LOLrioKart Update 11: [/drivetrain]

I got the last drivetrain parts on Tuesday afternoon. So, I guess I’m proud to announce that the power transmission side of LOLrioKart is complete!

With this, work is shifting onto the electrical system and last touches. What does this mean? You’ll probably not hear about this project for the next 8 months, of course, if the scooter build was any indication. However, the sheer level of ridiculousness this project has attained so far, coupled with my lack of usual robot projects, means I’ll most likely keep moving quickly.

Here’s some drivetrain assembly pics.

Disassembled drivetrain awaiting finalized parts

I tore down all the temporary mounting provisions to await the final parts and hardware. The bolts that temporarily retained the wheels (so we could putz around the hallways under human power) have been swapped with the actual half-shafts. I bought some 3.5″ long socket head cap screws to mount the Etek through one of the transaxle side plates. The only thing I needed to make were longer shafts for the differential so I could use shaft couplers to link them to the wheels.

New half-shafts installed in the planetary differential

Making the longer differential shafts mostly involved taking the center shaft section left over after making the halfshafts , splitting that in two, and cutting a retaining ring groove for each half.

Unfortunately, I couldn’t find my Convenient Lathe Bit of Groove Cutting (+1?), so I had to grind one from a blank.With no proper tool grinding provisions, that was quite an interesting feat.

The shafts now poke out another inch, enough to secure the shaft couplers.

Fully assembled rear transaxle, minus chain.

Like so. With the “pacmen” secured and all collars, set screws, and whatnot tightened down, the assembly is rock solid.

The couplers are simple 3/4″ bore keyed-plus-set-screw jobbies from McMaster. A long key connects both shafts together. So, under optimal conditions, it is the keyway (not the coupler itself) that will transmit the torque. The coupler’s just there for moral support.

In a similar fashion, in a properly designed bolted joint (Gee, how many times do I actually design my bolted joints?), it’s the compressive friction force between the pieces being tightened together that transmits the load – not the bolts themselves, which are only there to provide the compressive force.

Drive chain spliced and mounted.

And the chain is mounted.

It’s a single-strand ANSI #40 setup. I didn’t expect that mounting this would be so straightforward, give that I had two odd-toothed sprockets at a non-integer inch center distance. But, to my delight, the chain length required is indeed extremely close to a whole number of links. It’s “droopy”, but not loose.

After the chain stretches from breaking in, it will probably be more on the “loose” side of things.  In a pinch, I can route the chain over one of the Etek’s lower mounting spacers, and drop a bearing or idler sprocket on said spacer.

So that rounds out the drivetrain. Well, a drivetrain isn’t any good without a power source, so onto the batteries…

Giant nickel cadmium batteries!

Here they are again, the giant nicads of last year. I have no real facilities for taking care of batteries this large. At the same time, I figured batteries this large could take quite the beating before permanent damage occured.

It’s not like they’re healthy after sitting for two or three years before sitting for 8 months, anyway.

To revive the packs again, I first applied the voltmeter on each cell. Most of the cells in each pack still showed a reading above 1.1 volts. Others, however, were pretty much zeroed.

Nickel cadmium batteries, when left sitting a long time, like to grow tiny crystalline filaments of nickel within themselves which cross the electrolyte and separating layers and poke the other electrode. This, of course, internally shorts the cell, causing rapid discharge to zero volts.

To wake those cells up again, there is a scientific process known as zapping, which uses a high voltage capacitor to momentarily dump a large current into the cell. The filaments vaporize, and the cells can be cycled again to get rid of them.

Of course, scientific processes I had not access to, so I made do with a large lead acid battery and some meaty alligator leads. A very short ‘blip’ on each cell released a shower of sparks and put some life into the cell. It was then immediately put on my peak charger to bring it to 1.2-1.3 volts.

This process took a while, since each cell had to be measured, zapped if necessary, then charged after zapping. I think I’ve been babysitting the batteries on and off for the past week.

Eventually, however, all of the cells were at a level close enough together to charge their entire respective  packs at once. This just involved setting my charger to 7 amps, its maximum current capacity, then leaving the whole thing overnight.

Yes, the building is still standing. I’d come back and see the charger stopped somewhere around 400 to 450 minutes, after the amp-hour counter had fully rolled over. This calculates out to ~40AH put into each pack. My guess is that alot of the charging current went into slightly overcharging the good cells while the weak cells caught up.

For the past two days(!), all4 packs have  been wired in parallel and the whole thing charged at 5 amps, to attempt to equalize all the packs. I had to turn off all the limits on my charger for it to run that long. The 3-digit minutes counter has rolled over several times.  Each pack is getting about 1.25 amps at this charge rate…which is like C/30. They will probably never peak.

I have no facilities for testing the discharge of cells this large, so I’m just going to lob them on the kart and drive around with a voltmeter. For now, I’ll assume they’re all in decent condition.

What’s next? Oh, designing that battery + electronics mount that I never really got to!

Test-placing the giant nickel cadmium batteries!

I’m a bit stymied by battery layout. There’s several configurations I can use, each with their upsides and downsides.

This 4-in-a-row layout gives the best possible center of gravity placement vertically, but not horizontally. It puts alot more weight on the rear wheels. With me in the basket, it will be even further back.This means I stand a much higher chance of wheelying instead of launching, and it could affect steering also (heavily rear-weight-biased vehicles tend to suffer from understeer).

Since LOLrioKart is not a real car, I’m not as bothered by this fact. The cheap handcart rims will probably bend first.

Alternate placements include a 2-by-2 layout – that is, two packs stacked on top of eachother, then two of these metapacks side-by-side. Then I can place them either transversely (as pictured) or longitudinally (down the middle). The disadvantage of any ‘stack’ packs is that I don’t have immediate access to all the battery terminals, something that I think I’ll need because of the age of the cells and their demonstrated voltage instability.

Yet another placement is a “3-by-1” T layout. I can’t fit all 4 packs longitudinally side-by-side, but I can fit 3. Then I can have a single one at the back, mounted transversely. This actually gives the most centered (good!) layout of the batteries, but I’ll have to make a T-shaped mounting basket for them, which makes that issue more complicated.

In the end, I think I’ll just go with the 4-in-a-row. It’s the simplest to design a mount for. I can build a wheelie bar or add something else up front to compensate. I have access to all the cells (electronics and wiring can actually go on the back side of the basket to keep this space clear).

Measuring the allowed clearance between motor and battery

Another advantage of the “3 by 1″ layout is the ability to shift the whole pack forward, past the motor. Right now, I can only elevate the packs off the ground by about 2.75 inches before I hit the motor mount. This means at most, I can have 2.75 inches of ground clearance – throw in 1/4″ for the mounting provisions, and I can probably have 2.5″ of ground clearance, maximum.

Go-karts tend to be pretty low machines, so this may actually be acceptable. After all the sprocket in the back is 6″ in diameter, on 9” wheels – giving a maximum clearance at the back of 1.5 inches. The front steering brackets hang pretty low also.

But if I ever suffer from ground clearance issues, it’s always an excuse to move to bigger wheels…

Stay tuned for the next episode of Charles plays with large batteries in an inappropriate fashion!

LOLrioKart update 10: Everybody to the Back

With the front half of LOLrioKart done for now, it was time to work on the powertrain. I don’t have a plan yet for the pedals, levers, knobs, and switches in the front. This includes the all-important front brakes, which I have to figure cable logistics for. I will return to them after all four wheels are connected to something that is (or will be) controllable.

Since UPS seems to have taken off early today, I wasn’t able to get the parts needed to fully finish everything that I wanted. Such a shame, because with a few hours of rigging, I could have wired the motor up and went on a (linear, fast, most likely very short) test run.

Let’s begin.

Measuring out the back end

With my new 12 inch digital calipers!!!!, I measured the center distances between three critical points. First, from the rear axle to the shopping cart’s rear horizontal undercarriage tube.. Second, from said frame tube to the bottom rear basket support tube. And lastly, from said rear basket support tube to the rear axle.

Since I knew the side lengths of this triangle, all the necessary angles could be solved for.

Here’s a cool trick if you ever need to find the center distance between two rods or two holes, and don’t have a convenient handy-dandy centerline gauge handy.

Measure straight around the outside of the rods, or the farthest points on the bores.  Measure again between the insides of the rods (with the ID side of your caliper), or the closest-together points on the bores. Add these two distances and divide by two.

Sure, experienced machinists are probably giggling over this, but it was one of those COOOOOOOOOOOOOOOOL I LERNED SUMFIN moment for me. No more eyeballing.

With critical dimensions determined, it was time to buy a few Jolts and 5am-engineer something (Like I resolved never to do again.)

Wooden laser-cut prototype

The proper thing to do with any new design, especially one of questionable workability, is to prototype and proof the concept. Prototyping allows a more concrete observation, and subsequent interpretation, of your idea. Usually, prototyping is done with materials and tools immediately available.

Luckily for me, that happened to be leftover 1/4″ MDF and a giant lazer. I whipped up this 1:1 scale mockup of the motor-and-differential mount to see if my dimensions were actually correct. I did the same thing for NK5 before Dragon Con. It’s handy to prototype with a low-valued material, so you don’t bust money setting a flawed design in stone….

MDF to aluminum like boyz II men.

… or 1/2″ 6061 aluminum plate. After making sure the Etek actually sat in the mount and the mount actually sat on the kart (and by extension, the Etek on the kart), I headed back to the Media Lab with a half inch aluminum plate.

Only around MIT would you see someone sprinting down the street with a 3 foot long slab of metal. Nobody was decapitated in the process. An hour (40 minutes of which was trying to jiggle the 2D outline file such that it would actually cut properly – I blame the 5-year-old layout software) later, I had the semi-final, waterjet-cut parts. Semi-final because they still needed…

Putting parts next to eachother to test the fit

Threading, boring, and related secondary operations. Here’s everything thrown together as a visual. I used the boring head on the mill to clean up the bearing pockets. Some mounting holes were threaded 3/8″-16.

Also, I had to toss the rear mount back on the mill to add and/or lengthen some cutouts because I miscalculated the placement of some Etek features by a few degrees of rotation. Too lazy to drag the 200+ pound full size rotary table onto the mill, and not having a big enough head for the dividing fixture, I just manually interpolated a circular pocket.

It worked pretty okay.

Okay, so what’s the deal with those weird cutouts right next to the motor? I’m not seriously hanging a 25 pound high performance motor capable of gear-stripping, chain-breaking, shaft-shearing torque with a high-inertia steel disc rotor from a narrow cross-section of a material with a finite stress cycle life, am I?

Test assembling the transaxle.

That’s where those little Pac-man things come in. Here’s the whole transaxle test-assembled with hardware I found lying around. The Etek is indeed hanging on by one screw.

It is a well known fact that two objects cannot physically occupy the same volume in three dimensions. Since we happen to live in a (mostly) 3D world, I couldn’t actually install this transaxle onto the kart if the cutout wasn’t present (i.e. there was only a hole the diameter of the frame tubing) because it would involve passing it through the tube first.

Life isn’t Autodesk Inventor. So, the cutout slips over the tubing, and Pacman closes the gap from the other side. The two pieces are securely screwed together with 3/8″ cap screws. A circle of the same diameter as the frame tubing, 1″, is left in the middle. I will axially secure the transaxle on the frame tube. with some big 1″ two-piece shaft collars.

Test mounting the test-assembled transaxle.

And it’s mounted.

This whole assembly is some Serious Fucking Metalâ„¢. It weighs in somewhere north of 35 pounds, more once I make new shafts for the differential (I didn’t expect to be building something around it – oopsie) and add real hardware (including giant steel standoffs for the Etek).

The observant will notice there’s nothing attaching this assembly to the basket mounting tube. It ended up that the basket itself was welded to this tubing with enough overhang that I could not easily use it to actually mount anything pointing downwards.

Unfortunately, this meant that the rear axle itself had to become a stress-carrying member, or else the transaxle would have no means to react against moment loads (torque). Having your motor spin freely around a mounting point is no good, since that means your wheels don’t spin.

I figured loading the axle like this was okay, since no less than ten ball bearings support the axle: four in the differential, two in the metal frame, two in each outer bearing block.

Granted, they’re cheap-ass ball bearings.

Despite not being joined in the middle, the axle should be just stiff enough after the addition of two giant keyed/clamp shaft couplers to give a bit of “bounce” compliance to the motor – beneficial – but still transmit power and keep the gears straight.

Give me some slack. This isn’t going into space.

Not this version, anyway.

Side view of LOLrioKart with mounted transaxle

So here’s the profile of LOLriokart with transaxle installed.

Wait, couldn’t I have put the motor under the rear frame tube? It would drop my center of gravity!

Yeah, by maybe an inch. If I had put the motor closer aligned with the bottom of the frame, I would have no place to mount the batteries. The batteries combined weigh around 80 pounds, and I want them as low in the frame as possible. Currently, they fit widthwise, four packs, lined front-to-back, between the side tubes.

So what can I do without my shaft couplers, 3/4″ shaft stock, motor mounting spacers, proper-length socket-head cap screws, and other things that UPS didn’t get to me today?

BATTERY MOUNTING TRAY!

The plan is to whip up a cage that drops slightly under the bottom of the tube frame and that mounts to said tube frame. If I’m crafty, I’ll also work in a flat mounting surface for the electronics, right above the batteries.

The battery packs will need to be woken up after 8 more months of slumber (one metered in at an amazing 3 volts!).

This is the last weekend of IAP. Next week, I’ll have to start worrying about all that “preparing for the semester” bullshit.

And the week after that, 2.007 starts.