Con music video (skip to 8:15 if the time link doesn’t work!)

And finally, we make it to the Big Day at Robot Battles!

I’m a dope for not getting to it sooner – somebody had to point out to me that the dumpster was missing an UWU face (yes, it’s a face, not a sound…. today you learned?!), and so one was hastily appended with markers.

This thing managed to actually pull off a win. I did it by basically bull-baiting the opponent at the edge of the stage and letting nature take its course.

Of course, it couldn’t push anything, could barely get over the “arena” borders which are left beat-up on purpose, and had issues with turning in place under its own weight and the traction of the outdoor style carpet of the stage. So that was all! I also entered it in both rumbles at the end, but it got bowled over rather quickly in both rounds.  Nevertheless, I consider #YourWaifuIsTrash a win, because what is a win in #PostmodernRobotics? It’s whatever you want it to be!

This thing came back from Dragon Con working just fine, so it will live on as a stupid prop for all of eternity. I’ll probably repaint (read: just spray over it AGAIN in more shitty spraypaint!) it soon and begin adding choice tasteful decals.

Do you want your own? The CAD files in STEP, IGES, and Parasolid are available on the References page!