Archive for the 'Chibi-mikuvan' Category


Chibi-Mikuvan: Completing the Bodywork

Apr 21, 2014 in Chibi-mikuvan

Last time we visited Chibi-Mikuvan, it was still a mottled fleshy-pink blob of fiberglass, insulation foam, and microballoons. Over the past week, I went full artist and completed most of the paint and decal work that turns it into something vaguely resembling its larger namesake.

Corrections! I turned the pink blob of foam into something that looks truly diseased and disfigured.

This was taken after I already put one layer of primer on. I already knew that the fabric texture would show through because my layup was too dry – caused, it seems, by too much squeegeeing. I decided to not glaze or fill the entire thing and only focus on the really troublesome parts where the cloth looked basically unwetted (it was, but there was not nearly enough resin to fill the gaps). The primer was used more as a way to see where the true problem spots were. After the spot putty dried, I sanded it smooth and put on another coat of primer.

In retrospect, this was a bit of a bad idea. The spot fills with glazing putty looked great! after the paint, but everything else just looked so contrasting. If I were to spend much more time on this, I would have either microballooned or thinly Bondo’d the whole thing, to fill the texture, before any painting at all.

Oh well

After all the primer and spot putty wads dried, I went over everything manually with 220 grit sandpaper to smooth the high putty spots down. After that, it was time for several coats of white. This took about 2 days, because I decided to do everything accessible from the top, wait for that to dry, then paint the less-critical bottom edges.

With the white base coat dried, it was time to add the dark gray and black window detailing. I used an “inverse stencil” method here, which I am sure is a legitimate thing but I re-invented it in 5 minutes on a laser cutter. I basically cut the outline of the shape I wanted, masking-taped it to the location I needed…

…and very carefully trimmed out the original paper, leaving a masking tape outline. Then, it was easy to apply much more masking tape to close off the areas that had to remain white.

To give additional cover after I got far enough away from the masking tape edges, I just hung some spare poster paper down the sides.

The first window coat is a dark gray. I wanted to use a more pure black to highlight the actual windows, and the dark gray to form what is the blacked out portions in Mikuvan’s actual paintjob – which I called “starship” for the longest time, but it’s actually called raccoon. – _ – (Concept image by Cynthia, of giant mechanical RWBY scythe fame. It shows a window blackout continuation I want to do eventually)

The next day, after the dark gray coat dried, I repeated the same inverse-masking process to mark out the windows. I planned on only adding this detail to the windshield and front door areas.

When the window details were “dry to touch”, I moved the masking paper around a bit and repeated the inverse stenciling for the front bumper detail. I decided to make this a medium gray shade since it reflects the somewhat sun-faded color of Mikuvan’s front bumper.

The next day, it was time to peel the masking paper and tape off!

So my “dark gray” turned out to not differentiate well with the pure black under indoor lighting conditions. It’s more noticeable outdoors or under natural daylight. I could have gone lighter a shade or two with the gray and gotten the same point across. After a gloss clearcoat, the two colors might look even more alike.

Test fitting the body! At this point, it’s missing detail lines. I decided to hire this out to Brian Chan, Maker of Like, Everything, Man. The rationale was that if I tried to make “detail lines” they would for sure end up being detail squiggles. Brian’s artistic background means his hand is far, far more calibrated than mine. We chose to only do a couple of major panel lines to get the idea across.

The operation was over quickly. My only bodywork task left is to add the taillight detail, which is going to be only an orange, light silver-gray, and red patch. This version of the body will not have functioning lights – short of maybe an EL panel or two.

Chibi-Mikuvan is not impressed by your 2.00Gokart.

Notice the handlebar change between this picture and two ago. The body fitting told me that the big drop bars I got weren’t going to work at all : they were simply too huge. It restricted movement of the steering linkage, and the brake cables would run into my leg.

I went back to Cambridge Bicycle and had them yoink another junk handlebar from the pile – this time, it’s a plain U-shaped bar that was both narrower and did not curve back. This made it have way more upper body room. The brake handle position shown is likely the final one.

It also looks like antlers, or oni horns.

While my students were working on their vehicles on Friday, I popped out some more silly Shift-JIS emoticons out of the fridge magnet sheet. I plan to have a whole bunch of mix and matchable expressions, maybe for things like winning or losing or whatnot.

No, really, this was my whole motivation for building this entire thing. I swear.

The next post in this series will detail the extent of the mechanical work that’s occurred in parallel with the body shell. Finally, the Epic Inrunner has been mounted!

Chibi-Mikuvan: Adventures in Foam-Core Sandwich Composites

Apr 13, 2014 in Chibi-mikuvan

I would never make anything out of composites. There’s too much making of the part that makes the part that makes the part. I prefer to just make the part.

- Charles some time in 2011, probably not on this website but at least once in person with someone

Okay, okay, you win, 2014-charles. If this exact time last year you’d told me I would be a composites nerd today, I’d have left the room in a huff.

However, you could also have the said the same thing about becoming a van mechanic, or becoming a shady back-alley auto body man.

Welcome to Big Chuck’s Mobile Garage & Auto Body!

Seriously, there’s enough equipment back there to do everything short of rebuilding the engine on the side of the road, and that’s only for lack of parts.

Hell, if you told 2011-charles he’d end up becoming the very thing he loathed the most back then: a machine shop instructor…

Anyways, last time I left off, the two halves of the foam core were drying independently. The reason I had to split the core into two halves was because I mistakenly welded the axles 1 inch too far apart, necessitating a 1″ sliver inserted into the middle to make up the difference.

I used the hot wire cutter that I built in MITERS back in 2011 in anticipation of making Chuckranoplan 0005 (do not tell me that I’ll become a combination naval architect and aerospace engineer next year) and a MDF template to slice some 1″ thick foam up into the cross section of the sides. By this point, I’d gotten my own shipment of microballoons and epoxy, so I applied this patch the “correct” way with microballoon putty. This stuff beats the colloidal silica slime I used before, because it’s actually sandable – fairly easily. I used the heavier duty stuff because a bucket of it was easily accessible, but I see why it’s only used for very high strength areas, edges, etc.

I did a practice piece in the back first where the geometry was easiest.

I generally followed the procedures outlined in The Burt Rutan Book, since I had gotten a copy of it a while ago. This book was what ultimately threw me into this direction – my gripes with composite materials originated from witnessing people spend tons of time making molds and plugs just to produce a single part. The foam method seemed like a way to get to the end quickly, and if it’s good enough to, like, get in and fly yourself it surely must be good enough for this!

For this practice piece, I took no shortcuts and went ‘by the book’ as much as I could. That wouldn’t last long.

I tackled one of the broad sides next. To keep things as uniform as possible, I kept the outside faces as one large sheet of fabric. I laid an oversized rough-cut sheet on the surface, anchored it in a few places with some epoxy blobs, then trimmed to rough shape.

The front was done in the same fashion.

One thing I had trouble with consistently while “Rutaning it” was controlling how the fabric stretched and deformed. The first few square inches of contact with wet epoxy essentially determine how the rest of it will go on, and I found that I couldn’t really push one area in the direction I needed afterwards without having to pull from across on the other side or slowly work a wrinkle back towards the center and out the other side. This method clearly requires patience and methodicalness to get right.

The ideal shape of the rear inside would have been having the black edges roughly parallel to the sides. The edges seen are the ends of a long strip that I wrapped around the backside – the exterior is continuous, the inside is messy and ragged.

…so I said goodbye to the Way of Rutan, realized once and for all that this is in fact not going into space or even in the air slightly, and began slathering. In fact, if it ends up in the air even a little bit, we can assume something went terribly wrong.

Now I was on the interior, so I could reuse the scraps and cutoffs from the other faces. My tactic became laying the fabric out dry so I could have control over its shape, dabbing a few spots to anchor it, and then begin pouring on the epoxy-microballoon slurry, working it in like I would do to non-filled epoxy.

This inside right face is made from about 7 or 8 pieces, most of which were spent on the little concave wheel cutout in the front end. For the other side, I smartened up a lil’bit and cut a semicircle shape out of the cloth stock.

The process went by much, much faster after this. Here, the whole body has been covered in at least 1 layer of fiberglass. I decided to stop here, instead of trying to add layer 2 everywhere. I reinforced the corners and front outside edges where it will most likely bump into things with another layer only in those areas.

This whole adventure actually occurred through several non-consecutive days, so in between sessions, the layer of glass cured fully. To ensure good bonding at the intersection of cured and fresh, I returned to the recommended Rutanistani procedure of roughing up the zone of overlap with 220 grit sandpaper.

After a day of settling, it was time to start on the smoothing and sanding.

To finish the inside curves and radius portions of the wheel cutouts, I bought one of these…. fuzzy abrasive foamy wheel things (poly-abrasize wheel) that attach to a drill. This allowed me to grind off the edges of the cloth and make the area generally smoother.

I didn’t spend too much time on this portion since nobody’s going to be staring at it, but wanted to get rid of the stuff that was sticking up.

I did a first round of “orbital sander low-pass filtering” reveal the low spots and knock down some of the more prominent high spots where the glass fabric stops. The plan is to just fill every location the sander didn’t hit with some Bondo and call it a day.

One thing I wanted to take care of before reducing this problem to a known solved one (“Van bodywork”) was to make the mounting facilities for the silly Japanese style emoticons. As you might be aware, the whole existence of this project hinges on these silly emoticons.

I decided to try and make them out of ‘refrigerator magnet stock’ and have them be rearrangeable on a thin steel surface. Flexible fridge magnets are made of iron oxide powder mixed into a rubber backing. The idea is that the laser cutter can melt through the rubber and leave me with an arbitrary magnetic shape (even though the heat might demagnetize a small portion of the whole thing).

The steel stock shown here is common galvanized steel roofing flashing, for weatherproofing household rooftop HVAC implements. It forms and cuts very easily and is extremely, extremely sharp. I used to use aluminum flashing extensively in 1 and 3lb battlebots back in the day [when such robots could win anything].

Using a pipe and gentle hand pressure (Like, seriously, this stuff is sharp) I formed the flashing into the curve of the lower front side. I made it just a little tighter of a curve such that clamping the flashing down to the surface will cause it to flatten out fully and conform to the curve.

Here’s the extremely ad-hoc clamping configuration for the upper portion. The lower portion was being adhered to what was basically a smooth and flat surface, but I didn’t have a good way to clamp it.

I therefore decided to make the first attachment using E6000 contact cement (c.f. Goop and other ‘construction adhesive’ type products; the shit that reeks of brain damage). I painted a thin layer onto the metal, then slapped it into position and quickly peeled it back off. This created a layer of the cement on both fastening surfaces. I then came back 5 minutes later and slapped it on once more.

These contact cements count on the evaporation of the solvent to create a super tacky, microscopically porous surface that then instantly fuses to itself. Using it between two nonpermeous surfaces, such as the cured fiberglass and steel, means the solvent has to dry out substantially first, or it’ll never set (Guess who tried making a composite E6000 and aluminum flashing laminated frame for some of his first bots, and took apart the frame after the bot was destroyed and found that the center of the sheet never cured!)

Time for Stage 1 of Bondo. Shown here is the right side, which was a bit of a disaster because I made a cutting mistake that left the outer surface slightly concave, and tried to make it up during glass application with dry microballoon putty, but made it too wet so it just flowed and blobbed everywhere. D’oh. The biggest splotches of filler are therefore on this side. Otherwise, the only other major use location is where the steel flashing is bonded, since there’s some thickness to be made up.

After letting this cure for a few hours, I did another “low pass” sanding run. Upon the completion of this round, only small low spots and divots remained, which I stuffed a little bit more Bondo into. While it was curing, I turned my attention to the fridge magnet stock:


This stuff machines like thermite. Like, actually. I was fearing for the life of my lens, but luckily the air assist kept the flying molten iron balls away from anything important. Because the material is iron powder mixed into rubber, it sort of explodes when cut – the laser vaporized the rubber, and the high temperatures melt the iron powder into little droplets:

This is the end result. Not the cleanest cut, of course, but all the little balls fall out when you pop the piece out. I found through making a handful of these test pieces that there is indeed a demagnetized heat-affected zone close to the cut, so the less heat put into the part the better.

I therefore left the laser on a setting where it would penetrate maybe 75% through the thickness, then flexed the rest of the material out.

So with this operation successful…

I can stop now, right? This is all I came to see.

I knocked out the last few filler smears (one can be observed on the front left corner) with primarily manual sanding. The most versatile and flexible sanding block is still you holding a wad of sandpaper. Power sanding at this stage would have been too risky because I might get too zealous in one spot and go right through the Bondo and fiberglass. Nope. That didn’t happen at all. I swear.

After I was satisfied with the outward appearance of the body, I moved onto bonding the body mounting brackets. These were 3D printed over one of the nights when everything was being left to cure. They’re fairly thick and bulky brackets which hopefully will have enough attachment area to handle some bumps and impacts. Now it’s time to let this fully cure and come back in the morning to verify fit.

Once this is done, I’ll move onto priming and paintwork.

In the mean time, let’s return to mechanicals:

Check out that handlebar. I obtained this for free from Cambridge Bicycle after showing them what the hell I was building (MITERS and Cambridge Bikes go way back, too, which helps). However, as awesome as it is, I might have to ditch it for a smaller handlebar because it’s too huge. I do like having the ability to attach standard bike/scooter handle accessories like brake handles and throttles, so if I do let this go, it’ll only be for a custom-made one.

I’ll have to do a full ergonomics test once the shell and myself are both installed in their final positions.

The steering linkage is hooked up with 5/16-24 threaded rod. Unfortunately, one of them ended up having to go through the frame – I had to cut a 1/2″ deep notch in one part of the frame. I chose to do this rather than lower the center steering link further because it’s already pretty low – it would become the first thing to hit something I go over (ahead of the battery pack, anyhow…).

I intend to install more mechanical parts while paint dries in the near future. On deck to be completed in this realm include the following tasks:

  • Machining the 12mm brushless motor adaptor shaft for the gearbox
  • Machining the sprocket to fit the angle grinder gearbox output shaft
  • Slightly modifying the gearbox mount because the angle grinder box is a little bigger than my rough model on one end
  • Hooking up the brakes to the handlebar

After this, I’ll need to turn back to electrical system work, starting with remaking the battery pack endcaps – I have not observed any “95% scale” weirdness with this shell, so I can only assume that it was a fluke, or someone changed the setting back between my machining periods a few days in between?!


All-around Updates: The Anime Boston Recap, 2.00Gokart Round 3, & Chibi-Mikuvan

Mar 28, 2014 in Chibi-mikuvan, Electric Vehicle Design, MIT, Bostoncaster, Cambridgeshire

I have resurfaced.

Like last year and the year before, major gaps occur in my webospheric presence because I’m busy with corralling a group of undergraduate students before one of them sails right off the 3rd floor of the building on a silly go-kart – 2.00gokart is running again this spring, with only minor changes to last year’s edition. In the mean time, Chibi-Mikuvan has been progressing slowly.

Because this post is going to have some obvious… length issue, here are jumps to the directly relevant sections:

  1. The Anime Boston recap (a.k.a “My Fangirls Rejoice”)
  2. 2.00Gokart’s current state
  3. The work on Chibi-Mikuvan since a few weeks ago

amine boson

One of the quirks of running a shop is you actually never have time to use your own shop; go figure, because you’re either fixing something or helping someone else with tools or their own projects. And many projects there were in the past few weeks: the IDC has been the ad-hoc command headquarters of the MIT Anime Boston brigade, probably because I know everyone and am there all the time anyhow. I even joined in this time! Historically I’ve been an incredibly lazy cosplayer and have just done whatever involved a lab coat or sunglasses.

Except that one time at Dragon*Con, with the 6-foot steel venier caliper, but I digress…

I think the only other time I went above and beyond for a character was the RazErBlades, which were more independent project than not, and I threw together an easy version of the character for Otakon 2010.  My interest in this domain is primarily that of impractically large transforming mechanical weapons, but I have  never taken it seriously enough to work on one as a project in its own right.

But that doesn’t mean I won’t facilititate or encourage people doing so. Check out Jamison’s Impractically Large Transforming Hammer-Cannon (from League of Legends) That He Finally Blogged About, for example.

And the one I am partial to, Cynthia’s Impractically Large Transforming RWBY Scythe.

I’m a fairly close follower of the RWBY series because of its proportion of fancy transforming mechanical weapons, all of which I say “Monty Oum, you son of a bitch.” to because I’m almost certain there are some volume discrepancies going on, such that if you actually built the weapons and had them function visually in real life, there would be no place to put the actual weapon part. But Monty designs are hardly the worst perpetrators of this in all of impractical transforming weapon history. In fact, I think a lot of thought was put into their actuation and mechanical design for the most part, especially for Ruby up there. Things at least don’t magically appear and disappear.

Anyways, I strongly recommend reading through Cynthia’s design and build process because it covers all the bases of design to iteration to implementation, all in a first large mechanical build. My own first mechanical projects were way more, umm, ad hoc organized. I saw a ton of the prep work that went into it, and my involvement was limited to some backend hardware support only (e.g. “make 23 of these e-clip things” “ok”) during the final days of the build. So yeah, here’s the link again to Cynthia’s Impractically Large Transforming RWBY Scythe (which I am told will become slightly more practical but no less large soon)

I elected to make myself some accessories and attend the convention as the fan-made “Rule 63″ of this character:

Image links to creator’s DA page

Okay, you can’t give a character that weapon, call him Garnet (I must use literally a ton of this stuff a year), and expect me not to bite.

Luckily, it still ranked on the favorable end of “easy” to “You have to be cosplaying for as long as I’ve been building robots to do it right”. I even made up a “make versus buy” chart for it:


One of the end goals of all the “Makes” was to use some of the rapid prototyping tools I had in the IDC which could be reasonably accessible to anyone – the Replicator 3D printer (which has sort of become our “student beater” machine), and a laser cutter. Harking back to my Maker Resources 2013 presentation at Dragon*Con, this year our group of maker-oriented con-goers (not to be confused with conga’ers) decided to host a panel with much the same topics, geared specifically towards costuming:


Unfortunately, I don’t actually have many pictures of my build, largely because it took place during the last two days before the convention where I was mostly aiding the completion of other props and handling 2.00gokart. But stuff did pan out. For instance, here’s a rose sigil thing that I made from the scrap outline of Cynthia’s rose cutouts. The bracket on the back was designed to snap-fit into the buckle of a large tool belt I bought and chopped up for this purpose.

The material is “silver” PLA, which I bought from Amazon (and which Zenn Toolworks hasn’t stopped calling me about to review… yeah guys, it’s filament! It’s all melty and stuff! Yay!)

It’s not silver in the traditional sense, but more like a gray PLA with some metallic sheen in it, which I suppose gets the job done.

A ton of other parts were also made in this same PLA – the large ammo boxes, for instance, which I drew up to hold two side by side .50 BMG rounds up to 6 high, because I don’t know if that was the idea or not, but had the dimensions of the .50 BMG in front of me. Plus, the scythe is allegedly also a .50 caliber sniper rifle. The ammo boxes had removable lids and actually served very practical purposes during the con. Besides those, the decorative crosses were also made in the same material.

I also used it to print the decorative bullets for the belts. Unfortunately, due to time, I didn’t make anything else to hold them to the belts, so we went without them – that will come later. I had a few feet of black cargo webbing and decorative rivets ready for the task.

Termed by my compatriots as “nerd-cute” or “nerdorable”

I think I shot pretty close in the end, eh? This was on Sunday, after Cynthia had broken her scythe demoing it so many times, so she designed and made a comically small stand-in from leftover MDF.

I ended up not being able to attend the panel, unfortunately, since Friday night / Saturday morning I was plowed over pretty hard by a (what I assume was) exhaustion-driven opportunistic flu/cold. It was bad – as in, I could barely stand up and walk straight bad. I ended up sleeping most of Saturday off in the IDC, and felt better enough on Sunday to tag along to the convention, since I paid $60 for that entry pass, dammit.

Which I lost, by the way, some time in the chaos of the week and had to borrow Nancy’s. Also, I never thought I would ever buy suspenders for any reason in life, but this was it. (They’re black, so not really visible in the picture).

While I have no plans for Dragon*Con but to bring back this character in more detail, Cynthia plans on a full rebuild of the scythe, so I’d pay attention to her site in the mean time.

2.00gokart: Year of the Weird Angular-Framed Karts

It’s back!

This year’s 2.00gokart session saw record application and enrollment. I had 64 students, mostly mechanical engineering sophomores but there were some serious left-fielders like Architecture/Visual Arts (course 4) seniors apply, but could only take 20. Man, my acceptance rate is getting to be almost as bad as MIT itself.

I expanded the field to 20 students (i.e. 10 vehicles) this year because of the experience over the summer with dealing with 27. It’s incrementally not much worse, especially now that I’ve produced more reference and lecture material which has cut down greatly on the time-consuming basic questions. The operations are now much more streamlined now that I have experience myself.

The rule changes this year are very slight, but are again designed to put a little bit of a twist in:

  • Your vehicle now has to fit through a standard U.S. doorway, basically 33″ wide. Sorry Nelson. And like half of everyone from last year…
  • You now have the option of getting once nice wheel…. or two Harbor Freight pink wheels. Cue evil laughter here.

Besides that and some minor clarifications, it’s the same thing as last year.

As usual, the class started with a “demo night” where everyone got to see examples from last year, and past students dropped by to visit. Here’s dgonz giving a short informative talk about the dangers of “dgonzing” in your chassis design.

A few weeks in, and people have their first orders.

This year, I got sheets of hardboard (Masonite) from a local wood distributor for super cheap, so everyone got to prototype their heads off on the laser cutter again…

…before I committed it to metal.

There are a lot more teams daring to wander outside the safety of 90 degree angles this year, including Triforce-kart up there. Which I’m sure is not the actual name, but I’m naming it just like I’ve named other things “5-degree-kart” (for having a frame that is a 95 degree trapezoid) and “bus-ass kart” which… you’ll see later.

The hardware in general is more robust this year, I think in part due to the added lecture / reference content and the availability of more examples from past years. Remember – these don’t come from kits, each piece is cut or machined from the students’ own designs.

In general, the design diversity is up from last year, which is what I want to see. Here’s one of the three teams that have elected to do live-axle, but they’ve also went and bought a differential (and named it Humphrey…). All this equipment came from Surplus Center.

Electrical system-wise, Kelly + SK3 still rules, but there are more dual-motor drive setups as well as one team going super experimental and taking a shot at using the Trackstar 200 – the big one. I eagerly await the results of this test since I’m using one on Chibi-Mikuvan.

Speaking of which…


Ah, now the section that will take the longest since it’s about me!

In the past few weeks, I’ve completely welded and assembled (and painted!) the frame, plus gained much more experience using the Shopbot CNC router to make the foam cores for the body. I also tried making the battery pack housings, but there’s either some quirk of the machine I’m missing or my parts are all scaled around 95% in the Y direction. The frame is almost mechanically done, upon which I’ll focus on getting the electrical system installed.

In the last episode, I cut all the tubing parts to size but had not yet put anything together. I had one giant weekend of welding some time ago in which I assembled the entire frame. The first things to come together were the steering knuckles, which also mount the brakes.

After dialing in my practice again with those, I decided to work on the motor mount. Recall that I’m using an angle grinder gearbox to “preduce” the motor speed before it goes into a chain drive. What better way to mount an angle grinder gearbox than with the disc guard ring it comes with? I cut the ring off using the other large angle grinder, then wire brushed the paint off in the critical areas.

Here’s the motor mount welded up, along with some of the outer frame parts. I wanted to put together as much independently as I could before joining the long frame rails, just so there was less fixturing shenanigans.

In the welding room…

I made tack welds to the frame with a TIG welder, but then came back with the MIG welder to finish the beads. This strikes many people as weird – and it kind of is, since typically you do it the other way around. My rationale is, the TIG allows me to exert no force at all to make the tack, whereas the MIG will always have a little wire poking your part and could therefore move the fixturing.

Well why not fixture better? I think the reason I fixture tenuously – generally with only those red magnets – is the same reason why I can’t finish the weld with the TIG welder. Time and patience. I’m insufficiently patient to do a nice TIG joint, when patience is the key virtue in getting a good one.

Spray and pray!

You know what they say, though. A grinder and paint…

…makes up for a welder who ain’t.

To be fair, there’s no shitty welding on the frame, but I also don’t take myself seriously enough as a welder to say anything more insightful.

The frame was thoroughly cleaned with acetone first, then I put down a few coats of self-etching primer followed by black engine enamel paint.

Overall, after a day of drying, it came out very nicely.

I test fitted some hardware to gain more insight on the lengths of spacer needed for the front wheels. Now that they’re mounted, the front disc brakes look even more ridiculous. Seriously, I think I could stop 3 other sketchily-braked entries in the next PRS race.

Rear axle in bearing blocks installed.

A closeup of the front axle kingpin and spindle assembly.

It’s up on four wheels! No steering parts yet….

The steering column supports are a fairly classic tactic around here of drilling some holes in Delrin (acetal) blocks. Acetal is a bearing plastic, so it’s super slippery while being pretty rigid. Two pin-jointed blocks constraint a steering column at any angle you please, then two shaft collars (one on the bottom, one up top) constrain it axially.

The driving link at the bottom is welded onto the column – instead of bolting into a face-drilled shaft collar like on Chibikart. This is just for expediency.

This forms the extent of the mechanical work as of yesterday. Likely right after I hit “post” here, I’ll go hook up the steering linkage and use a vise grip as a steering wheel and get pushed around the hallways.

Here’s some Shopbot work making the battery pack sides and the bodywork!

The battery pack sides capture the Fusion Sticks into groups of 5 so I can parallel the cells. They’re 3D milled parts by design; I guess I could split them into 2D layers, but I wanted to learn the 3 axis milling mode of the Shopbot. The material of choice is a 2×10 chunk of sanded fir I bought from Home Depot. To stay within the PRS budget, I need to make this from something reasonably strong but cheap, and wood actually qualifies well there.

After roughing and finishing, the end result looks pretty good!

I made two versions. The one on top is both roughed and finished, and is the best quality. The bottom one was one of my attempts to shortcut the process by only roughing. The Partworks software that came with the Shopbot is sort of limited in the things it can do – it’s for beginners and general non-engineers, after all, so isn’t full featured like MasterCAM or HSMWorks. It won’t cut the part out in “roughing” mode, only finishing, so I tried to trick it into thinking the part was thicker than it actually is such that it would “cut out” in the roughing cycle by virtue of stopping too far down. This did work, but the internal features were then too far down also!

Seems like the only way to really make this work is to make sure the stock is thinner than you tell it, which couldn’t work in this case because the parts are 1.5″ thick and so is a “2 unit” dimensional lumber.

I’ll just put up with the extra 30 minutes of finishing. What’s weirder is that these parts are seemingly compressed in the Y direction by about .05 inches consistently, almost like someone put in a “scale 95%” in the program that I haven’t found. I’m going to try running another version sideways (long direction oriented in Y) to see if it is signficantly shorter, which would tell me “someone set a fixed scale percentage”, or still 0.05-0.1″ shorter, which would indicate to me an offset problem.

With some lessons learned making the battery sides, I started routing out the foam cores which will eventually be between the fiberglassy bread in the composite sandwich shell.

I bonded the foam together with slow-curing epoxy that was filled with “milled fiber” until it was pasty. Which is really overkill since it seems like 77 spray adhesive worked just as well, but let’s keep it legit since this thing is totally going into space after all. I made several bricks that were to contain the four sides of the body.

Foam is messy. You can’t use the dust collecting nozzle because it would hit the part, so the foam flies everywhere and generally covers everything. And it’s ultra static-y when you try to vacuum it up.

Doing a finishing pass after the initial rouging!

Foam machines like a dense air, so I set the machine to run as fast as it could. All of these parts finished in around 1.5 hours.

One  of the sides right after cleanup.

The foam was held to the MDF disposable surface by…. hot glue. That’s it. I drooled hot glue in a vaguely grid pattern, about 2 lines per foot, in roughly the shape of the part, then slammed the foam brick down before it cooled off. It worked great!

To dislodge the part, I used a giant dustpan to split the hot glue under the edges, then slowly pulled up with it.


Too bad that I welded the front wheels 1 inch too far forward – mistaking the end positioning of a dimension while jigging it up.

This means two things: One, that the frame needs to be cut in 2 places, and 1 inch subtracted from one side and added to the other, or two, the body has to get split and a 1″ foam extension added.

I went for the second, since it would also make the gluing easier to manage (I didn’t have any 48″+ clamps). So I split the body in half on my hot wire cutter. A 1″ cross section will be made and bonded to the body, bridging the two halves, and the whole thing rejoined with carbon fiber rods running lengthwise to give the bridge some structure.

Bonding the rear panel to the two sides was easy, since it was nice and arch shaped. I just piled heavy things on top to keep everything down, and used 1 clamp in spreader mode to set the angular displacement (It wanted to lean to one side). For added legitimacy, I used the fiber-filled epoxy here, as Burt Rutan would.

The front half was also easy, just another mess of clamps.

I received an order of 2 gallons of Nice Epoxy yesterday, and otherwise have all the supplies needed to do the fiber layup on-hand. I’m hoping to get to it this weekend, but it miiiiiiight involve a little more psyching myself out beforehand.

For now, I’ll work on getting the frame to mechanical completion because then I can wave it in my own students’ faces to encourage them to finish!



The Legendary Saga of Chibi-Mikuvan Engineering, from October Until Now

Mar 07, 2014 in Chibi-mikuvan

Welcome to another edition of Big Chuck’s Automotive Blog!

Well, I guess this time it’s Big Chuck’s Miniature Automotive Blog. I actually have not had a full engineering post on Chibi-Mikuvan, which has been in on and off development since late September. I’ve talked about it in bits and pieces, and did some parts investigation type posts:

It’s also gotten some random cameos and teasers in other posts. I feel bad, though, when I do too much work and then don’t say anything about it. So here it is – the start to finish of Chibi-Mikuvan so far, ending last night. This warrants another “get a drink” warning.

We begin with the first picture seen in the introductory post. I found a drawing from carblueprints that showed the Mitsubishi L300, the equivalent in non-US and Japanese markets, into a 2D sketch and began tracing it.

I imported the picture at an indeterminate size, but kept shrinking it until the nominal dimensions matched up with Inventor grid lines.

The “ground point” of this design was wheel size. I was planning on using the dreaded Harbor Freight 8″ pink wheels, which are actually 8.5″ OD. This was scaled to fit the standard 205/75-14 tires of the L300 / Delica, at 26.1″ nominal OD.

Coincidentally, everything lengthwise ended up almost exactly something nice: 28 inch wheelbase, 53.5″ length overall. I picked a neat-looking number for the height at 23.5″.

Here’s the more finished sketch. I simplified the geometry some (how can you get simpler than this?!) like using pure circles for the wheel cutouts and arc segments for body lines. The US version which real-Mikuvan is one of has a longer snout – it’s actually even longer than the scale model shown here, but the proportions were off in this scale model such that when I modeled it at the full length, it was just hilariously bad looking. I settled for a “looks reasonably funny” number for the length.

Width was the problem. Japanese cars are pretty narrow – at scale, the overall width of this thing would have been only 22″. That’s narrower than Chibikart, and not only would I not fit inside the body shell, it would be very tippy. I decided to start with a 26″ wide shell. Because I made all the sketches at the origin’s midplanes, I symmetrically extruded outwards in both directions.

This is the solid reference model for the body shell. At this point, I still hadn’t settled onto a construction method yet. One way was to build an internal ‘skeleton’ of sorts, around the perimeter, and cut out the sides and panels in very thin material like polyethylene plastic. Another way was Epic Thermoform, and another still was a milled foam shell.

Isn’t it wonderful to be able to model 90% of your car’s bodywork in three solid features?

Coming from the Beyond Unboxing posts, I made critical-dimension models of all the parts I wanted to use. The shell seen here is a fake hollow version to make it easier to work inside the assembly. Shown is the angle grinder box, the Pink Wheel of Maleficence, the T20 inrunner motor, and a Ford Fusion dynamite stick.

The frame layout game begins. This time, it was a bit easier because I already knew what the wheelbase and track had to be. The diagonal line shown is an approximate Ackerman linkage reference line. I also knew how big the batteries had to be, so it was a matter of designing a big ladder around that.

I went shopping for stock go-kart sprockets right away. Basically, the seller which gave me the most technical data on something won – I spec’d out a 60 tooth #35 chain sprocket from mfgsupply (not to be confused with, which can also supply you things if you so desire), along with a matching hub for a .75″ keyed shaft.

I’m too used to building long continuous one-layer frames with 80/20 rail. Not being able to make continous frame rails was kind of bugging me, even though I knew that I could just butt up steel frame tubing and weld. I briefly toyed with this idea of milling “Lincoln Logs” (muscle memory forces me to type Lincoln Labs)  to allow me to make a continuous frame rail.

Shown in this image is one of the bearing blocks I plan to use – these come from Surplus Center.

The steering parts are now getting a bit more fleshed out. Because of the need to support the battery pack as well as to pass a solid rear axle in bearing blocks, I designed an “asymmetric Chibikart” kingpin and knuckle setup – where Chibikart captures the steering knuckle in between the pairs of ‘winglets’ on each side (these would be A-arms in a vehicle with suspension), this design keeps it entirely off to one side, entirely using the winglets as bearing blocks. In some ways, it’s like upside-down thug MacPherson struts.

The downside to this is that it needs to be built extra solid, but it keeps the design relatively simple. I won’t have to fashion some weird angled or offset member to accommodate the steering linkage.

Time to add a seat. I picked a leftover (from the summer silly go-kart camp) Razor Ground Force go-kart seat as my choice. For budgetary reasons, I couldn’t splurge on a real ‘human press-fit’ go-kart seat. I modeled this up with best visual approximations and a tape measure.

Here’s a size comparison with Chibikart. The driving posture will be nearly the same, but everything is just a little bigger. The seat is shifted forward in this picture – I wanted to get a sense of seating position, and there were several spots. This one was in fact too far forward and I wouldn’t really be able to get in and out.

If I said I didn’t want to make weird angled brackets, well, that clearly all went out the window the moment I had to reconcile the height of the body with the height of everything else. This was my first brute force stab at the problem, and it became the solution for a while. Because the frame is ‘underslung’ – axles, bearings, batteries all on top of it, I had to either bend something up or angle it to interface with the bodywork.

With the frame more settled, I turned back to making the steering linkage. The only way the steering could work for this design, because of the battery taking up the entire middle, was “linkage forward”. This is the reverse of traditional go-kart steering setups, for good reason: to get an approximation of the Ackerman geometry, it’s way easier to design with the linkages behind the lateral axis joining the two kingpins. Because otherwise, hokeyness has to happen:

Yes, those are crossed linkages: left wheel on the right side of the “Pitman arm” steering link, and vise versa. Everyone who’s tried a linkage-forward design in 2.00Gokart has gotten it wrong, chiefly because I had not forced them to simulate their linkges in CAD before making them (That’s changed this year, so none of y’all get to fuck up). If you keep the linkages left to left and right to right like in the linkage-behind design, you get reverse Ackerman geometry for most of your steering travel.

At the generally low speeds these things run at, it causes excessive wheel scrub and unpredictable changing between understeer and too-much-steer: Oversteer isn’t the right word here since it’s the outside wheel suddenly getting traction and whipping you into the turn, not the rear end flying out sideways.

Moving to the back, I cooked up this motor mounting solution which actually involves reusing the clamping ring from the disc guard of the 9″ angle grinder. The guard is made from some hefty, basically 12-13 gauge steel, and the clamp already fits on the nose of the gearbox, so why not use it? The plan is to cut it off and reweld it to a bent bracket to adapt it to the frame.

To stuff the inrunner into the gearbox, I plan on machining a custom shaft that is 12mm in diameter, steps down to 10mm (with a keyway to be cut into both pinion gear and shaft), and has an 8mm hole with a split-clamp on the other end to fit on the motor shaft. It will be tightened down with an extra heavy shaft collar acting on the clamping region.

Short of machining my own tapered locking bushing into this area, I decided this was about the most secure way to interface to the motor’s otherwise nearly smooth shaft. It has a D-flat in it, but hell if you’re getting me to trust a set screw mount at nearly 30,000 RPMs.

I spent some time putting some thought into how I’m going to mount the batteries. I was going to arrange them in ways the manufacturer never intended. I did plan on reusing the tabs they came with – why not do so? The cell module shells snap into each other, so it was going to be easy keeping them in place. I’d just need to figure out the cell orientation and make a thing to mold around them.

I went through a few iterations of cell layout before settling on this one. Basically, the cell ‘sticks’ can be turned around inside the shells, so I could make several modules that had the same polarity (there were two different mirror-symmetric shell designs) and then bus them all together in parallel. Then, those meta-modules would be bussed together in series. That’s how this big bracket was developed.

It carries 15 modules, each with two cell sticks in them. Five cell sticks in the same row get paralleled together, then those series-feed into the next five cell sticks, and so on. At the end, the connection switches rows (which I ‘ll need to design a custom bus-plate for) and winds back on itself. It’s basically a more epic 6-cell “3×2 brick” battery of the olden Nicad days.

These 15 modules will end up providing 28.8v at 40Ah, or a cool 1.1kkWh, the biggest battery I’ve put on anything so far that could still perform to nameplate ratings, conceivably. LOLrioKart used to run 48v 25Ah nominal Nicads, but the cells were so trashed there was no hope of them ever being 25Ah.

Getting close to something that looks like a silly go-kart. I salvaged a road bike handlebar and quill, which has been crudely modeled here, but might not keep it – it’s positively enormous in real life. I might trim it to a more useful length.


Here’s how the frame looks inside the bodywork. This was the frame design as it stood for about 2 months. I didn’t like it even before finishing it. It was too squiggly – too many angled cuts and welds to get right, and a lot of extraneous material. The battery pack was going to weigh 60 pounds and the frame another 60 – plus a conservative estimate of 10 pounds of bodywork, and I was looking at a 130+lb empty vehicle. Yeowch – optimizing in the wrong direction is coming back from grocery shopping for a whole fuckton of humble pie.

Steel. It’s a terrible thing.

Some time last month, while the 2.00gokart students were still in the early design stages, I sat down and completely refactored the frame into something which made a little more sense. This time, the frame is inverted relative to the axle and battery pack. The battery is now the lowest point on the vehicle, and it most likely will still not clear the damned Maker Faire cable raceways.

Inverting the layout in the vertical direction relieved the complexity of mating to the bodywork greatly, allowing the frame to be made totally straight and therefore easy.

I re-imported many of the parts from the first design to use in the second. Things which remained intact include the steering parts and all the rear axle components, as well as the seat mounting tube.

Around this time, I also firmly decided to make the bodywork from a foam-fiberglass composite sandwich. I’m basically going to turn into the Burt Rutan of silly go-karts for this. After all, I didn’t buy a copy of this for nothing. I decided that a thin sheet-over-skeleton body was going to be too fragile.

To this end, I used the solid model as a reference and made ‘thickened’ sidewalls which can all be machined from a single slab of foam on the Shopbot. The intention is to do this, then bond the foam together, then apply 2 layers of fiberglass cloth – enough to give it structure, then polish it off like the real thing. It would be painted plain white and then the artwork will come after that.

If someone can find me a thermoformer large enough, Epic Thermoform is still on the table.

With the mechanical parts largely being copy-paste, I moved onto modeling moer electrical system parts. This is the contactor deck from the original Ford battery – it’s missing one contactor that controlled the battery negative, which I took out because in my system I’ll only be switching positive. I’m going to try and use it otherwise stock, just to get more usage from the pack. All hybrid batteries will come with a contactor pack like this or similar, so it can be a valuable resource in its own right.

To make the shape, I again imported a picture of it and began scaling until the size made sense. The 6″ caliper acted as the scale item in this case – between its arms is 6″ +/- about 0.01.

I’ve added more detail to the frame now, including the new front and rear bodywork mounting points, which double as bumpers. The extra front biased volume is going to give me much more legroom. The little holes in the bumpers are so I can attach the bodywork with quick-release pins.

The frame next to the bodywork model. The folded sheet metal brackets will be bonded to the body and travel with it – they’re only shown in place here for visibility purposes.

I mulled over this design for a few days to make sure I still liked it. It was time to start cutting steel.

I got a great deal on eBay for mild steel tubing – nine 4 footer sticks of 1″, .065 wall for only $35 total (shipping that much steel was a whole ‘nother tax break expiration). That’s not far above scrap pricing. Someone’s obviously trying to get rid of a lot of this stuff, and it might have been actually surplus or scrap – it was the dirtiest, most greasy steel I’ve ever had the joy to touch. An entire can of brake cleaner went into cleaning just the five rails I cut up.

Here is one, mounted on the coldsaw, about to be sectioned into frame pieces.

The resultant frame cuts! Not only did I process all the square tube, but the round stuff and hub parts too.

Here’s a dummy frame mocked up. The whole thing is 47″ from front to back – the bodywork brings it closer to 5 feet.

I had the joining plates waterjet-cut from 1/8″ steel. These include seat mounting flanges and reinforcement plates for the front steering knuckle area, as well as the flat plate profile for the motor mount. Overall, this vehicle features very little waterjetting, and it could all be bypassed if I were working to stricter budgeting requirements, but that would be taking Powerwheels racing too seriously.

The small tube chunks are all parts of my BurnoutChibi-derived hub solutions. In fact, BurnoutChibi was a prototype for this design, especially the fronts.

I decided to fabricate instead of buy drive hubs because 1. nobody made drive hubs for shitty Harbor Freight Pink Wheels, and 2. commercial cheap go-kart hubs are just slabs of steel welded to tube anyway. The bore of the tube is almost exactly .75″ – it even fits over the driveshaft I bought. It will straight up be broached for a 3/16″ standard keyway.

My first instinct is to all-aluminum-billet this, like old LOLrioKart hubs, but it’s being kept cheap and steel to more closely follow the spirit of the event.

The front hubs are an exact dimension change of BurnoutChibi hubs. I bought the steel tube such that its inner diameter was almost exactly 1 3/8″, necessitating minimal reboring work to stuff bearings into.

And check out these custom, almost 7″ disc brakes! Is this serious braking overkill or what? I think my actual van brakes are not much bigger.

I can make up for everyone elses’ shitty dysfunctional scooter brakes at the race.

To utilize these hubs, I had to carve off the existing bearing hubs from the Harbor Freight wheels. I chucked this into hugelathe and ran a standard boring bar into it, boring away at the face of the wheel until it flew off.

Using hugelathe, I also finish machined the hubs after welding them together. BurnoutChibi’s hubs did wobble a small amount due to welding warpage. These hub plates were all bigger, so the warp was amplified. Post-machining was also needed to make the boss that the rim halves sit on, and to clean up the interior of the tubing for bearings.

The internal bearing spacer dropped into place. I whipped up a “spacing donut” for the Makerbot to chew on while I did the welding and machining. This spacing donut is 95% air and only serves to keep that spacer aligned roughly with the bearing bore so I don’t have to play “chase the spacer” every time I have to pull a wheel.

Here’s a finished set of front wheels. I’m waiting for an order containing broaches to finish the rear drive hubs.

Meanwhile, I’ve been drilling the necessary holes into the frame tubing. I hope to be able to weld most of it together this weekend, as well as test-make a body panel on the Shopbot.