Archive for November, 2013

 

BurnoutChibi Repair and Revival!

Nov 28, 2013 in Chibikart, Project Build Reports

Chibikart just can’t seem to catch a break.

Even in non-hub-motor form, it’s been sitting in a pile of its own wreckage. During the late August summer go-kart finale, I raced against some of my own students and TAs, and about midway into the race, the left front wheel just completely locked up, skidded, and blew out. It’s even in the video:

 

I never really got a chance to drive it well, since it was built during the rush of organizing the two kart classes, then rendered mostly dysfunctional right afterwards. Since then, it’s been living alternately under the bench and on top of it, reducing my moral authority to command people to clean their workbenches to near zero.

I originally used the scooter rear drum brakes as a parts examination, and came to the conclusion that no. But, it’s not really their fault – most of the failure is from the fact that the wheelbarrow tires are completely and utter trash – they’re out of true in every possible fashion! It’s almost as if they were not designed to be used on go-karts or something. Imagine that.

Anyways, I’ve had a mind to swap Burnoutchibi over to front disk brakes almost since I tried stopping with the drums for the first time. I figured that if the solution involved a custom non-shoddy hub for the tires anyway, then there’s no point in trying to machine or adapt a custom drum when a brake disc is just a flat chunk of metal.

That effort finally got rolling last month some time when I began to fiddle with my left over brake calipers. The brake calipers I speak of are the generic $10-20 scooter calipers commonly found on small gas and larger electric Chinese scooters, the kind retained by the likes of Monster and other scooter parts places. They’re simple, but get the job done.

What I found in the course of my research about them is that nobody has ever modeled one in 3D. That I know of, anyway. Nor does an original model seem to exist at all. I’m strongly of the opinion that these things were copied once, then everyone else just duplicated dimensions without really thinking about what’s going on.

And so, desiring a model which I could line up in CADspace instead of assuming things about those mounting points, I set about correcting this issue by virtualizing the brake caliper. Now, these things don’t have a SINGLE straight edge to datum some dimensions off of, so I had to resort to tracing over an image:

I took a very long-zoom picture of the caliper, end-on, and used it as a the backdrop to a sketch full of freehanded splines. The tight zoom is to flatten out perspective so I could capture all the features on top of each other. I didn’t do it perfectly, as some of the mounting screws might show.

This sketch became the basis of several features that used the profile information, in conjunction with real life caliper measurements, to generate some solid model parts of the caliper. Critical dimensions such as hole spacings were found physically and then those dimensions overrode the raw picture traces.

And there it is, a model of that which has eluded modeling. These dimensions ought to work for most generic small caliper brakes – later on, I show which ones I ended up getting more of, and even though they looked different, the holes and spacings were the same.

I’m putting this file up in the References section as brakecaliperrh.zip, in which are Inventor files, Solidworks files, and a generic X_T (Parasolid) for importing to other CAD platforms. Hole dimensions are exact to 0.1mm, and the axial spacing and thickness of the parts represent worst-case scenarios on purpose, such as designing to clear parts around them will definitely clear the calipers in real life.

To verify, I grabbed a chunk of virtual Tinykart from Shane and dropped my caliper model into the ad-hoc physically located model they made for it. Result: Pretty much an exact match!

The next step was to adapt the new brake caliper to the front end of the kart. The placement of the holes and generation of the adapter plate was driven entirely off the position of the caliper. I made midplanes and mating axes to space the caliper out where I wanted. I was going to make a custom brake disc anyway, and the brake disc diameter was chosen based on not being so large as to hit the ground if the tire goes flat, using as much of the brake pad surface as possible, and being large enough to clear its own hub mounting bolts. Ultimately I decided on a 100mm (or 4 inch) disc.

A new ‘upright’ block, which the wheel spindle bolt tightens into, is required because the axial spacing of everything was so different. This set was made hastily and sort of out of square anyway.

This is the new combination hub. Brake disc mounts on the little end, and wheel mounts on the big end. The tube in the middle is a 3/16″ wall DOM steel tube with slightly machined ends. It’s basically a smaller, cuter version of the average commercial go-kart hub, like some of these. The discs are to be waterjet-cut, then welded to the steel tube.

I would have cranked these out of some aluminum billet, but I wanted to pilot this method for Chibi-Mikuvan, which will use a slightly larger version of this.

The planned assembly overview.

The idea is to strip the shitty rolled steel hub completely out of those wheels, using only the stamped rim halves and bolting them to my own hub which will, in the best of circumstances, be more true-running and less… I dunno, shitty than the stock wheelbarrow-grade hubs.

An overview of the new assembly to be manufactured. No matter what, I couldn’t get the disc brake design to have the same front track (width) as the older version, so this one will have slightly wider track at the front – but only a half inch. Not enough for anyone to notice but my alignment-obsessed self. That’s the one downside of using disc brakes with these smaller scooter- and handtruck- type pneumatic wheels. Unlike a car tire which is deeply dished on the inside to fit the hub, steering knuckle, and braking hardware without it all sticking out, here you usually just have to deal with a scrub radius of 2 or 3 inches.

(Unless you want some seriously wild camber. Stancekart?)

Brake rotors, hub components, and the caliper adapter all cut out from 11 gauge cold-roll steel (0.120″). To the right are some replacement steel steering links. The aluminum ones, while they worked in principle, started smearing their hex bores out eventually so the steering backlash began growing with each test. I decided to just make a weldable version; the geometry is otherwise identical.

The beads were dropped with a MIG welder for expediency. I machined each steel tube end to be a slight press fit in order to keep the pieces from warping while being welded.

An interesting feature I noticed about the steel tube I got from Speedy Metals was that the outer and inner surface were both very shiny and tight-tolerance. This is in contrast to most DOM style tubing where there is a visible weld seam on the outside and inside, and the sufaces are dark with oxidation. This makes me wonder if it was actually extruded – I’ve heard of extruded steel but wasn’t sure if it was a common process.

So far, this only applied to the 1.5″ OD, 3/16″ wall stuff I got for BurnoutChibi – the other sizes that came in the same order for Chibi-Mikuvan are normal looking.

To cut off the stamped rim from the hub, I took a boring tool to it, cutting on the back side with the machine running in reverse, until it just popped off.

Using some 1″ OD by 1/2″ ID aluminum extruded tube, I made the internal bearing spacers. The faces of the spacer were turned down to leave a little shoulder so the whole thing wasn’t rubbing on the bearing seals and races. There’s no reason the spacer has to be so thick radially, but the close match between spacer diameter and hub ID means there’s less playing “Guess where the spacer is!” when shoving the axle through.

Also shown are the non-trashy R8 flanged bearings to replace the very trashy stamped steel lawn mower bearings which I swear use MIG welding spatter beads as their balls

Finished hubs with holes tapped. The wheel side threads are M6 x 1 to reuse the bolts they came with, and on the other side, the brakes are #10-32. Yes, mixing metric and U.S. threads. Yes, I am a lazy american.

Wheel mounted…

And both wheels completed.

The hub is designed such that the small projecting boss of the steel tube is the right diameter to align the center hole in the wheels. This contributes immensely to true running. Combined with the flat hub and real bearings, the amount of wobble in the wheels is now minimal, and practically none at the brake disc.

Moving onto the new steering links, this is how they fit onto the existing kingpin bolts.

And they are welded the same onto them. Now there’s no more need to constrain them from the bottom with a bolt and washer, and ideally this system will have no backlash…

What I have to do now is to remachine the upright blocks and then mount the wheels. At the least, then I can try a test drive to see what difference it makes.

Next on my list with Burnoutchibi is a rebuild of the shifter mechanism – while the principle was sound and it worked great in the short tern, the aluminum ball detent slots rounded off pretty quickly and right before the race, it was pretty hard to keep in gear. I’m going to move towards a spring steel bar (incidentally the same spring steel bar that I made 12 O’Clocker’s front legs from) that can bend in and out of slots, like a lawn mower shifter gate. Since there’s only two require positions, it’s not exactly difficult to come up with a pattern for it!

 

Beyond Unboxing, Chibi-Mikuvan Miscellaneous Engineering Edition: Inside a 9-Inch Angle Grinder Gearbox; Hobbyking T20 Inrunner Motor

Nov 24, 2013 in Beyond Unboxing

This post will wrap up some more of the components I’m aiming to incorporate into this build. Recall that part of the mission of Chibi-Mikuvan is to use a jumble unconventional parts together as a technology demonstrator of sorts, so I’m exposing the inner workings of a handful of potential part sources not typically seen in public together Previously, I dove into the Motor Controller of 1000 Cool Story Bro Amps, then the Dramatically Over-Engineered Batteries of Doom. This time, the teardowns aren’t as epic or novel, but as usual I figured the more pictures of things, the better. The story now moves to the drivetrain parts; in particular, one way to get a compact 3:1 or 4:1 reduction, and then a few pokes inside the motor I settled on using.

First up:

cheap angle grinder gearboxes

Angle grinders are three things: A way to really quickly embed little abrasive rocks into your face, a fast and powerful motor, a high-speed right angle gear drive, and a doubly-supported output spindle that usually even comes with a little nut to attach deadly centrifugal grenades to. In the past, I’ve personally seen them used on some Battlebots in a weapon application for overhead bar spinners in the “sublight” weight classes (12 and 30 pounds). At one point during my early days, I had the parts of 4 or 5 cheap Chinese angle grinders floating around my combination bedroom & machine shop.


I never said I was smart, that’s just what everyone else says… (Picture from 2004)

Angle grinders tend to also come in two major classes: For parts, or to be used as tools. I’m concentrating on the former here – the so-called “Harbor Freight-class” angle grinders that typically sell new for $20-30, if not even less on discount.


Some time featuring such things as “noise reducing gears”.  (Picture from 2004)

Okay, so I literally haven’t seen a single plastic geared one since then, but the precedent is set!

These days, I assume the Chinese gear-hobbing industry is better established. You can even buy individual grinder gears on eBay nowadays, if you want to build your own housing, and by far they are the cheapest way to get a right angle drive; however, like repurposing car parts, they aren’t sold by tooth counts and equivalent diametrical pitches, but exact model replacements. So to use them, you’d need to do a bit of ‘shotgun designing’. You can even use them to make differentials like God intended. Bear in mind that the steel quality for most of these gears is likely a little on the shady side – I recall being able to machine them easily using my primitive garage tools, so the steel is most likely a low or unhardened medium carbon type.

I specifically picked these out of the back aisles of my mental design warehouse because I was in need of a way to make a very high gear ratio, on the order of 20:1, in a small space and without being too expensive. Spur gears were essentially out of the question right away due to price, even Andymark gears, since at least 3 stages would have been needed. Chain drive was a little more feasible, but still, the cost of support materials like the bearings and shafting to form the intermediate stages was high. I tried to think of clever ways to get a high reduction without causing the material cost to exceed the PPPRS $500 threshold.

And then it hit me! Something I forgot about for many years now seemed like the obvious strategy. I began the eBay hunt and tried to cross-correlate different angle grinder models with their gear tooth counts, but what I found about the smaller 4 to 5 inch disc grinder size was that their ratios were really low (2.5 or 3:1, or thereabouts) because the torque levels needed to drive a small grinding disc were not that much. Ben’s differential build above shows a pretty typical 4 to 5 inch class gearset. This wasn’t going to be sufficient for a first stage, since I was constrained by wheel and sprocket size to no more than roughly 5:1 in the second.

I decided to try a different method of finding out what gears were in which size grinder: Going to my local Harbor Freight and literally taking apart their display models in the aisle, with a screwdriver sourced from their hand tools aisle. The manager was nonplussed, but backed down after I explained that I was actually doing an engineering study and would buy the display model that fit my needs. Sadly, I didn’t have any pictures from this excursion.

I got the 69085 9″ grinder display model (an older version; all of the boxed ones were this new gearbox design) for $30 after some explanation, without any of the frills. My hunch settled on the 7″ through 9″ sizes having bigger gear ratios because the torque needed to swing such a large disc combined with the motors not being that much larger across pointed towards it. So let’s see how this looks inside.

Four longitudinal case screws and four dorsal gearbox screws later, the whole thing sort of falls apart. The motor itself is a hefty universal motor – a brushed DC series-wound motor with laminated stator to enable it to run on AC with less losses.

In my opinion, this motor can be rewound to run effectively at 24-36 volts just by replacing the many turns of thin wire on the stator with a few turns of very fat wire. The stator coils measured 2 ohms, so the stall current of the motor is quite low if used stock at that voltage. The armature resistance was around 0.25 ohms – high, but not the end of the world, and you can find low voltage motors that have a higher resistance easily. Maybe I should just do this instead!

The gearing is already starting to look promising.

The pinion is only retained by a nut on the end of the motor shaft, and as I found out, there’s no other power transmission medium in it except the torque of that nut. To remove the nut, I stuck the rotor in a vise and uncranked it with the appropriate sized wrench. Then, a little rubber mallet coaxing of the gearbox housing popped the shaft out of the gear and input bearing.

Here’s the gearset! A solid 49:12 reduction, or 4.083:1. Why 49 instead of 48? It’s so the teeth wear more evenly. The greatest common divisor of 12 and 49 is 1, and the least common multiple is 588, their products. Not only does this mean that it will take 588 turns instead of 4 for the same two teeth on pinion and gear to meet again, but it also disturbs any potential 4:1 mechanical resonances and harmonics that can pop up, contributing to smoothness.

The gear pitch is somewhere around module 1.5 (or about 16 pitch). The little gear has a plain 10mm more, and the big gear has a 15mm bore.

The output gear is retained axially by a single snap ring (which makes me feel really good about hanging a giant grinding disc off it, I’ll say). Rotation is ensured by a 4mm thick woodruff key. This is the same as the little grinders, just more metal.

Short of machining a custom housing, the most useful form of angle grinder gears is inside the angle grinder gearbox itself. The idea that I settled on is to machine a 10mm shaft that has a standard metric keyway cut into the end, broaching the small gear with that size key to make for a positive power transmission coupling. I’ll retain the threads on the end to lock the pinion in place axially. The nice thing about the pinion is that the right spacing relative to the output gear is attained just by running it against the input bearing – a good move for repeatability.

I’ll probably purchase hardened woodruff keys for both sides because I’m inclined to believe the ones that are included are very soft steel; at least, the “file test” made a huge divot in the output side one, which is the most likely to shear.

I made a ‘important dimensions only’ model of the gearbox for use in the design – it will be released once I validate it.

Full disclosure: All the bodywork on Mikuvan has involved a 4.5″ Harbor Freight angle grinder. It works just fine.

the unnecessarily large inrunner that will beast into it

There’s some pictures that just shouldn’t exist. For instance, this:

No, not the Moxie, but the inrunner that’s almost as big as it. Disclaimer: I have no clue what Moxie Cola is; this was given to me by someone, and I’ve actually been too scared to open it.

That motor is the Aquastar T20, a “1/5 scale” class inrunner for boats. So, I don’t understand the 1/5 scale R/C vehicle class in much the same way I don’t understand model airplane scales that are indicated in percent, like 33% or something. To me, when a radio controlled model gets that big, why don’t you just fucking get in and drive it yourself? A 1/5 scale model car is already a go-kart!

I typically advise people to stay away from inrunners because of their tendency towards extremely high speed (high Kv, or RPMs/volt) and consequently lower torque than an equivalently sized inrunner due to the smaller rotor size. It’s not as optimal a setup for small vehicles, in my opinion – that, and they are far harder to append Hall sensors to unless they already come with it.  However, when they get ridiculously sized, it’s a different story.  This motor is just slow enough that you can build a rideable vehicle using the Burnoutchibi principle: running a fast motor with a very high gear ratio to divide down your own apparent mass, and using a high capacity R/C controller instead of a dedicated EV controller.

This is where my number of 20:1, previously mentioned, came from. With the motor’s wye-terminated speed of 730 RPM/V and running the 28.8v system described previously, it works out to about 21,000 RPM, which isn’t far from what the angle grinder motor would have made anyway. Geared 4.08:1 and then 5:1 externally, the output speed at the wheel is theoretically 26mph. That’s just in “gear 1″.

One of the major reasons I selected this motor, besides straight up motor pen0r (that’s a technical term), is because it can be externally terminated in Y (wye, star) or Delta. The difference is how the windings interact with each other inside the motor – in a motor power system that is otherwise the exact same except for the type of termination, the Y-terminated motor spins 1.7 times slower (actually √3) with 1.7 times more torque. The science of it is more complex and has to do with the windings being placed in series in the Y termination, among other factors (see Mevey Ch. 2 for the rundown). My hub motor instructable of yore assumes you wind in Y.

This means that Chibi-Mikuvan could have two electrical ‘gears’, to contrast with Burnoutchibi’s two mechanical gears. Switching between the terminations without actually pulling wires means I’ll need an additional multipole switch or contactor rig to splice phases and connections. I have a few designs for this, and I’ll also post those once they’re validated. The hypothetical top speed in ‘Delta gear’, as compared to “Y gear”, is around 40mph, though realistically it will be less due to the nonlinear effects of wind resistance. I haven’t really thought about anything going that fast since the LOLrioKart days.

Let’s crack this motor apart:

Well, that was easy. Six faceplate screws and the rotor pops right out after some tugging. This is a 4-pole, 3-phase, 12-slot (or tooth) motor – most inrunner motors are build to be integer-slot like this.

This motor has a shorter rotor than what the can would indicate, with the extra space taken up by a spacer bushing. This is because the Hobbyking version is actually the smaller one. Elsewhere, this motor is called the “X520″, and yes, they make a longer one called the X524 (example 1, example 2, example 3… in case they vanish one by one), if you need an EVEN BIGGER MOTOR PEN0R (that’s a technical term).

The stator windings are very cleanly done up, though they don’t seem to be lacquer-coated for heat resistance. In most industrial motors, they dunk the whole stator in a resin that seeps into the windings and helps secure them at high temperatures and prevents the magnet wires’ enamel coating from coming apart. The whole wound stator seems to be mashed into the can, and that red heavy paper layer is presumably there to prevent it from being mashed too much.

I tried my best to pull the rear cover of the motor off, but the bullet connectors are very tightly press-fit into those plastic pass-throughs. Furthermore, the stator itself is also pressed into the can, so it wouldn’t have done much good. This was as good of a picture as I could get of the distal end of the motor can. You can just barely see where the terminations are brought out and soldered into the bullets.

The rotor length is 50mm…

…and the rotor diameter across the magnets is 27.8mm. The 4 magnets are wrapped in some kind of resin impregnated fiber. Now, it claims to be Kevlar, but I vote dental floss.

The boat variant of the motor has a water jacket, which I think will aid greatly in continuous power dissipation. It’s a very simple ring type one, with no internal turbulation devices or flow channels, so its effectiveness might be limited in comparison to a more rigorously designed one, but the tube structure is easy to make. My only beef is that the inlets are too small to use with regular PC water cooling equipment. The nozzles are 1/8″ ID barb fittings, so 1/8″ silicone or PVC tubing is the best you can do, but most PC stuff is 8mm or even 10mm. I haven’t designed or even thought about parts for the water cooling loop, but it’s something I do want to incorporate because inrunners seem to love to run hot.

The motor has a flatted 8mm shaft, but I almost wish it didn’t, since a flatted shaft makes it more difficult to use a collet or friction grip system to transmit power. I suppose for the application these are intended for, the flat is welcome, but my shaft coupler to run this into the angle grinder gearbox is a simple collet-like system identical to what I keep putting on my 3d printers and battlebots. I call it the “ninja coupling”. I also have just a bad impression of flats coming with my motors because of the nightmare that was aligning the collets on old Deathcopter’s ducted fans – a combination of poor quality collets and the flats on the motors meant that balancing the damn things was basically a crapshoot. In this case, a sturdy and well constrained motor mount would prevent that.

That does it for these parts! The only thing left to do is the Chibi-Mikuvan global engineering update itself, and that will come in due time; plus, I have some update for BurnoutChibi. So now it’s time for….

daily van bro.

I often have to avoid main thoroughfares and their associated never-ending traffic by ‘leaking’ through neighborhood blocks and side streets, where I have a vague sense of where I need to get and just navigate ad-hoc, re-orienting every once in a while by trying to find the Prudential Center. This has shown me a good chunk of the vehicular underbelly of the area, like when you lift up a rock in the woods and about 80 different species of bugs and small mammals all scatter. One day, I found this quite lovely Dodge A100. Along with the Chevrolet G vans, and Ford Econoline gen1, it was part of the American trio of derpy vans from the 1960s.

Maybe these should all be on my hit list too – it’s like Pokemon #800-951 (they’re up to that many now, right?)

Another potent candidate for Vans Next To Things! Here’s a great size comparison – even the “compact vans” were still bigger (which is weird, since Mikuvan is larger than a Greenbrier?)

Incidentally, right up until 2009, you could purchase a new 3rd generation Mitsubishi Delica in Mexico as the “Dodge Van 1000″.

Y’know, I really, really think one of the reasons cab-over vans never caught on in the U.S. was because we kept giving them names like “Van”, or “Van”, or “Van”, or the like. Would you buy that shit? No, but I would totally buy the SUZUKI EVERY JOYPOP TURBO. Or, since we’re American and macho here, the FORD BEASTMASTER GREAT ADVENTURER SRT, perhaps?

Either way, I know what I’m doing if I ever need parts on a greater scale…

Beyond Unboxing: How to Take Apart a Ford Fusion Hybrid Battery While Minimizing Death for You and Onlookers

Nov 13, 2013 in Beyond Unboxing, Stuff

Alright everyone, it’s time to get serious! This is the first of most likely several engineering posts directly relevant to Chibi-Mikuvan (see its introductory post here), in which I’ll delve into the parts that I’m intending on using in the build.

No, I was not kidding one bit.

・ω ・

It’s the hallmark of this whole thing.

Anyways, recall that most of my recent EV-related builds are in the interest of pathfinding and trailblazing. Taking a stab in the dark at a potential useful part for makers and builders and seeing how it works out, some times involving a little bit of science, and possibly incorporating it into my personal go-kart kindergarten. That, and ultimately looking silly. With this build nominally trying to fit under (or at the least, not outrageously around) the PPPRS $500 parts and materials budget, I’m intending to try several new tricks at once, and all of them will be exhibited here.

This post in particular will focus on the one pivotal strategy that I explained in the post-Maker Faire post – cracking apart a second generation hybrid car traction battery and using the internal cell modules. These modules are typically NiMh chemistry; this was before everyone switched to lithium ion batteries this and last year. NiMh lends itself to highly favorable energy density over lead-acid batteries. Being several years and tens of thousands of miles out, you can find these batteries sold by auto recyclers for between $300 to $500. For people looking for huge capacity in less weight, I think the cost per watt-hour of energy of a used hybrid car battery is the best you can get.

In fact, my final cost in this investigation was $300 per 2.15kWh, or $0.139 per Wh. This beats out even Hobbyking lithium polymer batteries at 33-50 cents per Wh (which aren’t allowed in the first place in the series!) and most lead-acid battery options short of ‘found objects’ or purchasing at heavy discount. This is now on record as being possible, which, really, satisfies me even if I don’t end up build my own entry with it.

Someone else can totally take the torch from me right now! I’m certain somebody’s already dropped a hybrid battery into a custom project, but this is where I declare pics or it didn’t happen. Otherwise, feel free to scoop me on actually using these, because I’m sure I’ll never get around to it at this rate.

So here’s how it went down:

Initial reseach

The first step was hopping around on Wikipedia, hybrid owner forums, and Google Images compiling a list of battery candidates. I had a few in mind after a day’s worth of looking around:

  1. Toyota Prius, 2003-2009, or the “Gen 2″. This was the duh option – millions of these are on the road, so their parts are bound to be the most plentiful. The prius battery is a block of 28 “blades” composed of 7.2v modules, each 6.5Ah. The total battery energy of the car is  302V and 1300Wh (total, not useful – there’s a distinction). I was particularly interested in the Prius at first because the 28 blades could be arranged into a 28.8v, 45.5Ah battery by putting four 7.2v modules in series, and seven of those in parallel. 28.8v happens to be near the maximum voltage rating of many inexpensive R/C power system options.Along with the Prius, other same-gen hybrids like the Camry Hybrid and Toyota Highlander hybrid share the power system.
  2. Ford Fusion Hybrid, 2009-2012. This option was familiar to me because the MIT EVT chopped the powertrain out of a 2009 Mercury Milan Hybrid and converted it to full battery electric. The car’s advertised battery is 275V, 2.1kWh. We took apart the large hybrid module about 2 years ago, long before I had any delusions of using it on a small go-kart (since we were, and still are, up to our nostrils in donated A123 cells). I knew already that this battery was composed of 28 9.6v, 8Ah NiMh D cell modules, which can be arranged into a 28.8v  72Ah battery. Similar year Mercury Milans and Lincoln MKZ share the same powertrain.I also briefly looked at the older Ford Escape hybrids (2004 to 2008 years) which used a higher voltage, lower capacity (330v 5.5Ah) 1.8kWh battery, but I did not find many for sale, and the familiarity with the Fusion pack meant I focused on it harder.
  3. Honda Insight and Civic Hybrid, first generation, 2000-2006. I knew little about these except that they had 7.2v, 6.5Ah modules arranged in a 144V battery pack, which means they came in at a clean 1kWh. These are getting rare, however, and I did not find many for sale that weren’t very expensive.

I found an interesting price curve when studying eBay’s listings. Well known hybrids like the Prius commanded high prices because you know people are going to need parts, and will pay for them. Rare hybrids like the Insight command high prices because of the few people who are still keeping them on the road. It was the hybrids that nobody really cared about, like the Civic Hybrid and Fusion Hybrid (and relatives) which seemed even remotely plausible. Since I knew what was coming in the Ford Fusion battery, I decided to focus harder on it, and still keep options open by inquiring about the Prius.

Purchasing

I was aiming to get $300 or less for one of these units. This was the number that I decided would make sense for a PPPRS entry – you take the battery, split it in two to obtain 28.8v and 32Ah so you can have two of them to swap out between, and each battery half will only count for $75 on the budget under their accounting rules: $300 purchase price, $150 budgetary price, and half of that running at any time.

I began by scoping out these batteries on auto yard listing websites such as car-part.com, to identify potential candidates. The plan was to get a handful of leads, then call the yards themselves pretending to be an instructor for a MIT electric vehicle design team whose students are carrying out a project where they want to utilize a hybrid car battery, and would you happen to have any as-is, repairable, or core return batteries we can use?

Only 10% of that sentence was bald-faced lying through my teeth, so I felt clean!  The hope is that someone would drop their price for such a noble cause. You might call this unethical, but it was also a good probe of how open these yard operators were to the idea of makers and hackers (mis)using their parts, especially young ones. If you weren’t actually going to put it on a car, the yard doesn’t have to worry about a warranty or whether or not the part is damaged or not functional to OEM spec. Even a worn out battery from an automotive perspective is most likely still useful for small vehicles. Hence my emphasis when I called on these packs being rejects in some way.

This was an adventurous roughly 2 weeks. I would call to find they have no stock, but be told to “call back next week in case” – so I did. So many of these were “Call this number and ask for Billy” type situations, which I frankly love. Here is a sample of the text file that I was compiling:

Chuckran’s 508 697 6319  x

Robertson’s (800) 551-7000 x

Linder’s 800 521 8000 ($900 07 Prius ) ($700 09 Fusion) Jesse sent you

D Richards Jack 800 776 0459 $300 ! 10 Fusion

Tom’s 800 255 6656 (left my info)

A-rite Auto 800 874 7116 $700

LKQ 800 500 8733 Candia NH call back next week JC

Bishop’s 860 346 2336 x check later

Goyettes 800 640 7548 fax 508 207 1546 x

Self-Serve Used Auto Parts 1-508-763-4442 Check in a week

Route 128 Used Auto Parts 780-980-0025 (calling back soon?)

I ended up calling 18 yards in increasing distances from Boston, gathering prices and references (When a yard didn’t have stock, I asked for a recommendation on other businesses to call). Several of them led me to each other by accident. The price was slowly falling the further I got from the metro area – this made sense, after all, since I’m sure not only the cost of doing business is higher closer to an urban area, but more cars means more wrecks and more throughout and more people needing their car back on the road; all leading to higher prices. I started breaking into Connecticut and New Hampshire.

Ultimately, my lead came from Burlington, Vermont. So one Friday morning a few weeks ago at 5:30AM sharp, I blasted out of Boston and ended up 45 miles from Canada about 4 hours later, collecting no maple syrup or moose in the process.

Or bears. As I wandered further up I-89, the “_____ Crossing” signs got more and more hardcore. Bears? What’s next, T. Rex?

I must say, if I spent my life here, I’d be a nature hippie too. Burlington is pretty damn beautiful, especially down by Lake Champlain.  I wanted to pull down by the waterfront to get a Vans Next to _____ picture (the VT-NY ferry was a good choice), but the area was blocked off and full of tour buses. I decided against pushing my luck.

The yard itself was about 15 minutes north, in a small town called Colchester:

D. Richard Automotive was the name of the place, and I didn’t exactly look out of placed parked amongst their selection of used cars. A few what the hell is that questions were received from some of the customers and staff, to my amusement. Apparently Mitsubishi vans just never quite diffused evenly throughout the country.

The owner, possibly Mr. D. Richard himself, was very receptive to the idea of student-built EVs using these packs, and I made some idle chit-chat with him about 2.00gokart and the Makersphere while his… Ninjas? Cronies? were retrieving the part from stock. It’s a little far to drive every time something good comes up, but more links in the network, the better!

This is the thing. If they made Hybrid Mikuvans back in the day, I’m guessing the battery will go in a not dissimilar position in the rear. It weighs about 150 pounds (judging by my non-ability to lift it). Check out those bullet connectors.

A four hour drive back home and some Bro Assist got this patient on the operating table. The T-square is to illustrate how big this is – about 20″ wide by nearly 4 feet long. By this time, I was pretty much worn out from driving totally randomly into 3 different states to get a part for a silly go-kart, so I passed out. The following pictures were taken some time after the fact.

That thing above the battery is a cooling fan that pulls air through the cells and exhausts it to the side.

These are the side vents in the case, through which you can see the cell modules.

The fan shroud is held on by a handful of small hex head bolts. Use a 8mm socket (5/16″ works too) to remove them.

Before I totally took the fan shroud off, I decided to have a look at the fan itself, since it’s a brushless fan for long life and greater efficiency. What was it – a really hardcore Hobbyking controller driving it? Could I turn it into a hub motor?!

The fan pops out with the same little bolts, three of them. Under the big white sticker are 4 little Phillips head screws, and after you remove that:

Holy hell, it is a Hobbyking controller! A really, really hardcore one…

This thing appears to have its own DC-DC converter in line with the power input. Judging by the 35V output capacitor, it probably modulates the speed of the fan only through this DC voltage, either raising it past 12V or lowering it, commutating the phases fully on or fully off at all times. This architecture is called a current-fed inverter, and it saves you from having to PWM all the 3 phase bridge FETs, cutting some losses and complexity.  I would have literally lobbed a HK controller in this, startup beep and all. I guess this is why I am not working for a car company.

(If I had it my way, everything would have a V8 made of Melons in it.)

Once you pop the retaining ring on the back side off, the rotor dismounts:

It’s a copier motor.

I literally burst out laughing uncontrollably when I found it. It is the exact same dimensions as a typical 68mm copier motor I’d harvest from some poor Xerox to use in a scooter hub motor. Hey, this winding might even be fine as-is for a hub motor, since it’s a fairly high voltage motor!

Some larger Torx screws also hold the lower fan shroud on. A T25 Torx bit is the proper way to attack these. With the fan shroud tossed aside, it was time to start cracking the battery itself open.

Here’s where I pause for a second to say:

The following images depict opening up and messing with a 275 volt DC power system. It will be exposed, be unfused, and be extremely dangerous. If you touch the wrong spots, you will commit suicide instantly. If your metal tool touches the wrong spots, it will explode in a plasma cloud and deposit itself on your face. It is your responsibility to educate yourself on proper handling of, and safety-oriented behavior around, high voltage systems.

tl;dr be safe around batteries, kids. These pictures are for your reference and educational value only – I am not here to train HV electricians. As you’ll see, the battery pack is extremely well designed and anticipates service and disassembly, but it makes best practices no less necessary.

Now, onto removing the battery’s shell.

First, I had to remove the various other warts and boxes on the bottom of the battery. These include, from the left to the right, the main contactor set and output connectors, the orange  service disconnect, the battery charge monitor controller, and the battery computer which interfaces with the rest of the car.

These are all held on with nuts on studs, and I used a 12mm deep socket to reach them.

I decided to ruin the resale value of these OEM battery parts by prying them open and looking inside. (Seriously, I could have made $150 on each of these battery controller parts!)

This is the “Battery computer”, as it seems to be called . I’m assuming this takes battery voltage info from the other box (with the HV interfacing hardware) and pipes it to the car’s ECU. I’m not focusing on reverse engineering this stuff, so it’s here for visual stimulation purposes.

 The other box contains the high voltage interfacing hardware – check out that isolation gap! My guess is this has an array of multiplexers and ADCs to sample the battery voltages at the module level. That or a huge string of op-amp subtractors, Hobbyking style, with 400 volt opamps?

Some more prying and biting later:

Looks like the two theories might be combined. On the left board (the first module) are what appear to be several 14-pin, possible op-amp or comparator ICs. The part numbers are all proprietary – even though they feature logos like NXP and Texas Instruments, the etched numbers all seem to be serial numbers of some sort. The myriad passives on top and on bottom suggest to me a possible series of op-amp circuits.  The right board has a shoreline made of optocoupler solid state relay devices, KAQW216 (the chips are marked W216HC5). Most of them seem to eventually pour into (or out of?)  a large Analog Devices chip, an ADUM1401 4 channel digital isolator, surrounded by some quad op amps (AD8554).

Again, I’m not out to decompose the battery management system, since I won’t be using it and it’s surely very proprietary, so this is as far as I’ll go here – component staring and hypothesizing. My fan theory is that the 4 channel isolator turns on the solid state relays in a multiplexer fashion to select groups of cell modules, the top end and low end voltage of which flow into the op amps to be subtracted to get a voltage number; combinations of groups of modules being read will let you deduce the voltage of each individual module if you keep track of what is being read at any one time.

Moving on:

This is the contactor bank of the battery pack. There’s three separate contactors: one for positive, one for precharge (connected to positive), and one for negative. In vehicles, traction batteries are generally fully isolated when the vehicle is at rest. It’s not frame grounded like you would do for a 12V auxiliary system. It’s great to see all the systems my students typically build into a small or medium size vehicle (like the precharge circuitry) because I tell them to, being used on an industrial scale because you neeeed it in order for the system not to grenade instantly.

The set of two contactors at the top left control positive – one of them fills the HV rail through the precharge resistor (the big sand block resistor next to them). The single contactor at the bottom is the negative. The only thing that looks a little cheesy is the great big negative bus bar that spans the entire length of the thing.

Finally, the black hole that the positive busbar disappears into is a Hall Effect current sensor which, again, seems to be a proprietary part. Maybe I will take it apart in more detail and try to extract the functionality later.

Update:  The current sensor is a LEM DHAB-S/34 type, which outputs an analog voltage proportional to current. It has dual sensitivities – one pin for 40mV/A, the other for 10mV/A, so you can monitor low currents and high currents alike while minimizing quantization error and sensor accuracy error.

I decided to remove the back cover first to check if anything else needs to be removed. More 12mm nuts and T25 headed screws hold this on.

The yolk of this egg is beginning to show. Twenty-eight D cell modules, bound together by (non electrical) aluminum bars and plates.

I came around to the front again to disengage the service disconnect. Without this connector in, there is no voltage present on the contactor pack. I had to thread it through its panel hole to release the cells eventually, however. This also meant disconnecting the large ring terminals that fasten the battery to the contactor pack first – they’re at the very top, isolated by a plastic wall to prevent “mishaps”. A 14mm nut holds the terminals together, so I used a deep 14mm socket to extract them.

YEP, IT’S A BATTERY ALRIGHT

266 volts from non-isolated negative (before the contactor) to the service disconnect.

After I removed the contactor pack and service disconnect, I needed to remove the handles to gain access to the side plates. These handles are held on by some 12mm hex headed bolts that also retain the upper two sets of structural aluminum bars.

The bottom set of bars have giant Torx screws holding them in – T45 to be precise. I only had this size in a ratchet drive.

And both the side plates pop off…

There it is – the delicious Roe of Fusion at the center. But we’re not done yet – I still need to take apart the pack module by module.

The battery modules are retained by these non-electrical aluminum bars (I keep wanting to call them busbars, dammit!). So really, I could have just removed both top and bottom cover (and contactor pack), then only one side plate removal was needed. I could slide the whole thing around at this point.

Little temp probes that contact the cells at certain distributed locations to get a pack temperature reading.

Some intern had a lot of fun designing this engraving, I bet. This bar is totally something you’d hand off to an intern to do. It’s a straight piece of metal with holes in it! There is no way it’s fuckup-able!

Here’s how the pack comes apart. The moment you take the end module off, the whole pack is split into two, so you can’t even drop something across the primary positive and negative. Furthermore, you can only remove one module at a time: the busbar retainers interlock and are retained by two adjacent modules, so you must necessarily disconnect a whole module and remove it before moving onto the next set.

This construction is quite ingenious and must have taken weeks if not months of engineering committee meetings to come up with. I and everyone present at the first autopsy of this battery have said that it’s best example of design-for-assembly and design-for-service that we’ve seen.

I would never, ever have the patience to design such a system.

The back side of the battery is one solid strip of bus bar endcaps, removed one at a time at your leisure. The module endcaps are M5 threaded, and all the heads are T25. Someone really, really loves T25.

The full disassembly would look like a messier version of this – a small mountain of cell modules being slid out one at a time from the main pack.

The busbars live in snap-in mounts and are 1mm thick copper. On the “front side” of the battery, they simply are friction fit in place – they slide right out.

On the other side, they are retained by little snap fits. Unsnap them to take them out if you want to reuse them.

THIS IS THE BUSINESS END OF THE PACK. To avoid having to deal with this at 275 volts, take apart the pack starting from the right side of the contactor pack, if you’re looking straight at it. That’s the way I have it done here. If you start from the left side nearest to the contactor pack, you get to disassemble 275 volts first. Have fun.

If even one module is removed from the right side, this connection is isolated.

The sticks are made of 4.8 volt strings of D cells, with (I think) a nominal capacity of 8Ah. As you charge and discharge them faster, the useful energy will decrease – I put these on an aggressive 1 hour charge, so they won’t read 8Ah. One that I left overnight on a slow near-trickle ended up at 8250mAh.

Here they are inside. The cells are held off from their cases by rubber insulators; this is so air can flow through the case and around the cells, allowing them to cool or heat using the fan.

I may delve into the contactor pack in a future post. I think it’s quite useful stock and as-is, so I did not take it apart. After all, it’s just contactors with wiring already built in, and a precharge to boot – it doesn’t care if it’s being run at 275 or 27.5V – the coils are all 12 volts.

The contactors are nice – way better than the cheesy relays the students rig up remote-power circuits with in my class (but then again, that’s all they need). They’re Matsushita (Panasonic) type AEVS contactors designed specifically for EVs. It’s hydrogen filled!… wait, isn’t that bad?! Hindenburg Relays!

I’m sure if you had to buy these they’re a hundred dollars or more alone, but you get three of them with purchase of a cheesy old hybrid pack. Limited time offer.

Since I did not need this pack immediately, I decided to not take all the modules apart. Instead, I put the end plates back on so the whole thing is rigid, and it currently lives on a handtruck in the shop. I left the rightmost ‘crossing busbar’ out so there’s nothing 275 volts about it in public.

I made an almost-accurate model of the Fusion Sticks and have uploaded it on my reference page as a Parasolid (x_t) file, which should import into most CAD programs. The dimensions are accurate to 0.1mm for the first inch of length on both sides. There’s a lot of little randomness that sticks up in the middle, including the aluminum structrual bar channels, that I didn’t model since I was planning on mounting them on their ends only in my application. You could, if you’re crafty, just saw the bars in half and arrange the stock copper pieces to your liking on the cells, then redrill the end holes and use the stock endplate and handles.

Each module weighs 3.66 pounds, so this full pack of 28 is 100 pounds. And 2.1kWh! Beat that in lead.

So there we have it – another episode of Charles Takes Something Apart for His Own Amusement at Great Personal Expense Beyond Unboxing! I would actually love to do a series of this just on hybrid car packs, but it might have to wait until I find another one such as a Prius module or Leaf lithium ion module for super cheap (or someone comes and drops one on my desk). Mass produced commercial things always have the potential to be low cost and highly functional, and I think the gradual increase in market share of hybrid and electric cars is a boon for the hobbyist or home builder.

Beyond Unboxing: One Thousand and One Hobbyking Amps

Nov 07, 2013 in Beyond Unboxing, Reference Posts

And we’re back!

After passing the mother of all blogstones, I hope it’s had time to sink in with everyone, because it’s time to replace it on the front page. Just because I wander away from this site for a week doesn’t mean I haven’t been doing anything siteworthy. In fact now I’m once again in a situation where I need to backpost like 3 weeks after the fact. Luckily, this time, it’s all loopy engineering content! That doesn’t mean I’m done with the 2.00gokart coverage – it will hopefully appear on Make soon, and I might actually be wrapping it up in a more presentable style for a conference next year. We shall see.

Beyond Unboxing began with me taking apart a derpy Hobbyking controller. While it will not end with me taking apart a different Hobbyking controller (at least, I hope not…), if you haven’t gleaned from the title yet, this is once again about them!

No, Hobbyking hasn’t come out with a 1000 amp controller yet. That will be the day. What I got instead is five different “200 amp” controllers!

(more…)