The intervening week between the Dragon Con 2014 event report and the New York Maker Faire saw a torrent of Chibi-Mikuvan repair and overhaul work. I’m glad to say that Chibi-Mikuvan, despite not winning anything outright, did last to the very end of the endurance race, and the performance at the last sanctioned PPPRS race of this Maker Faire season means I have plenty of lessons learned to go through, and deltas for next year to start working on. Here’s the story the last week and a half, in pictorial form!

the finishing

I began by doing some body repair, since that would need time to cure and I could do things in parallel with it. The front body mounts all broke off in Detroit, necessitating drilling zip tie holes (shudder) through the bodywork to hold it on. I was going to rebond the body mounts and fill those holes. By the rear body mounts, there were also portions of the fiberglass which were caved in from handling and side impacts, exposing some pink foam.

All of these areas were remedied with some glass microsphere and epoxy paste. I whipped up a concoction that was more or less the consistency of peanut butter, so it would stay in place, and used it to fill the areas with exposed foam as well as to rebond the body mounts. The whole thing was clamped sideways overnight to cure.

This. This is the last clean shot of the Detroit Breadboard of Shame before I rip everything out and just start over. Check out the burned ground trace on the bottom left. Something tried to ground itself through the thin trace on the Arduino, blowing it up and pretty much killing everything on the board due to the resultant voltage transient.

I wasn’t kidding – the way I thought would be cleanest was just starting over on the whole thing, since all of the signal components were declared unfit for flight after the breadboard smoked. If building 19 motor controllers has taught me one thing, it’s that some times, parts aren’t worth salvaging if they might give you headaches in the future, and debugging is best if you can concentrate on 1 problem at a time and not worry about cross-coupling causes from malfunctioning and half-assed parts.

The wooden board that mounted the water cooling equipment was also cracked in half, so that had to be replaced, and I also wanted to replace the PVC tubing going to the motor. So off everything came!

Chibi-Mikuvan’s original wiring, contrary to my usual practice, is completely undocumented. Things were wired as I needed, as seen by the 3 or 4 identical looking black cables funneling into the breadboard looking like a New York City tunnel entrance. The new design will consolidate all of the external wiring into 1 cable, and all the internal wiring will be spread out into terminal strips, which I spent some time allotting for wiring positions that made more sense.

The “Cable” in question is an Ethernet cable. I’m fond of using repurposed PC cables for signal running duty, as they’re often twisted, shielded, and several feet long already. An ethernet cable has 8 pins for me to do whatever with. To use the cable without having to cut the ends, I bought some breakout boards for the 8P8C connectors.

One architecture change from the Detroit version of the electronics is the addition of a small backup battery for the 12v subsystem. In the first version, I had a hacked BEC DC-to-DC unit running the logic board, and an independent one running the water pump for the cooling loop.

The important one failed (The water pump was working at the end of Detroit!) and destroyed the logic board. This time, I’m only going to have 1 BEC, but it will continually feed the backup battery. It’s made of a row of A123 26650 cells, worth 2.2Ah, so the logic board will ideally keep running from this for two hours, based on the ~ 1 amp draw from the logic board, contactor, and water pump combined.

I’m also hoping for the odd side effect of the backup battery destroying the DC/DC unit completely (blowing it up and rendering it open circuit) should the unit fail for some reason!

 

To disconnect the backup battery, I added an independent logic power switch. Strictly speaking, this is just a battery switch – hitting the main switch will power the DC/DC unit, which is connected directly to the logic, regardless of the battery presence.

Getting power to the fans was another challenge of the new logic board. The fans on the Trackstar ESC are 5 volts, and I wanted to keep it as stock as possible for fast swapping in the field. So I now needed a 5v power supply capable of pushing about 1/2 amp.  Attempt #1 harks back to my analog electronics days: An inline-4 made of 7805s, stuck on a used Jasontroller heat sink.

Sadly, while it was in good faith, the continuous power dissipation made it get too hot. Half an amp dropping from 12v to 5v is still too much!

So the next day, I was wandering at Microcenter in the DIY section to see if there was any product that was readily available to downconvert 12v to 5v, thinking to myself “what small consumer gadget does this?”, I suddenly remembered the gift to humanity that is shitty, cheap in-car USB chargers:

They contain a small 12v to 5v switching buck converter, and are pretty compact. Problem solved for $3!

Here’s the board, mid-completion, with the chopped down car USB charger wired up. The reason why I didn’t just use another Hobbyking BEC unit is because I only had those large 3A switching types – the LOGIC blob is how large they are, and I didn’t have any more space on the board for another.

 

Bench testing the new logic board. The circuitry is pretty much the exact same, so there were no surprises here.

And the new logic board installed and well-potted with hot glue. The entire underside is also coated. This whole assembly is already so much cleaner than the last!

Finishing touches and running the new cable up to the steering column. There’s now no more need for a big wiring loom containing 2 cables and a few loose wires.

I made the new cooling components mounting plate from two 1/4″ thicknesses of birch plywood epoxied together – reason being I don’t have a laser with long enough focus distance to get through half an inch cleanly.  Afterwards, it was just final assembly!

While generating a replacement catface using fridge magnet stock on the laser cutter, I decided to have a little fun with the remaining material I had left.

Unfortunately, as funny as it would be to have this as a race livery, there’s too little magnetic area to have it stay on well.

So I gave it to the laser cutter itself. Now, the hapless freshmen who use this machine will have to overcome its disapproval of their newbie handling procedures.

the event

 

Your obligatory vans-on-vans photo here.

Since New York Maker Faire is only about 3.5 hours away, we’ve historically loaded out around 4:30AM and arrived just in time to be stared at very disappointingly by the NYMF traffic-directing staff since you guys were supposed to have finished unloading 30 minutes ago, but we’ll let you in anyway because we’re nice.

In the background is Chibi Atomic Jeep Thing (I guess the strikethrough doesn’t need to be there…). Having tossed the hastily-made Jeep body in Detroit, Dane and company were left with a curious problem of not having a shell. In the last six hours or so before we left, a VW Thing was created out of leftover plywood. It was actually quite accurate, since the Thing was made of flat panels anyway.

We got there exactly on schedule – that is, at like 9:15 when they were hustling everyone with loading vehicles out, but once again, they let us pull right up to the MITERS table anyway! Chibi-Mikuvan was hastily dropped off and MITERS gear pushed out the door. At this point, only a few teams were around.

MITERS was right next to PRS, and even as they were still unloading, they started getting interested audience members.

Clean shot for the weekend! This is the last time it will look this nice.

Qualifying rounds began at 11am. The fastest timed lap gets pole position in classic qualifier fashion.

Very quickly, qualifying devolved into a 3-way battle between myself, Phantom Power (behind Chibi-Mikuvan in this photo) and the NIMBY Ferrari. All 3 of us were consistently fast. It took me a few rounds, but I gradually understood the handling envelope of Chibi-Mikuvan enough that I was able to shave off time in the corners, leading finally to a 23.78 second lap time which the other two continued all the way to the end of qualifying to beat.

And that’s how Chibi-Mikuvan got the quickest qualifying time.

I don’t think I ran that hard again for the rest of the races, since at that point, the track was fresh and dry (it would be wet on Sunday) and the tires and steering were still in good shape – they’d get more trashed as the weekend progressed.

When I go, I go hard.

Here’s Chibi-Mikuvan demonstrating some oppo.

Moxie this time was an unlimited-power drifting session to the Hatsune Miku cover of Let It Go. You didn’t know that existed, did you now? The New York audience seemed much more receptive over the Detroit one, who I assume is uncomfortable with topics other than union relations and economic depression.

The new tires I picked up before this event – Kenda K473s, performed admirably. They handle very linearly, and have a highly predictable traction envelope.  I knew exactly how much power to put into a turn if I wanted to push the rear end loose. Even better – after this whole round and like 10 qualifiying lap attempts, they barely looked worn! Go figure, real tires.

Over the rest of the races, I also got a better idea of seating position if I wanted the vehicle to handle more loose (oversteer prone) or … opposite of loose (which isn’t tight, but describes understeer). Since these vehicles are all so small relative to the rider, shifting in your seat is a huge influence in handling.

We brought back the MITERS all-vehicles parade for the Chibi-Thing Moxie round. This time, I slowly followed the parade while playing the Hatsune Miku cover of the Disney Electrical Parade theme song. You didn’t know THAT existed, did you now!?

Dane hovering on the Flying Nimbus, which I’m sure you’ll hear about soon.

By the start of the first heat race, the Trackstar was already shedding fan blades. I think it has a lot to do with the fact that I cut out the circular fan grill to increase the air flow. This could have exposed it to much more street debris. These little fans are fragile, as years of robot combat has taught me.

I swapped them out before heat #1 with a set from another Trackstar ESC.

This is it! It’s the first heat of the weekend. 25 laps, and I was in pole position.

Chibi-Mikuvan’s delayed launch due to the sensorless startup meant I fell back to 3rd place immediately, as Phantom (position 2, to my right in this photo) and Hack to the Future (behind Phantom) are either DC or sensored AC, so they have much faster starting response.

I caught up immediately and began chasing Phantom for a few laps.

Attempting an inside pass on Phantom on turn #2…

That didn’t end well. I got mashed between Phantom and the curb, which did some body damage to the right underside, as well as sheared off a tire valve stem. I in fact continue driving for a lap without knowing the tire deflated, and it was the frantic waving of the pit team that got me back in!

Swapping tires…

The dynamics and thought processes involved in racing are way different than Battlebots, and I’m still getting used it and learning from people I know who are more into auto racing. Jim and the organizer crew kept goading me with “This isn’t Battlebots!” during the event, which is entirely true! In racing, often it’s advantageous to hang back and wait for an opportunity to pass.

I did no such thing – I was highly aggressive the whole time, trying to overtake Phantom at every conceivable opportunity, and it ended up costing the race as a result… since in Battlebots, you stay on your opponent constantly and don’t back off unless it’s to regroup after a good smash, for example. Now, it could be argued that if the car was built better, it could have stood that curbing action and continued…

Some damage to the front end here. This was the mild side, consisting mostly of tire scrub.

…and this is the side that got mashed against the curb. That’ll require some patching…

Sadly, that wasn’t the only pit-in during this short race. A few laps after the tire swap, what I thought was the Trackstar dying on a fuse-out event was actually caused by one of the six motor wires of the Aquastar motor popping out. Even though the whole bundle had been cable tied, it seems like the impact with Phantom and the curb might have partially loosened the wire, which was then free to flop out after some more vibrations, and the resulting burst of current from the motor losing sync with the controller caused the fuse-out.

I really dislike the ‘split tube’ style of bullet connector, as most of these larger-than-4mm connectors are, because they lose their springiness and start fracturing at the base of the fingers after only a few insertions. For instance, this is a 4mm bullet connector with the ‘circular spring’ style that I prefer. I don’t buy the argument that they’re higher resistance because of the spring contact over the metal-on-metal contact – at least to me, the extra milliohm is preferable to less connector durability.

I also found out that the voltage transient from the fuse-out also killed the DC/DC converter, which was just heating up when plugged in and no longer regulating. In the interest of not running any potentially dead parts, I took it out. Remember, the 12v backup battery meant that the logic and water pump could still run! I had a backup one, but it needed to be hacked and wired to be compatible with the board – I left this to be done by one of the MITERS crew members while I continued.

For the remainder of the race, the motor wire bundle was just…. very well zip tied down, and exhibited no more problems.

Here, the reinforced motor connectors can be seen (post-race photo):

Although Chibi-Mikuvan ran beautifully in the last 10 or so laps of the race, the two pit-ins caused me to come in, well, basically dead last. In a race this short, if you have to visit the pits at all, you’re toast.

If I’m not working on one van, I’m working on another…

Here I am straightening out one side of the steering linkage that also got sympathetically bent up with the Phantom collision. After the race, I noticed I had a significant amount of toe-out, and it was caused by one of the follower links (“rocker arms”, more commonly) twisting. Gradually, over the course of the other races, all of the linkages began bending and twisting slightly because they were single-supported.

At the end of the race was when I discovered that one of the fuse holder’s fuse-mounting screwed was stripped out. As it turns out, a very excited new team member decided it would be a good idea to save time and use the cordless DeWalt drill with a screwdriver bit to untighten and retighten the fuse screws….. with the clutch setting on “Drill” (locked)!

As a result, the fuse holder’s brass threads were stripped completely, and there was rather little torque that could be put on the fuse’s mounting tabs. We figured this was okay, because of the amount of contact area available…

The 30 lap race went “better”, but I face a fuse-out problem almost right away as the flag dropped. The fuseholder hack obviously wasn’t tight enough. My guess is the relatively poor contact made for much faster heating of the area, which would make the fuse blow much easier. Above, a member of the Hack to the Future team helps pull Chibi-Mikuvan off the track.

We spent about five minutes back in the pits trying to secure the fuse better. Here’s the fuseholder I was using.

The stripped block was rotated on its side 90 degrees so the wire-pinch set screw was facing up, sticking out about 1/8″, and a 1/4″-28 nut partially threaded on (M6 is roughly 1/4″-28 for a few threads…) to retain the fuse tab. One of the MITERS members who was ‘on rotation’ for this race ended up forcefully driving a wood screw through the stripped out M5 fuse retaining screw hole to pinch the wire. The whole thing was then zip tied over several times to keep it in place. Changing fuses would have involved cutting a zip tie, but whatever.

So why didn’t I just replace the fuseholder? Well, I have a spare one. It was even in the same box of parts as everything else. But I didn’t pack it in the race-specific box of parts, because “Well, how the fuck can we mess up the fuse holder?!”

After the fuse surgery, Chibi-Mikuvan then proceeded to run 18 laps in the remaining minutes of the race  with absolutely no flaws. Here’s some “Luigi Death Stare” action going on with FUBAR Labs.

 

Passing some slower cars around the outside…

 

Going up on 3 wheels was very common. Going up on two wheels also occurred a fair amount.

With the two heat races complete, Chibi-Mikuvan wasn’t doing too well in standings, but remained fully functional at the end of heat #2. It seemed like something would break and then be repaired, and it would show no flaws for the entire rest of the race and go back to chasing the leaders…

…but only after it stopped mattering. The announcers and I agreed tongue-in-cheek to only report me as being in 8th place, no matter the actual position, so CMV wouldn’t know whether or not we were winning.

In preparation for the endurance race on Sunday, I only charged both batteries (plus the 12v backup battery), and changed the rear tires, but not before some wet track induced drifting:

Jeremy taking Chibi-Mikuvan for some drifting action.

 

I was just a little too excited here.

 

The starting lineup for the 75-minute Endurance race…

 

Well, that was certainly embarrassing.

No, nothing bad happened, but I powered on in the wrong order. With the DC/DC converter, just flipping the master switch would have been enough – it would power the logic immediately. But I forgot that we took the broken DC/DC unit out, so I had to turn on the logic switch separately, which I therefore forgot to do.

If the Trackstar ESC wakes up and doesn’t see a signal, it will beep at me for a few seconds before realizing there is a signal. So I had to turn it back off, then turn the logic on, then power on again.

Oops. Most everybody got the jump on me here, but lost ground was made up quickly.

After the first driver switch (15 minutes per stint – all cars must roll through pit lane, even if  you don’t switch drivers or batteries), here’s Chibi-Mikuvan alt driver Mike holding off Hack to the Future. At this point, Chibi-Mikuvan was somewhere around 2nd place for lap count, just behind Phantom and ahead of HttF.

 

That didn’t last too much longer. Near the end of the 2nd stint, the steering linkage finally let go on a straightaway – check out the combined toe-caster-camber-whatever!

One of the teams brought along a flux-core wire welder. It clearly isn’t the best welding technology, but it’ll have to do! What happened was the steering’s driving link (“Pitman arm”) sheared its thin tube welds that I made months ago. I’m guessing the numerous shocks and bumps and aggressive handlebar ripping finally did it in. There were 4 welds to the tube – I didn’t weld fully around the circle, which is clearly the right thing to do.

 

A post-race photo of the weld made. This is a classic gorilla weld. Huge, ugly, but it held together…

Here’s a short video combining the footage recorded using my camera from pitside!

Just schmoozing along nearing the end of the endurance race.

After the third stint, Chibi-Mikuvan ran out of batteries. Like, literally.

The hybrid batteries can’t charge fast enough to run for 15 minutes with only 15 minutes of charging. I only brought two batteries, which was a bit of a mistake – I should have assembled a third (or a fourth….) , but time was always of the essence.

At first, I mistook it for an overtemperature or undervoltage protection on the ESC, which I thought I had disabled, but nope – a quick check with a voltmeter showed the battery sagging under 26 volts just running the drivetrain no-load!  Nearing the end, we were just coming into pit after not too many laps – as low as 10 or so – to swap the battery and use whatever surface charge had stuck to the other.

However, despite the last-minute energy problems, Chibi-Mikuvan still pulled off a fifth place lap count, and also won the most moxie points awarded during the Endurance race, and came away at the end with a 3rd place overall in Moxie points. I was informed afterwards that during the Endurance race, there were two children just following the Moxie vote board around, constantly voting for Chibi-Mikuvan. Niche appeal saves the day!

 

Here’s something different now. I’m giving a quick driver orientation to none other than Limor, or the Ladyada of Adafruit Industries.

 

Limor piloting CMV a few laps after all the races were over!

 

Finally, the “team photo” of this season. Most of this crew came along for the MITERS booth and to attend the Faire in general, but always stuck around to help out with PRS efforts when needed.

summary

Coming away with two medals – one for the 3rd place Moxie overall ranking, and one for 1st place Moxie during the Endurance – and retaining functionality at the end, Chibi-Mikuvan had a pretty reasonably successful first season, all things considered. I think the issues it experienced can be summarized into two categories – design related, and operations related.

The design related problems include the “little detail” things I seem to overlook habitually, such as constraining wires properly (“Psssh, I can build Battlebots fine, how hard can this be?”).I’d also include the poorly welded steering driving link under this heading, since I really should have welded all-around after putting the tacks down. The issue of the steering linkage being rather pliable is a more in-depth design issue that would require some rework before next season.

Ops related problems were those caused by us during the race. The overtorqued fuse holder, for instance – a “more serious” team with clear roles would solve that (but where’s the fun in that?!). The fact that I didn’t bring the spare fuseholder. The fact of the hybrid battery pack not being designed for fast-charging, and more packs not being made beforehand.  And the drivers being excessively aggressive and, in general, not experienced racers, me especially. Mike? He’s okay, since he did go-kart things as a kid, or something. I just rage through Boston rush hours.

One of my favorite moments relating to the latter was chatting with some of the other teams after the race, and one of them remarking “You guys have the SCARIEST drivers”. Without pause, Dane fires back “We’re from Boston.” Case closed.

Some helpful advice that Jeremy put on the front of Chibi-Mikuvan, before the endurance race. If I had doubts, I could just look down…

idea emitting

I spent some time recently thinking about how I’d change up the design for the next season, and here’s some of the key issues, including the previously discussed design problems, that I think I’m going to focus on.

 

•  The steering linkage is transitioning to all double-supported linkages, especially the kingpin and upright. The uprights right now are made of 2″ box tubing, and they started going all parallelogram as the number of impacts with tree roots, curbs, and potholes increased. After each race (and twice during the Endurance) I’d take the big Channel-lock pliers in the toolbox and bend them back into place.
• I will bring a spare “part I don’t think I need” because I will probably need it.
• The “right angle ball joint linkages” from McMaster-Carr are not exactly vehicle duty. Seems like they’re built for industrial machines, only moving every once in a while and without much shock loads. By the end, they had plenty of slop, which added to the steering imprecision. I’d like to transition these to real rod ends, which would also work well with the double-supported linkages, can be slop-adjusted and lubricated, and should handle shock loading much better.
• I will bring a spare “part I don’t think I need” because I will probably need it.
• I’d also like to greatly reduce the steering ratio. In combination with the handlebars being really wide (see below), you had to pull or push quite the distance to effect a sharp turn. Right now, the driving link to follower link ratio is about 1.5:1 – I’d like to drop it to close to 1:1 so you don’t have to move as much.
• I will bring a spare “part I don’t think I need” because I will probably need it.
• I’m considering adding some proper “caster angle” to the steering as long as it will be entirely redone. I still remembered how NIMBY-Ferrari handled in such a linear and predictable fashion. The addition of caster will help unload the rear inside wheel on a turn, which will aid turning at lower speeds.However, the effectiveness of such a change is something I and some of the other builders are considering.
  In the races, I was clearly already traction limited in the rear – I was basically throwing the rear end out a little bit on every single turn during the heats and some during the endurance race too. To unload one wheel and load the other even further seems like a recipe for severe oversteer to me. Combined with the fact that throwing your own weight around modifies the vehicle’s center of gravity significantly, and I see this as a secondary – nice thing to have – change that is pursuant to tightening up the steering precision in general.
• I will bring a spare “part I don’t think I need” because I will probably need it.
• As another corollary to the rear traction being more limited, I’m considering moving to wider 9×3.50-4 type tires in the back. They’re only half an inch larger in diameter, really, than the 2.8/2.50-4 handtruck wheels, and it wouldn’t be really noticeable from an appearance perspective.
• I will bring a spare “part I don’t think I need” because I will probably need it.
• The handlebars are set rather widely – they’re stock road bike handles – and you basically have to elbow yourself in the stomach when turning.On the left side, where the brake handle is, it’s tolerable since you can move your elbow and forearm. But on the right where the throttle is, you can’t, since else you’d have to drop the throttle. My right turns were more seriously limited in speed because of this. I’d change it to either a much narrower U-shaped handlebar (so the arm motion is reduced) or a regular T shaped straight handlebar.
• I will bring a spare “part I don’t think I need” because I will probably need it.
• Speaking of the throttle, as much as I love the thumb throttle, I’d rather have a “half handle” style twist throttle. The reason is the thumb throttle actually requires quite a lot of travel, and it was seriously fatiguing on my right thumb to drive 15-20 minutes at a time. A half-twist throttle would work great with a flat T style handlebar, but not as well with the U-shape one since it would be facing a non-intuitive direction.
• I will bring a spare “part I don’t think I need” because I will probably need it.
• Chibi-Mikuvan has a speed edge of a few mph over most cars, but I’d like to trade it off for some more ‘kick’. I bought a 10-tooth sprocket to up the gear ratio from 20.5:1 to 24.6:1, but it was…. well, it looked kinda like this. It had almost no meat on it at all, and so I elected to play it safe and not risk literally shearing the whole thing off! An 11 tooth sprocket is a bit more tolerable looking, and I’d get a roughly 22:1 ratio for a net top speed of 24mph instead of about 27. Having this option is nice – obviously, depending on the size of the track, it could be switched out.
• I will bring a spare “part I don’t think I need” because I will probably need it.
• The size of the front disk brakes might actually need some reducing. They’re enormous, and lock up very easily because of it – the joke is that Chibi-Mikuvan can stop itself and the 3 karts behind it. There is not much braking linearity between engaging and locking up, and I found myself performing some “manual ABS” brake pumping during the endurance race when the track was wet in some spots and I was coming in way too hot towards a crowd of cars taking it slow. Because of the amount of travel in the thumb throttle, I didn’t utilize regenerative braking as much as I would have liked – though Mike was all over that.
• I will bring a spare “part I don’t think I need” because I will probably need it.
  I will bring a spare “part I don’t think I need” because I will probably need it.
  I will bring a spare “part I don’t think I need” because I will probably need it.
  I will bring a spare “part I don’t think I need” because I will probably need it.
  I will bring a spare “part I don’t think I need” because I will probably need it.

Other ideas include changing the gearbox design and exploring some of my more wild power transmission concepts, but I may let the angle grinder  and boat motor run another season, because why change it when it’s not yet proven to work 100%? Maybe when CMV finishes a race, problem-free, in the top three.

For now, CMV works, and I’ll likely be marshalling the MIT Mini Maker Faire on it this coming Saturday. Be prepared for the story of that!