Archive for December, 2013


The Equals Zero Christmas Special: eNanoHerpyBike; or How to Hack Your Cheap 5V BEC to be a 12V DC/DC unit

Dec 30, 2013 in Beyond Unboxing, eNanoHerpyBike, Project Build Reports

Hey everyone! It’s time for another “Did I say later… I mean like now“! Recall this from a few weeks ago:

Going above, beyond, and way further on the highway to collect dumb shit than ever before?

I decided that it was so simple and required so little fabrication that I might as well do it in the post-semester downtime when the students all go home continue to use the shop for their own derpy projects, but luckily they don’t involve my oversight as much. It also presented me with an opportunity to finally use one of the One Thousand Hobbyking Amps ESCs, and I got to try the “mechanical waterjet” on aluminum to test out how well it was able to perform; even though it’s not really meant for that.

So here’s what it is.

Here it is on the dissection table in preparation for removing the existing powertrain.

What it comes with:

  • 12v lead acid battery
  • “80W” generic scooter motor – it metered 0.18 ohms, which means this motor could actually push quite some power. I couldn’t find these on sale anywhere, so they might have been a one-off made for this vehicle by Razor by Unite Motor (which makes OEM motors for Razor)
  • On-or-off controller that was really your finger switch pulsing a relay. Finger-PWM!
  • Roughly 8.5:1 reduction, belt drive. I found this impressive since it means I could easily make use of it for sensorless drive, to which a high gear ratio is crucial.

What it’ll get:

  • 25.9v (7s) lithium battery, from my Nuclear Arsenal
  • Hobbyking “200a” controller. I decided to start with the “Birdie” model.
  • A left over T600-880 motor. It was way too fast for the application, but is the closest hardware direct swap!

The big plastic belt cover comes off to reveal the driveline mechanics. They put quite a lot of engineering effort into these, so I’m sad that these things are no longer being sold. An adjustable sliding motor mount and a real spring-loaded belt tensioner, commanding a quite unconventional 14mm wide 3mm HTD belt (size 549-3M-14). Typical 3mm belts are only 12mm wide. I was only about to find the belts listed specifically for this vehicle, so again it might be a dedicated thing for Razor.

I designed a quick mounting plate that adapts the T600′s 25mm bolt pattern to the 56mm pattern of the original motor:

I was going to lob it at someone in MITERS to sneak onto one of the waterjet cutters, but somehow this felt immoral. One thing I had been itching to try was cutting metal on our Shopbot PRS Alpha. It’s something which has been done in limited amounts and generally very slowly, often on thin material. The Shopbots aren’t metalcutting machines by design – their construction is lightweight aluminum extrusions and little wheel guides on steel rails. This doesn’t lend itself to maximum rigidity or vibration damping.

Here’s where I found out that our machine has a bit of head backlash- look at the oval flattening of the holes. I guess I’ll have to dial it out now.

I drilled and screwed a spare 1/8″ plate onto our cutting bed and sprinkled the area in cutting fluid to keep the cut cool. Another reason why I’m not about to buy aluminum in 4 x 8 foot plates is the lack of a coolant system and chip extraction system – the vacuum next to it isn’t gonna do that for you.  If this becomes a more common practice, I might try to invest in a standalone mist cooler, which doesn’t make a huge lake of coolant and delivers just enough to keep the bit clean and cut area lubricated (Example)

Using a feed and speed calculator, I came up with a conservative feed rate and spindle speed assuming 0.001″ of cut per tooth – pretty light. Using an 1/8″ carbide endmill, I was going to run it at 30 inches per minute at 10,000 RPM.

It broke pretty much instantly.

I’m going to guess that the machine just can’t absorb that level of force and not vibrate, which is weird, but when your bit is brittle and tiny, even a small amount of vibration will destroy it. I began at 50% of that speed (15 IPM) and gradually put it back up to 20. In the picture of the two parts, the left was done at 20IPM and the right at 15IPM, and the right one exhibits better surface finish.

I suspect dialing out the backlash and tightening the axis rollers more will help it, but it seems that slow and steady with a “one flute” cutter wins – check out the work by this crew.

Anyways, the “is this insane” part of the mission is complete with a satisfactory part, so I’m good for now.

The adapter plate and motor installed!

To make an adapter for the 6mm D-flat shaft of the motor to the 8mm D-flat shaft of the pulley bore, I turned a piece of aluminum tubing to a 1mm wall, cut off an arc with a Dremel tool to make it the rough proper circumferential length, and then mashed the pulley and motor and itself together on an arbor press. The 1mm tubing section cold forged into the shape of a D-flat that bridged motor and pulley, and all was good!

I wouldn’t send it into space, but I would totally send it down the hallway.

The slot is to pass the pre-assembled motor and pulley through, because the pulley flange was too large to fit between the mounting holes, and the arbor press installation was pretty much a one-shot thing.

A better view of the motor installation, with the same M4 screws it came with.

Moving onto electronics:

I whipped up this really quick housing to be made from on-hand high density fiberboard. It served two purposes – first, I hate zip tying electronics together (and explicitly ban it in 2.00gokart!), and I was going to actively fan cool the controller, so it needed a “wind tunnel” of sorts. This thing hangs down on the center frame tube and is secured by some circular sectors with matching holes. The fan is mounted to the left (the front), and the controller sticks out the rear. I designed it to cradle the controller by the output wires.

For signal processing, I brutally butchered a small servo tester – cut the LEDs off, cut the mode button off, and cut the potentiometer off. I only used a soldering iron to clear the potentiometer’s pin holes to add my own 3 pin cable. The potentiometer’s voltage output is to be substituted with a typical small EV throttle (example).

These little testers swing from about 900 to 2100us with the pot’s 0-5v output, so with a typical Hall sensor throttle’s 1-4v output, I’m still in a reasonable range to interface with the R/C controller. This is the quickest way to get set up with an R/C power system – more details can be found, of course, in Scooter Power Systems.

The ESCs I got in the Hobbyking vivisection post are all “Opto”, meaning they do not supply power to the receiver and they take signal in through an optocoupler (hence the name).

Well, first off, there’s no optocoupler – in Hobbyking language, these “Opto” ESCs just have all the parts for an onboard 5v power supply taken out for cost reduction.

Smart. Well, there’s always a very small 5V supply – the microcontroller’s own power regulator. It’s too small to run any servos or bright LEDs, but if you only need a few more mA to run said servo tester, then it’s an easy wire jump from the 5V regulator to the red wire of the servo signal cable! In these applications, the regulator is already very heavily stressed (they’re not supposed to be dropping 20+v on their own), so a small auxiliary micro is about the most that can be gotten from it.

Now, to be fair, it’s not true for all of them – most of the higher value controllers, such as my eternal favorite the 100A Sentilon, have legitimate optocoupled signal inputs and isolated grounds. For high current applications, it can greatly reduce noise glitching.

I bunched the old heat sink back on, in a less precarious location (recall that they ship with the heat sinks touching the output solder blobs), and threw a new heat shrink shell around it.

This leftover 4400mAh  7S Thunder Power battery from the DERPA robot fit perfectly in the 12V lead acid tray. It was so perfect that I just used some Velcro cinching straps (not zip ties – never, ever zip tie your lipos) and it sort of hangs out there, oblivious to all that is about to occur.

I needed a 12v rail to drive the cooling fan, which was not going to tolerate 26+ volts straight from the battery. I decided to check if my collection of cheap HV 5 volt BECs was hackable to yield 12v. The answer is yes, and I’ll keep it as a separate section under Beyond Unboxing, so for now, this is the 12v DC to DC converter!

The electronics box was laser-cut from 1/8″ and 3/16″ high density particleboard. This stuff is actually reasonably strong and water-resistant – not like Ikeaboard. I am still on the hunt for a high strength laser-cuttable plastic that isn’t either terribly messy (ABS, PETG) or expensive (Delrin). Or really soft and plushy (Polyethylene). For now, wood was fast and someone left a plate of it without their name anywhere, so it automatically becomes state property.

Here’s where the controller sits.

The 12v DC-DC module and servo tester have been added. That fan is a 40mm harvested server fan – the ones that scream at 10,000+ RPM to move the same air as a big computer case fan, but it’s a server in a room by itself and flatness is more important than being quiet. I whipped up an adapter block and had it 3D printed while the rest of the wiring happened.

A better shot of the fan, as well as a view of the master switching. The male Deans connector is where I jack in the battery; the vehicle is turned on using the “Georgia Tech Switch” that I now use on almost everything that doesn’t matter. It’s also known as a ‘removable link’.

A shot of the completed vehicle! I used left over wiring twizzlers spiral loom from the Electric Vehicle Team to keep the wire bundles neat.

First impressions: It’s very menacing wheels-up when you gun it, but on the ground is a different story. Remember how I said the motor is “too fast”? Basically, the theoretical max speed for this design is 61mph:

  • 880 RPM/V at 25V (assuming some loss in the system) to yield 22000 RPM
  • Reduced 8.5:1 to yield 2588 RPM at the wheel
  • with an 8″ wheel, that results in a ground speed of 61.5MPH

That’s an impressive number to throw around to the uninitiated, but what it means is that at any speed under half of that – or about 30mph – the motor is dissipating more watts in heat than it is giving you in mechanical watts of ass-haul. So, in other words, this thing just pulled 200+ amps and didn’t do that much. The takeoff was still extremely strong, but at the expense of all the wires and the battery being hot within a minute. The controller is only surviving due to forced air cooling.

Despite all this, I rode it home and back a few times for sheer shits. I might even say I love it more than Melonscooter, since it’s so light and nimble. I just had to have a very, very sensitive throttle finger since if I accidentally gunned it, it’s not going to take off without me, but just light on fire.

This stopped being funny within a few hours. I decided that nobody else was ever going to be able to experience the joy of this thing since it would self-eat so easily. And self-eat it did – for some reason, one day Jamison was taking a spin and it made a popping noise and stopped working.

Whoops. Well, it’s time to “downgrade” the motor while I’m at it:

I spent a while on Hobbykong searching for a replacement motor. I decided to try and get the speed down to about 30mph tops – which I assure you is plenty fast enough. I had to juggle which motors were in stock with how much work I wanted to do to replace the shaft (since almost all of these small outrunners have 6mm shafts), with which ones actually had my required RPM/V.

I settled on this 700-class heli motor with 500 RPM/V. It would yield a “max power speed” of 16mph, so you could actually stand a chance at blitzing down the hallway and be able to kill yourself instantly at the end. I also trade an unrealistic top speed for more useable launching torque.

I wanted to do away with the irritating forged sleeve adapter. I had left over 8mm precision shafting, so I turned this replacement shaft on tinylathe and made retaining ring grooves on the right end to keep the pulley on without a set screw (It came with a snap ring on the original motor). The dimensions are otherwise on-the-fly measured directly from the heli motor.

The new motor installation was easy; it shared the same bolt holes as the T600. I replaced the burnt Birdie with the Red Brick.

This thing now really hauls – I can legitimately hand it to someone and have them throw themselves off without potentially destroying anything. Not to mention that it somehow became 100% practical too! The suspension and pneumatic tires means it’s actually very smooth in handling bumps, and a little exciting in acceleration since it compresses and you are not sure if it will keep compressing until you land on your head. Just don’t try to stop. It has 1 brake, in the rear, and your center of gravity makes it lock up if you are even thinking of stopping.

eNanoHerpyBike (because it’s electric and smaller than Herpybike to the left) is up for some test video soon, whenever it stops being disgusting outside. I do have some hallway footage, but this really needs space.

That’s all build-report wise. I said, super simple and minimal fabrication. I’m truly sad that these things are now getting rare.

Beyond Unboxing: Turning Cheap BECs into 12V DC/DC units

Now a little more about the BEC hack. This falls under the category of “you might find this useful if you already have $part but want to do this thing with it”.

You can buy dedicated 12V DC/DC converters for your contraption, but they’re either fairly expensive for the job ($30-100+ dollars) or are a fixed, narrow voltage input and output for industrial use. Good quality R/C BEC (Battery Eliminator Circuit – originally for pilots who didn’t want the extra weight of a receiver battery) are usually wide-range input switching converters, but they’re tuned for 5V.

Update: For thoroughness, and at reader suggestion, here are some examples of where you might be able to get DC/DC converters:

Current Logic is one place I’ve gone to frequently for commercial/industrial modules.

End update!

But since they cost $4-5, a little bit of legwork can turn them into 12V units which will often put out more than enough current – 3 to 5 amps – to run gaudy lighting or auxiliary systems.

I have a pile of these inexpensive 8-40v things specifically for robots and vehicles, so I decided to tear one apart and see which resistor I need to jump to get the output voltage to change.

Inside almost all of these, it’s just a small switching regulator chip, similar to the LM2576 – the design has been genericized to hell and back. This one is by “XL Semiconductor“.

The circuit is pretty much exactly the application note:

Essentially, the converter doesn’t actually “know” what voltage you want it to output. It only decides if the voltage at its feedback pin, measured through the divider R2 and R2, is higher or lower than an internal reference voltage (usually 1.23V). If it’s higher, it’ll lower its output duty cycle percentage to compensate, and if lower, it’ll raise the duty cycle. This is a brick simple, classic DC/DC buck converter.

The feedback circuit is right here. The two resistors that make up the main feedback network are R4 and R2 (the small resistor horizontally displaced to R3′s left).

In this application, R4 is what the schematic above calls “R1″ (the lower half of the divider), and R2 is… well, R2.

R2 is designated 49B (3.16K) and R4 is 01B (1.00K).  Small SM resistors use some god-awful lookup code instead of a numeric ones – here’s a table of them. Let’s see what voltage this yields:

Vout = 1.23v * (1 + (3160 / 1000)) = 5.11V

The way it selects 5v or 6v is by jumping R3 on the board in parallel with R4, reducing the effective value of the low side of the resistor divider,and causing the regulator to sense an artificially low voltage. So it tries to make up for the deficiency by outputting a higher one. The resistor that gets jacked in by the jumper is a 472, or 4700 ohms (4.7K). This results in a net low side resistance of 1.0K || 4.7K, or 824 ohms.

Vout = 1.23v * (1 + (3160 / 824)) = 5.95V

And that’s how you get 6V.

So if I wanted 12 volts, I can do one of two things:

  • Keep lowering the low-side resistance value (lesser R1 in the example schematic, or lesser R4 on this board)
  • Raise the high-side resistance value (make R2 larger).

To do the former, I would need a R1 (slash R4) of:

R1 = R2 / ((Vout/Vref) – 1) = 3160 / ((12.0 / 1.23) – 1) = 360 ohms

To do the latter, I would need an R2 of:

R2 = 1000 * ((12.0 / 1.23) – 1) = 8756 ohms

Now, most of these datasheets recommend keeping R1 to 1-10K ohms for best stability, so the second option is more palatable. I could use a 9.1K resistor in place of the R2 on the board to get about 12.4 volts.

I didn’t have a 9.1K resistor of any kind. And then, only SMT resistors of the utterly incorrect size and value, a 1206 package (the board uses 0603 package, half the dimensions in every way!).

So I’ll just glob it on sideways. Whatever.

This 10K resistor nets me a extra volt or so:

Whatever ¯\_(ツ)_/¯

What is this, science?!

If I put the 5V-6v jumper into the “6v” position, the voltage becomes about 14.8 volts. In other words, damn perfect for charging a 12V auxiliary battery in constant voltage mode. These things automatically enter constant current if the current load exceeds 5 amps, but they heat up and can be damaged quickly. So, float charge only.

Anyhow, I’m doubtful of the utility of this hack for most hobbyists because it requires SMT surgery. Because the external jumper only adds a different resistor, there’s no Clever Jump It With a Different Resistor hack possible – it’ll need to bypass the internal R to ground to get it done. Adding more resistance to the path will make the voltage differential lower.

To avoid SMT work (trust me, it’s not that bad: sharp, clean tip, and a tweezer), you could solder a regular 1/8 watt small resistor directly to Pin 4 of the chip, the feedback circuit, and just solder blob away the smaller feedback resistors.

Regardless, this is presented in the interest of aiding anyone else who might think of this bad idea in their own quest. I’m certainly ordering a bucket more of these things – I can’t believe I didn’t think of this until now.

The War of Eternal Mikuvan Improvement: Operation DEAD CAT BOUNCE; SHOCKS AND AWE

Dec 21, 2013 in mikuvan

Ongoing van facility upgrades have been continuing since September. Basically, every time I go to Rock Auto to spec out one part:

I needed ONE thing…

This ends up happening. But everything is so cheap ;  _  ;

So I need to find an excuse to use them! Improvements have focused on non-engine work, since I still don’t believe in internal combustion. . This isn’t to say that the engine and transmission have been purposefully neglected, but I’m just not going to add a turbo kit any time soon. Believe me, I did want to “just add a turbo” with no other changes and see how long the engine would last, but apparently you don’t “just add a turbo” – everything else has to accommodate it or it would be a waste, so there goes that.

Mechanical work has been at the forefront – upgrading all sorts of other infrastructure first, such as the suspension, doing bodywork repairs, and maintaining that which would prevent spontaneous grenading of any component. And in continuance of forced and hackneyed tradition, I’ve given all of these major ops a cheesy bad pun name. Don’t worry, I spend only a few seconds thinking up each one, at most.

Operation: Dead Cat Bounce

Ever since I got it running, one characteristic of Mikuvan has been “That Weird Rattle”. It only occurs under high loads at low to moderate RPMs, such as pulling onto the highways or going up a hill without dropping a gear. It was a very hollow metallic rattle, not like something was interfering mechanically, but was always worrisome since it indicated that something could be interfering mechanically all of a sudden. I had a number of theories revolving around the transmission and loose bolts in the crankcase, but careful interviewing of passengers focused the sound to the middle of the right hand side, coming from directly under.

Well, there’s very few things going on in that area – pretty much only the exhaust system. And a rattle under high engine loads probably indicates something loose or jiggling with it.

A tap with a rubber deadblow mallet confirmed that something small and jiggly was indeed trapped in the front of the catalytic converter (or “the cat”). Since I automatically assume that every component on this thing is near, at, or past its end-of-life, I figured that the catalytic element was completely destroyed and has slowly become a pile of ceramic pebbles. But it never sounded like that many pebbles, so maybe a little chunk just fell off or something!

Either way, time to shake them out. Except everything on the exhaust path – and I mean everything – was rusted solid. Half the time, I couldn’t tell what was bolt and what was nut, and the rear cat hanger had become this sort of malformed rust icicle that I think had fasteners on it at one point, but which were now amorphous lumps protruding from a bracket likely salvaged off the Titanic.

Out comes the cutting wheel. I’m going to have to do some fabrication to get the system back together anyway, so I had no remorse in removing these by force.

As I learned after cutting this apart, it was in fact a nut.

After removing the rear spring mounted hanger (why the hell is this specific area on springs when the whole thing is on flexible rubber hangers?), off comes the cat. First, the gasket is utterly disintegrated, and there are erosion channels in both it and the mating flange. So I’ve definitely been having an exhaust leak issue along with it.

This is one more layer of refab I’ll need to take care of. Haven’t I learned by now that any time I touch something, I’ll eventually have to re-mine the iron ore from scratch and derive its engineering from first principles?

Well, the catalytic element is clean. Very clean, in fact – I would have expected even a little graying or blackening, but this thing was still stark white. No missing chunks or holes, so what is going on? Let’s turn it upside down and shake it:

I literally laughed out loud, in real life, in the garage, at like 2 in the morning.

Yep, that’s the shield off an oxygen sensor. The O2 sensor is located at the exhaust manifold, so it probably fell off and has been jiggling in the cat for who knows how long. I haven’t seen any Idiot Lights being thrown, so I can only presume the O2 sensor itself is still working, maybe just a little naked.

What to do now? I had taken the exhaust system apart in the garage with no way to jam it back together. I’m out a gasket, two studs, and need an evening’s work in resurfacing the mating areas. I’d need to buy some exhaust gasket material or sealant beforehand.

Solution: Leave the cat off, and drive with straight pipes.

If it weren’t so very illegal, I’d post a video. It actually didn’t sound bad. For the tiniest sliver of a split second, I wanted a ricer exhaust.

After stopping by Advance Auto Parts (all customers wondered what the hell just pulled in) to gather new exhaust studs and high temperature sealant, it was back to MITERS for the surgery. I first cut the studs off nearly flush with the mating surface, and cleaned off the old gasket with a wire wheel on an angle grinder and then some hand-scraping. Then I slowly drilled through both using incrementally larger drill bits to keep the hole centered and the drill controllable.

Usually, for fixtured parts, I just blast with only 1 intermediate step, or even straight to the final hole size, but I was counting on using a small drill bit to start the hole in the dead center of the stud remnant. But this was not a traditionally fixtured part – it was hanging by its folded sheet metal flange in an extremely beat-up vise in MITERS, on a table that has non-lockable 12″ caster wheels. MITERS has different definitions of fixturing.

New studs installed with a healthy dose of antiseize grease.

I wasn’t able to locate any high temperature gasket sheets, so I resorted to mainlining the sealant and used it in between the two mating faces as an in-place gasket. Then I went totally overboard around the outside. So far, this seems to be holding up, even though it is way up there in terms of bad hack fixes and might not last long.

And that was it for That Weird Rattle.

In one of the “well, it was on clearance” orders, I did pick up a new O2 sensor because (again) I figured the current one was at the end of its life. For $10, I got a Bosch OEM-replacement sensor. It has a way cooler shield:

Back in the confines of the garage at 3 in the morning (when all the middle managers and white collar types who live in this condo complex aren’t awake to nag me for ‘servicing’) I pulled the O2 sensor from the exhaust manifold and replaced it with the new unit. The old one seems to be indeed intact, but missing its shield. Gee, I wonder where it could have went.

I discovered upon removing the exhaust manifold heat shield that some asshat had attempted to weld this thing at some point, and either failed miserably or didn’t know that welding cast iron is a very difficult to get correct kind of thing. You don’t just wave a “splatter stick” at it like this weld. There’s a barely visible hairline crack running parallel to the weld bead, about 2 inches long.

Welp. That’s going to have to wait. It doesn’t seem to be causing any problems, and is covered completely by the heat shield anyway.

New O2 sensor in, with the heatshield removed. This operation required me to resort to vise grips, because the only wrench I had that was big enough  to fit over the hex shank was too big to maneuver into the region. I’m sure there is a special bent wrench that is in the list of special service tools just for this.

Another small, random part has been replaced. One at a time! In the future, the Ship of Theseus Paradox will be known as the Charles’ Van Conundrum.

Shocks and Awe

One of my favorite van activities, occurring at about 50% of all stoplights, has been this sort of thing:

I’m not as pro as him – I can’t maintain it at-speed, only rolling very slowly or at a full stop. With a load of 6 people, the amplitude can get pretty ridiculous.

Overall, what I mean is, the shock absorbers are toast. As much as I love doing “The Thing” (one of many things we call The Thing around here), I’d also like to avoid riding like a battleship – even like a Space Battleship would be okay, but not a normal one – or dying in a rollover.

This time, I explicitly went hunting for “cheap but good” shocks. Through recommendations and haunting the sales of Rock Auto, again, I ended up with these KYB units.

I went for the “Improved Handling” category because I figured “Original Ride Quality” meant “slightly less battleship-like, such as perhaps a trash barge” and that wasn’t appealing. The smaller boxes are sway bar connecting links. While doing the suspension inspection, I noticed the rubber bushings on the “original” ones were pancaked and cracking – the suspension has been getting more and more creaky in colder weather. So this was a “Oooh, it’s on sale!” item which I actually did need.

A few days after, the operation commences on my loading dock of choice. I attacked the rears first because they were much easier to get to and did not involve removing wheels. The rear left was super easy to get to.

The rear right required some Technical Contortion.

The worst thing about it was the upper anchor bolt. I had one click of ratchet movement for this whole thing, and I had to wrap myself around the rear axle to get to that. After the initial tension breaking, I actually used a second wrench driving the T-handle of my awesome right angle handle-drive ratchet thingie to fully remove the nut.

I should be getting brownie points from Harbor Freight, since I think I’ve personally sold 10 of those to people. On the other hand, I’m not sure I want to eat a Harbor Freight brownie.

A removed rear shock absorber unit. This thing was covered entirely in black undercoat grease, and was also very, very dead. Like, dead to the point of needing only around 10-15 pounds of force to push around. I heard air gurgling inside whenever I changed directions. On an automotive force scale, this was probably almost nonexistent.

I wiped off the grease to find a Mitsubishi OEM part number, as well as KYB logo. I’m going to guess these are literally “original equipment”.

Staying true to the brand (which I didn’t know anything about just days before).

And the rears are done. The fronts had to wait for a day when I could roll into the garage to use the lift.

And up we go.

Note that between the rear shock replacement and now, I suffered a rear right tire failure, so the narrow wheel is a “donut” type spare I purchased for all of $25 a while back when I had to dispose of the very rusted-out full-size spare. A full complement of winter tires is on order.

With the front wheels removed, the s way bar links were extremely easy to get to. Here’s the new front right side installed. The old rubber bushings were hardened donuts, and the whole thing was actually loose, having lost tension long ago most likely.

Not pictured here, but while I had the bar loose, I also took apart and inspected their frame-mount bushings (They look basically like that). These weren’t in bad shape, so I cleaned, greased, and reused them.

I dumped some penetrating oil onto the shock absorber anchors and let them marinate a bit while doing the sway bar links.

I cleaned the old mating areas with isopropyl alcohol and some very fine sandpaper to loosen the debris. Underneath all the underbody spray, I can tell more parts of this used to be white.

There’s nothing very exciting about the shock installation, since like the rears, there were exactly 2 bolts involved. I had trouble with removing the lower anchoring bolts, which had nuts that were more rusted on their threads. Not wanting to shove everything right off the lift by using a bigger wrench, I just whipped out the impact driver again for those two bolts.

All in an evening’s work, and hopefully now nothing will creak either. I have yet to get a chance to road test for robust anti-ship suspension behavior, since there were some other things I wanted to take care of as long as I had the opportunity to stand completely under it.

little things

There’s some things I do that aren’t so epic and thorough as to warrant a poorly thought out name, but in the interest of completeness, I recorded them anyway.

I bought a can of underbody coating spray a while back, meaning to treat some not-yet-problem-spots to prevent them from becoming such later. Up on the lift, I could get at them much easier.

One of them is this front …. brackety thing that holds the steering rack and lower half of the A-arms. It was fairly rusty, with much of its previous undercoat treatment having flaked off, but not bad rust – just superficial. I knocked off much of it with a wirewheel, then manually with a handheld wire brush – whatever flaked off on its own was removed and then the whole area cleaned and resprayed.

Another not-yet-problem spot was this corner of the rear right wheelwell, which was showing the same signs of cracked undercoat with superficial rust. Luckily, this isn’t a hole yet. The same procedure was followed here.  With my rear hatch repair going off pretty well, I’m plotting ways to tackle the sides that DO have holes.

But before that, back to the interior!

What good is christening my rolling Craigslist wreckage MIKUVAN if I can’t blast Hatsune Miku from it obnoxiously? Well, I can, but not very well. The sound system is what could be called “stereotypically 80s” – cheap paper cones, tiny drivers, hollow panels and dashboards making for all midrange and nothing else.

I wasn’t out to completely re-engineer the system or add 12″ subs, but I did score a nice deal on some Pioneer TS-D1002R speaker sets for $25 a pair recently. Nobody really uses 4″ speakers any more, so I was pleasantly surprised when I was haunting eBay that there were many low and closeout/clearance type sales going on. The Pioneers seemed to be the ones with the least hype. I figured anything made in this century (literally… this century.) would be an improvement.

I began with an overall inspection of the stock sound. The left speaker is really easy – a little panel pops off and it can be replaced without hassle.

There is a very pitiful but honorable attempt at a subwoofer installed in the center underneath the dashboard. It’s a little plastic, stuffed box that is smaller than most computer speaker subs.

It was also empty. Like, no driver in it at all empty. I can’t imagine Mitsubishi leaving in literally everything but the driver if someone decided to forego the upgraded sound package or something, so I assume it was taken out some time over the past 24 years.

I didn’t have a voltmeter or oscilloscope at the time, so I couldn’t verify if the internal amplifier was alive. Given that it might have been driving a speaker with infinite! impedance, it’s probably fine.

Comparing new with old… The stock driver is a 7 watt unit. I think the PC speaker inside my old desktop from 1998 was larger.

Left: Super easy!

I only had to splice the two (20 gauge, such power!) wires – my little nano-iron is seen on the bottom right. I used up the rest of the exhaust silicone glue on it, by the way, since someone dropped it years ago and cracked the handle near the base of the heater. The element was jiggling inside (sounds familiar…), but that’s no more.

Center: also easy. At this point, I verified that the system was working, but very weak. It could very well be “broken”, or the amplifier IC half-dead.

This is left to a future investigation or replacement by a Ragebridge pulling a class-D at 100kHz.

The right side: You must be shitting me.

Nope, I didn’t get a nice little hatch out of Mitsubishi this time. The right speaker bracket is second to last on the list of things to do in order to remove the dashboard. After scouting the Internet, specifically the Delica forum (this thread), I found how sneaky people would do it.

The operation involved maneuvering an 8mm socket (the right speaker is mounted with 2 nuts on studs, not 2 screws) into the removed glovebox and around the corner. In many ways, it reminded me of flying a guided missile into a secret switch panel in Descent II’s level 13 to obtain the Omega Cannon.

I was a huge Descent player back in the day – don’t worry about it.

Unfortunately, the instructions pertained to a Japanese or Euro spec L300. Clearly there’s something else going on in the US spec Van, because there was a Bracket of Irritation in the way and I could not reach the mounting nuts without removing it first. This was a half hour operation on its own as I figured out the exact right combo of strangely angled sockets and extension bars to get all of the hardware.

It was also really rusty (surface only, fortunately), so it left this disgusting smear behind. Most of it came off with petrochemical distillate scrubbing.

Said Bracket of Irritation.

See where the Bracket of Irritation goes? Almost immediately behind it is the necessary mounting screw. Anyways, this is a post-installation shot – after the BoI was removed, I could remove the speaker, splice the cables, and reinstall.

With this mod complete in one night, I’d say the results were worth it. It’s not a sonic drag racer – I’d rank it somewhere near “Mid-2000s economy car”. The giant hollow dashboard doesn’t really help the case, but with appropriate EQ diddling, I get tremendously improved low-end response (read: discernable).

I may revisit the subyiffer (does not qualify as subwoofer) issue later, but I am still satisfied with the two stereo channels alone.

Stay tuned for the latest battlefield reports!

Winter is Coming.

Dec 16, 2013 in Land-Bear-Shark, Stuff

BurnoutChibi Reassembly: Inside the Vex Ball Shifters; RazEr as Art

Dec 08, 2013 in Chibikart, Project Build Reports, RazEr REV2

Quantity of Working Chibis: 2

I’m now back to no longer having any dysfunctional vehicles! So I bought a new one.

This…. is my life.

It’s a Razor “e-Punk” minibike similar in flavor to herpybike (Left), but much smaller. But it still uses standard 200×50 scooter wheels, so it lends itself to all kinds of potential shenanigans. I have no immediate plans for it, but at $20 for the wreckage, it’ll be around when I or an eager froshling have them. It’s actually engineered pretty well in the back end with a proper spring-loaded belt tensioner and sliding motor mount and all, so motor swaps are easy, and the battery tray can be repurposed for other components.


To continue the repair of BurnoutChibi, last time I finished making the new front wheel assemblies. Over the past week, I put the rest of the new front end together, and the thing is now operational again.

I also have a (sadly a bit brief) word about the redesign of the shifter mechanism – it hasn’t changed much physically, but this time I put just a pindrop of Design Mojo into it and the functionality was greatly enhanced. However, it was both short enough of a hack, and risky enough in my mind that I didn’t take a picture of the actual installation, but enough should still be present.

First up, to attach the new wheels to the frame, I had to machine a new steering upright (or knuckle) since the offset distances changed due to the new brake arrangement.

I took a liiiiiiiiiiiiiiiiiiiiittle more care in machining them this time, from 1″ 2024 barstock. You know, care when tapping so the first 8 threads weren’t stripped or something. The axle spindle is a 1/2″-20 bolt, and now the Set Screw of Kingpin Retainment is a 1/2″-13 thread instead of a 3/8″-24 thread. The latter, which were used in the first blocks, tended to wiggle loose since they didn’t have enough tip area.

The new upright, with brake caliper mount and caliper, and installed with kingpin. These calipers are a bit different – beefier – than the ones I actually designed up last time, so they came too close to the kingpin nuts. I had to space the kingpin out of the bottom with more bronze washers (I just gave up and have a small box of them now), then use a thinner jam locknut up top.

Hmm, the toe angle is a little wrong…

I hadn’t uninstalled the old steering follower links yet, so they’re still hanging off the tie rods.

Wheel mounted on one side. I can tell it’s better already. The caliper can be adjusted out slightly using the knurled screw such that there’s very little disc scrub but still rapid engagement. The right side disc does have some runout, possibly from the hub being slightly warped from welding. A post-machining operation to smooth them out would have been beneficial. The left side has almost no visible slop.

Here’s the installation from below showing the steering follower link attached to the tie rod.

Back up on all wheels! I stole the Hella switch from Burnoutchibi to borrow for… something. I’m not sure, but one of the fleet lost a Hella key (probably Melonscooter2), and I did not have a spare at the time. I since ordered a stash, so I picked a new one.


A closer view of the new front end arrangement. I test drove it like this with the worn out shifter (holding it in gear like I usually had to do anyway) to confirm and tune the brake operation. Since the two brakes were mechanically ganged together (fixed distance travel), I had to adjust the cable in and out to get the two engagement times as close as possible. This is one inherent disadvantage of multiple ganged cable brakes instead of hydraulic ones.

The shifter was having a harder and a harder time holding gears before the front tires blew out, and after this front end repair they were not particularly any better. I decided to break down the Ball Shifter transmissions completely and inspect them for damage.

I cracked the transmissions back apart to check for excessive wear, but the shifter mechanism is hidden in the shaft assembly in these, so I’ll have to go deeper. I never really tore deep down into these during the initial build, so here it is!

One thing I noticed right away is that the motor pinion and first stage gear were wearing through their hard anodized coating. Meaning now I’m going aluminum-on-aluminum. I suppose it’s ultimately a consequence of building these transmissions to survive a FIRST season or two. Even though the gears are 7075 alloy and would likely be stronger than a typical carbon steel gear, aluminum doesn’t wear nearly as well as steel. One thing that could be done is swapping the 14 tooth Vex aluminum pinion with an Andymark steel one; pinions tend to go first over larger gears.

So far, none of this has been the cause of any issues, nor do the teeth look substantially thinned, so I left it alone. \

I took the big retaining clip off the center shaft, and here is the result. The two shifter gears slide off, and then a bunch of little balls fall out. Well, nothing’s bad looking – the ball splines have some “Oh God It Hertz” divots, but nothing major. The balls aren’t deformed, their crossdrilled ‘tunnels’ weren’t smeared. So why wasn’t it staying in gear?!

It turns out that my shifter throw distances were incorrect. When my shift plunger hit its first gear hard stop, it was too far in. The action of applying torque forces the internal round lobe further into the transmission, but since it’s backed by a hard stop, it “works” anyway. The result when I shift is that the plunger stops too far short to properly engage 2nd gear, and the application of torque forces the balls radially inwards, tugging the internal lobe back. What I found was that I could not actually adjust the cable travel out to meet 2nd gear properly.

I’m thinking I must have missed one axial dimension, or applied it in the wrong direction, because the Vex CAD model does actually line up with what I got in real life; I had at first assumed the model was different. The proper plunger offset in first gear is 0.25″ from the inside face of the plunger bearing (on the far left end there) and where the hollow output shaft begins. My original plunger design was only 0.2″.

The travel was confirmed to be 1/2″ as measured center to center on the ‘ball tunnels’, so only my endpoints were potentially incorrect.

So, dammit, I’m gonna redesign myself some shifter plungers. On the left, the old style. On the right, the new one. Making the plunger stick out further required moving some of the geometry around to be compatible with existing spring lengths.

The difference between the two parts; old one on the left (too short offset), new on the right.

New variant shifter plunger installed. Result? Success – the balls not pushing on the ramped part of the internal lobe means no translation of radial force from torque application to axial force.

So, tl;dr start at 0.25″ and end at 0.75″.

Even though it did “work” with the existing shifter knob, the aluminum detents were getting extremely worn out; there was basically no feeling of which ‘gear’ you were in. This has, again, been the case for a while, and I had contemplated going to an all-electric servo shifter or some other fancy electronic method instead. But, as long as I was warming up my 3D printer to make the new plungers, I decided to apply a little flexural mechanics:

This is a drop-in flexural detent shifter module that replaces the steel ball plungers with a solid “spring” and two speed bump detents. I roughly calculated the force needed to click the solid spring’s contact surface over the detents as about 10 pounds at the knob – this is quite a lot, but at this point I wanted to be a little paranoid. It will probably get smoother with time and use.

The estimation method was:

  • Calculating (read: FEA) how much force it took to push the little ball up around 1mm, about 15 pounds force.
  • Dividing by the cosine of the roughly visually estimated tangential angle that the detent is acting on the speed bump at “about 60 degrees ish, kinda”which yields a “How hard do I have to pull sideways to generate x pounds of upwards force” number (30 pounds)
  • Taking that through an estimate of the coefficient of friction of ABS plastic on itself (0.35) to yield 10.5lb force
  • Noting that the lever ratio is roughly 1 to 1 between the handle and the point of contact, so let’s call it 10 pounds.

Various forces will come together to make that better or worse – it will, for instance, definitely get worse as time goes on when the speed bumps wear down and decrease the angle of action. If I grease it, it’ll definitely lessen the actuation force.

This is probably not going to last very long, but at least enough to get some grins in, and it was put together in one evening. I do want to machine the ‘shift gate’ version eventually.

This is the installation. There’s 2 different colors involved since I went back and changed the yellow piece a bit to force the machine to fill in the region solidly; before, the profile was too thin and the machine did not add any infill, leaving the spring area hollow and weak. By that time, I was printing yellow stuff for students.

The device as it stands now. This thing has been a total blast to drive around now that the front wheels actually, you know, roll properly. And the brakes are sensitive enough to decelerate extremely consistently. Actually, they were so sensitive that I almost brake’d myself off the front handlebar. A strong return spring has since been threaded over the cable in between the actuation arm and the cable stop on each side to alleviate this.

I went outside to see if I could “manually ABS” the new front brakes – it IS possible, just not very practical since the vehicle’s mass is so low. I’ve been practicing cadence and threshold braking on occasion in pre-widespread-ABS-era Mikuvan out of curiosity and an abundance of caution.

As usual, it’s missing good test footage. I’ll need the weather, availability of photographers, and mutual desire to go out and test stuff in this season to build up first. Inside, the vehicle is severely traction- and space-limited, and indoor testing video won’t really be worth watching.


As implied in the title, I’ve managed to somehow break into the design gallery scene:

What?! Yep, that’s RazEr REV2! On display, in a fancy architecture design gallery in Boston. Fancy.

How the hell did that happen? Long story short, the IDC’s population has a large Architecture and visual arts/design minority, and two of the researchers happened to be affiliated with the BSA. All of my small rolling contraptions seemed to be a fit for their “Urban Mobility” display this winter and next spring, so I was solicited for potential display items.

Originally, they wanted Landbearshark as a stark contrast to everything sensible. I almost agreed, but with the season of subarctic melancholy on the horizon, wanted to keep it around for shenanigans.

What’s nice and practical, looks reasonably well finished, and completely useless in the snow? RazEr.

It gets its own Fancy Display Platform complete with ipad scrolling some of the build pics.

There were other Fancy Hardware examples too. You may remember the Tribey from Mt. Washington. And the green thing in the background is an Lemelson-MIT Prize winner. Winning at what, I am not too certain. The display runs through May 2014 and can be viewed at the BSA public gallery.