While you were away…

Wednesday, March 4th, 2009 @ 5:32 | Project Build Reports, Project RazEr

I finished the rebuild of Project RazEr. Because the site has been down, I never wrote up any day-by-day build reports, but still took pictures accordingly.

So, here’s an epicly long build report all the way from start to finish. The whole process wasn’t long and drawn out this time, mostly because the fabricated components were conceptually simple. The majority of work was done in a few evenings at MITERS (as usual.)

Here’s the plan. Unlike version 1 of RazEr, the module is one-piece, not a weird jumble shoved into the tube frame of the scooter and an add-on pack underneath. This time, I intended to cut out the entire bottom of the scooter in order to fit everything in.

Benefit: A vast increase in usable, contiguous space. That’s the biggest difference. While I only gain maybe 2 cubic inches from version 1 (in the form of the aluminum bottom of the scooter), that 2 cubic inches of metal divided the battery pack, electronics, and everything else from eachother. Losing this barrier meant I could use bigger battery packs for MOAR P0WAR!. Not only that, but I could use stock R/C hobby lithium polymer batteries.

Like this 5AH, 4S1P battery from UnitedKingoftheHobbyCity (or whatever the hell they call themselves now). I got two and intend to run them in series for 5AH, 29.6v nominal – or almost 150 watt hours of battery power, quite substantial for something this small.

One pain with using Li chemistry batteries is that they require balancing. Unlike Nickel chemistry batteries, you can’t just let a Li cell bake off an overcharge while weak cells in the pack catch up. Usually, that involves catching fire.

Having more contiguous cubic inches in the design meant that I could use more real electronics, such as this 5S balancer module. The plan is to mount two inside, each balancing a single 4S pack. There was no danger of ground-looping a pack because these balancers have no other physical connection to the world.

Alright, so let’s see how everything fits. So far, so good. It’s hard to get a sense of physical size on a 2D computer screen. When am I going to get fully-immersive 3D holographic CAD programs?!

This is “The Stack” of electronics. Of course, the end result will be more organized, and use actual mounting screws, standoffs, and the like.

The 100A 42V brushless controller is on the bottom – the intention is to strap the MOSFET side directly to the thick aluminum for heatsinking. On top of that go the balancer cards. On top of that still is the 5V switching regulator that will power all the logic (no more linear regulator dropping 35 volts to 5!). Barely visible, between the balancer cards, is a servo tester. I decided to just buy one since they make them smaller, better, and more durable than I ever can – come on, it fits *between* the two balancer cards.

Alright, enough visualizing. Let’s begin.

What happens when you need 18 inches of a 3″ square aluminum tube? You buy 6 feet, because McMaster-Carr says so. I had to split the enormous 6 foot stick into two three foot sticks, then cut the 18 inches I needed off that.

So I lied – I only needed half of that 18 inches.

Except it’s half in that other way. With  the new SLITTING SAW!!!!! I was able to take a clean cut down the longitudinal plane to separate the tube into 2 U-channels.Two little clamp fingers set into the T-slot and dialed in provided a flat stop to clamp the tubing against.

18 inches is a long way to crank a mill manually – this and many other operations were facilitated by the mill power feed.

Wait, what power feed? Charles, the MITERS Bridgeport doesn’t have an X-axis table feed.

No, but I have a cordless drill with a socket wrench that fits the handle nut. Worked beautifully.

Finishing the U-channel sides to proper height. I dug up all the random clampy things we had for these operations – I think we need more.

After making a bunch more profile-shaping cuts and drilling holes, it’s time to throw it on the body as a test!

Note that the wheelmotor is missing at this point. I still needed some form of transportation – walking just seems so inefficient after putzing around on wheels – and the hub motor, while smooth, still added massive drag compared to a plain skate wheel. So I dismounted the drive motor and threw in a scooter wheel while construction was going on.

As can be plainly observed, I have a few pebbles worth of ground clearance. Surprisingly, it’s still more than the average 90/100mm-wheeled Razor scooter.

Here’s one result of planning ahead and designing things before building them. This is the front endcap of the module (the rear is similar, just without the holes). Eventually, when everything is mounted, I’ll go around the edges with silicone sealant and make this sucker splash-resistant.

The “window” is laser-cut acrylic, done on the Media Lab’s GIANT LAZER.

Throwing everything into the unfinished module. Hmm, space is a little tighter than I had imagined up int he front.

Well, the important bits fit, so it’s time to commit the changes to the scooter body.

I set the aluminum extrusion frame of the Razor scooter up in the mill and carved out the entire bottom. After this, the frame is no longer structural, and the module itself has to be attached with more security since it has to take a load. That was my concern with switching to the current design. We’ll see how it goes, I suppose.

The no-longer-tube frame cleaned up. I guess now it’s more of a double-T shape, or a U-channel with a fat bottom end.

Let’s see how the back end fits. Verdict: Pretty well. I was afraid of misjudging the height of the interior as well as that radius at the interior top of the frame, but it seems to be correct.

Nothing a giant bead of sealant can’t remedy.

So, the bounding dimensions are correct. Time to add the details!

Milling the angle at the front. I could have left it square and gotten a little more space, but the angle complements the existing angle on the scooter frame, and doesn’t make the whole thing look like a box.

Also, I drilled/milled holes and slots for the eventual battery plugs.

Test fitting after milling the angle.

So what are those five little holes in the front for?  Well, I figured as long as I was making a cap for the module, I might as well add some spiffiness to the whole thing. So, the five holes are for superbright white LEDs, which cast a low-angle light in front of the vehicle to act as a sort of headlight.

The entire physically finished frame. Essentially, this module is an extension of the internal cavity of the scooter. If it weren’t for my obnoxios bad-setup-induced chatter marks, it would almost look like it came this way.

Now that I was mounting a bunch of electronics right under the folding joint, the large conductive cylindrical protrusions (otherwise known as screws) had to go. Simple – attach the joint, then cut off the screws flush with the mounting plate so they don’t stick out any further from the bottom.

Making a precisely (okay, semi-precisely) engineered frame extension module for a scooter may be easy enough, but now how am I going to securely attach it to the original frame? The plan called for simple set screw pressure (the six holes in the sides), but I pretty much knew from the start this wasn’t going to hold anything.

Set screws by themselves exert strong point pressures, but this can easily exceed the yield strength of the metal and cause epic joint strength fail because the connection point simple deforms away from the screw.

This is why removing anything from a shaft that has a slipped-set-screw ring cut into it is horrible and usually involves a giant press or grinding wheel.

How to overcome this? While reading about how to adjust the cross-slide gibs on the MITERS lathe, I realized that if I slipped a soft piece of into the gap (between my module sidewall and the original scooter frame), and then applied the set screw pressure to that, the pressure gets distributed by the soft metal into a friction force that covers a wide area, binding the module to the frame.

This is the same mechanism that some small hobby lathes and mills adjust their slide gibs with, except, of course, you don’t want to go as far as to bind the sliding surfaces together.

So notice the strip of aluminum crammed into the gap. When all six set screws are tightened against it, the whole assembly is rock solid.

I christen this the “Ludicrous Gibs” mounting method for attaching two channel-shaped objects together. ROTT was a great game, by the way.

Wiring work begins. Note the neat laser-cut Deans connector mounts. Each one snugly holds a Deans female plug, aided by a drop of thin CA glue. In turn, they are screwed into the side of the channel.

The left plug is the battery, and the right plug leads to the controller and other electronics. This will be bridged by a removable power link, like how I make most of my robots’ main power switches.

It’s kind of silly to try and aim a plug into a socket as an EV switch, but I couldn’t find a switch of the proper rating (or form factor) and invented this idea in a few minutes. However, the benefit is that I have easy access to both sides – battery OR vehicle, should I want to run experiments or fit a power meter between them to take runtime data.

Dropping the components in… There’s no turning back now.

Looking pretty good so far. After mounting the batteries (read: tacking them down with foamy double-sided tape), I performed a quick fit check.

Again, pretend the cruddy chatter marks don’t exist. That was the result of “Hmm, I think clamping on a 1/8″ thick section of material and then face-cutting 2 inches away is a good idea”.

In a moment of disappointment, I discover that plugs do indeed have a real volume – not complex, nor purely imaginary.

Instead of re-engineering everything to fit in the balancer cards, I decided to go dig up a 9-pin connector of some sort to let me access the cells individually from the outside.

It ended up that the battery balancer cards weren’t going to work anyway. They begin working whenever they are connected to a battery, and are always on, even if not balancing. Therein lay the problem – I couldn’t just reach in and turn them off. That means once connected, they will always be draining the battery with the minuscule current needed to run themselves.

It isn’t much, but is enough to discourage me.

I found this neat 9-in single-row header in a box of cables. A .1″ male header row fits it, so I was in luck if I need to make a cable.

I did have to make a slot to mount it, however. So I threw the module onto the mill with giant, live lithium batteries strapped in and cut the slot.

As to not cause heart palpitations in safety freaks, I’ll refrain from posting the picture.

Anyways, magic happens and it works. Here’s the finished shot! No external electronics this time – the box mounted on the steering column is empty. And I even have a real thumb throttle.

The front is covered in electrical tape because I didn’t want wheel grunge fouling up the shiny new acrylic window. Since this picture, it’s been pretty well dirtied up.

Weigh gain over version 1 is approximately 1.5 pounds, mostly in form of SERIOUS FUCKING \M/ETAL from the aluminum module frame, and slightly larger batteries.

Still light, however. The lightest around, as far as I know.

The question that everybody asks after “What the hell is THAT?” is “How fast does it go?”. I’ll need camerafolk to answer this, but the answer is “At least 15MPH on flat ground and fresh batteries”. It’s hard to gauge speed when you’re on it, just like 60mph  seems much slower in a car than standing on the side of the road.

However with a properly calibrated controller (such that it recognized me not as a stuck motor), I metered 1.2kW on good launches. The wheelmotor actually has substantial torque – just from the feel, I would rate it around half that of Snuffles the First, i.e. “it will fly out from under you if you’re not paying attention”, which is what happened to me the first time after calibrating the ESC.

I need to actually to “back-run” the numbers here to see if they match up with my initial predictions before building the motor.

So that wraps it up. Now that the power electronics side of the project is done, I need to rebuild the wheelmotor, which is falling apart day-by-day and is only running at the moment because I pumped all the empty space between the tire and rim with hot glue.

Yeah, I hot glued my motor together.

And here’s a shot of the headlight working.

i can haz 5-minute cross-campus commutez back plz?



  • More About the #RobotTrapShop and Building Up a New Workspace
  • A New Beginning: The 17th Chapter; Back to the A-Town
  • Robot Ruckus at Orlando Maker Faire: How to Somewhat Scale-Model Test Your BattleBots
  • Überclocker 5: Finishing Up The Everything Else
  • Überclocker 5.0: In Which I Actually Have to Build the Bot, Not Just Talk About It
  • Überclocker 5.0: The Big Post of Designy-Stuff
  • The Overhaul of the Future Begins Now: Überclocker 5.0 (Also, Welcome Back to Robots)
  • Operation RESTORING BROWN Part 7: The Epilogue; or, Dragon Con 2019
  • Operation RESTORING BROWN Episode VI: Return of the Van Lights; the Conclusion
  • Late Stage #PostmodernRobotics: Welcome to Your Waifu is Trash, the Robot Dumpster Fire

    Comments are closed.