Archive for the 'Project RazEr' Category


Wheelmotor Beta 2

Mar 12, 2008 in Project Build Reports, Project RazEr

So after the last wheelmotor build worked but failed miserably, I paused effort on it for a while as classes started again, TB needed prepping for Motorama ’08, and the Media Lab beckoned. However, with springtime approaching, I’m picking up the build where I left off.

There will be a few changes made to the design.

  1. Forget “minimal chassis modification needed to install”. Who the balls am I building this for anyway? A minimal chassis modification for me might as well be building a whole new back end.
  2. This allows me to use a wider motor. The diameter will not change much since it will still go inside a 5″ wheel. However, torque is proportional to motor length, all other things held constant, so I hope to get some more push from this next build.
  3. Also, I will use a real stator. Waterjetting .032 steel plates, coating them in wood lacquer, and slamming them on a hub doesn’t really count as having an iron core stator.
  4. The stator will have fewer, larger poles. Torque is pretty much determined by how many amps you can shove through how many windings, and for a motor of a given diameter and power rating, there’s only so much that adding more poles can do. Beta #1 was almost a stepper motor. I will never push either motor to its maximum electrical speed as dictated by the ESC anyway, so by using larger poles I can shove more windings in.
  5. The magnets will be ginormous. I settled on some 3.5mm thick,15mm x 8mm rectangular N50 magnets this time around. Each will be stacked end to end to give a single 30mm long metamagnet.

Hopefully the above combinations will make for a better drive motor. Just about anything will be better, actually, since eddy current losses were probably through the roof with my ghettorunner.

Some work pics over the past week or so:

The stator of choice was harvested from a large Kodak copier’s main drive motor. I actually stole this from said copier a while back when it was placed up for grabs as a freebie. I snuck into the building during the early morning hours, dismantled the copier, took out the motor, then put everything back together, leaving no trace.

70mm diameter and 35mm high, but unfortunately I cannot use all of it since there is no space (I don’t actually intend to rebuild the whole back end!). So it will be separated into two pieces, one 25mm for my use and one 10mm for wall decoration.

Rendering of the new motor. It is a full 46mm wide, which is almost twice as wide as Beta 1. The middle shaft also sticks out a bit for easier mounting. Also, the wires will no longer exit through the hollow center of the shaft. Instead, they will pass through a gap between the shaft and the bearing. This will make installation significantly easier.

The center shaft. This was made on the lathe using NEW TOOLING! WHEE! and then milled. The stator mount will end up flush with the end of the flat, and so the wires will be able to exit directly, without snaking through holes and the centers of screws.

For the next step of the process, I had to bore a precise hole. This was rather hard, since we don’t have a boring bar. I snaked around this problem by grinding one of the new carbide tipped bits down to approximate a boring bar. It actually ended up working, to my utter surprise, though there were rough spots.

Here’s the product of said boring bar, a press-fit hub for the stator. The three holes are to guide the power wires so they can be connected to the windings. This hub, in turn, will be pressed onto the shaft.

Stator stripped of wiring. This was an amusing process – loosen one wire, then drop the stator, letting the wire unwind as it flails in the air. Repeat 3 times. It almost resembled a fish flailing in the water resisting getting reeled.

Plastic endcaps removed from the stator. I tried my best to keep them intact, since having insulative caps would help the winding process immensely.

Next, I had to trim the 10mm off the stator. This was the most disaster-prone part of the evening, and required the most setup, but it actually went smoothly.

The stator, with hub pressed in, was mounted in a milling vise, sitting on parallels. With a NEW EDGE FINDER!!!!!!!! the center was located. Then, using an 1/8″ endmill, I mill/drilled one of the crimp-like things which held the stator together.  One by one, the laminations fell off as I drilled deeper.

I set the quill stop at the height of the hub, so the final product was a clean 25mm stator.

I think this is some of my best machine work yet.

The plastic endcaps were now a few millimeters too long. They’re also thin, floppy, but brittle at the same time, so that precluded most methods of trimming them (scissors, cutters, milling, etc…). I ended up holding them to a belt sander for a few seconds. Worked great, but a few bits still came off, so this side looks a bit grungy. The key part is that the stator arms are all insulated.

Shaft mounted in the hub, with plenty of Loctite for security and the fancy fruit-like smell. This whole thing is now one unit, like Beta 1. I thought about making removable mounts, but decided it wasn’t worth the effort.

The next step is to machine new side plates to accommodate the wider motor.  Then, I need to find a chunk of steel pipe to turn into the motor can, onto which magnets will be mounted. It turns out 28 of the magnets mentiond at the beginning, lined side by side, fill up the motor can to within a millimeter of circumference, so I’ll probably be winding the motor “LRK” style, with 14 magnet poles. Each magnet in this case is four small magnets.

Stay tuned!

Gyroscopic inertia is your friend.

Feb 02, 2008 in MIT & Boston, Project Build Reports, Project RazEr

When MITers get bored and there are spare bike frames lying around, weird things happen.

Had to crank up the brightness in Photoshop, otherwise nothing would really be visible. Yes, it was raining rather hard.

Oh, also, I gave Snuffles Reloaded its own project page, since it does exist.

It’s legit.

Jan 20, 2008 in Project Build Reports, Project RazEr

It moves. Finally, after k months, l complete redesigns, $m invested, and n setbacks, I have a working model of the in-wheel motor for Snuffles Reloaded. With the MIT Mystery Hunt now in full swing, I haven’t been logging the last few build hours over the past few days, so now it’s time to catch up.

The stator got a bit mangled as I tried to wrestle it onto its mount. No worries, a few dabs of epoxy here and there to secure the plates and a cuddling session with all 8 of my various off-duty clamps fixed that quickly. The heater it’s next to helps the epoxy set quicker and stronger.

Everything looks hardcore with a few flathead cap screws on it. For some reason, three of four area hardware stores did not have the cap screws I needed. The fourth had exactly 16. The motor needs exactly 16 screws.

Back at MIT, starting work on the magnet ring. There are 40 magnets and 8 mounting holes, so five magnets are spaced equally in the arc segment between two mounting holes. I placed these eight “keystone magnets” first and set them on top of the radiator to bake the epoxy for a while.

I also keep food warm on this same radiator, and if I need a quick chilled drink, I stick it outside the window.

Magring completed and epoxied. Index card stock folded twice over worked perfectly as a magnet spacer, so there are no odd gaps or areas where two magnets are unequally spaced. Constructing this was the messiest thing I had ever done – I think I absorbed more epoxy working on this than I have in all my life so far. Also, the stuff started to gunk up after 10 minutes or so (protip: if building motors, use some good quality 24hr epoxy). The magnets loved to leap off whatever I’m using to hold onto them and make a mess as they splat into the steel can. Adjacent magnets also like to stick to eachother, and the whole can would consume whatever tool I accidentally waved over it in the process.

If life were Autodesk Inventor, I could suppress the Magnetism attribute of the magnets as I mount them.

To the Media Lab shop! For some reason, I keep forgetting to bring my camera-camera to the ML and other off-dorm construction sites, and have to rely on the cell phone’s camera. Here’s some final machining prep to mount the motor to the scooter. The motor shaft center hole was threaded, the original wheel bolt holes on the Razor were drilled out (along with the wheelie bar), and the stator had its high spots and epoxy-gunked areas sanded flat.

At 4AM in the morning, the entire Media Lab building is empty. I was probably close to being the only person in the whole place, and couldn’t help but look over my shoulder and listen to every sound that did not originate from me.

Motor without windings test-fitted on the chassis. It’s almost there! It’s just missing wires! And a center hole in one of the cap screws. For this task, I went to MITERS and chucked a cap screw up in the lathe.

These cap screws are made of some serious, serious steel. I blew through two 1/4″ drill bits before damning it all and popping a 1/4″ endmill into the lathe tailstock chuck. Surprisingly, it devoured the steel capscrew very quickly. This leads me to conclude that endmills rock.

When freewheeling around, the cogging of the iron core upon the magnets creates a resonance in the tube chassis at a certain speed. It’s incredibly annoying at this speed – a rather high pitched siren noise. Although typical of brushless hobby motors, the rectangular thin-wall aluminum tube that forms the Razor chassis amplified the noise a whole bunch of times.

It had to go wrong somewhere, of course. I came too close to tempting the robot gods by being able to buy precisely the number of screws I needed. So of course when winding the first phase, I was off on the length of wire required by mere inches. So close, in fact, that I was only two loops away from being able to end that phase. If I had pulled the preceding coils tighter, I probably would have had enough!

This sucked. I had to cut off that coil, wind it individually, then join the two bits of magnet wire. Magnet wire does not solder well at all – I had to use a torch to burn the enamel off. Of course, the torch runs out of gas after one joint. I then had to resort to scraping the enamel off and waiting for the soldering iron to burn it off the rest of the way.

And then I couldn’t apply the heatshrink to insulate the joint because normally I use the torch at a distance for these things. It had to wait.
In the end, all was well. This is actually the back side of the stator. It’s actually quite neat and orderly, like good shrubbery. Five and a half turns per pole (probably two few) of twin-strand 21 gauge wire for a rough equivalent of 18 gauge windings. Real 18 gauge wire was too inflexible for my non-rugged not-engineer hands, which were pretty much bloody stumps after pulling hard on 30 coils of wire ten times each (that’s 300 good tugs not counting repeats).

Yeah. Here’s the other side. It’s a total clusterfuck since I couldn’t figure out how to overlap the coil transition windings and kept switching methods between coils. Surprisingly, everything rotates clear of the side plates. I’m very glad I left the extra .125″ of space on those.

But of course there has to be a catch to this somewhere. Yes, there is continuity to the chassis in the windings somewhere. The bane of hobby motor builders everything – improperly insulated stator combined with brittle magnet wire enamel make for shorts between phases. Again, if this were a real stator, it would have been dipped in an epoxy compound to remove all the sharp edges. Pop quiz’s motor is actually chassis-shorted also. I’m just going to run with this for now.

The proper thing to do is to strip the windings and start over, hopefully avoiding shorts. Fuck that, seriously. Fuck all of that.  It looks shiny. Can’t you guys cut me some slack!?

Closed up and tested for wire access. At this point, I was able to hook up a test rig and hold the motor using visegrips to spin it up. My large ESC was being nonresponsive, so naturally I used a small 25 amp controller I bought in China two years ago.

If you think Chinese parts suck nowadays, you probably thought they REALLY, REALLY sucked two years ago. You’d be correct. The controller was able to spin the motor up for about a second before something went poof.

I wasn’t sure of the amp draw since I didn’t have the RC meter hooked up, but it was surely not 25 amps.  Then again, the motor was a bit rough to turn by hand, so that might have caused a current burst long enough to explode something. Oh well.

I later found out that the big ESC was nonresponsive because the receiver battery supplying +5v was dead. It doesn’t have a BEC.

With the motor mounted in the scooter chassis and the wheel off the ground, it was time for a motor test!

The video of the first firing is here. MOV format, 3 megs. Yes, it’s  rather rough and the controller burps a few times. With subsequent runs and some loading, it (oddly) smoothed out. I never knew you could break in a brushless motor.

Getting the bundle of wires through the mounting holes at the same time as mounting the wheel was quite an adventure. I had to disassemble the wheelie bar, thread the bundle through, then through the chassis, then pop everything into place with a sharp whack. After that, the other side slides in normally. There is zero play in the whole assembly.

In the optimal design, the wheelie bar is replaced with a suspension assembly that makes the whole thing a module. No threading wires through too-tight spaces.

Observe. This is the first duct tape test rig I have built for anything in almost a year (Snuffles 1 didn’t move until everything was mounted in its final position). The ESC, battery, power analyzer, receiver battery, and about 10 alligator clips were all bundled up near the back and strapped in place with duct tape. Surprisingl, it held together long enough for me to get in a few test runs down the hallway.

The result? Dismal, as expected. Acceleration was extremely sluggish and only noticeable near full throttle. Peak current draw was over 50 amps  with steady current during acceleration of about 30-40. There were probably lots of bad contact areas and high resistance spots thrown into the mix.

At the end of a few heavily strained runs (the motor always sounded close to detonation) everything was surprisingly cool,  The motor will probably get perimeter cooling holes drilled in the side plates anyway, since I want it to be (somewhat) continuous duty. It’s good to see the duct tape rig didn’t short directly against the frame and set the building on fire.

In the morning I’ll do a concrete-asphalt run to confirm its suckiness, since the hall carpet probably added alot of drag. It is also awkward trying to hold the big Spektrum transmitter in one hand, the handlebars in another, move the (non-sprung) throttle stick, and stand on one foot. At once.

Hopefully I’ll be able to video that run. In the mean time, IT EXISTS! IT’S LEGITIMATE! IT MOVES kinda AND CAN CARRY ME marginally AND ROCKS! It’s also a prototype, rather rough around the edges, and can stand alot of improvement.

Verdict: Needs more engineering before becoming viable.

Stay tuned for more! Meanwhile, bot on.


Jan 13, 2008 in Bots, Project Build Reports, Project RazEr, Test Bot 4.5 SP1

It’s done! Structurally, anyways. And it’s really shiny. Besides being shiny, it actually stands a chance of working.

…because I’m fairly certain it has the same center of rotation throughout. If it does not, the discrepancy is too small to see or feel. The outer holes will be countersunk when I can get a sample of a 4-40 flathead cap screw.

Outer plates of 6061 aluminum, can of steel.

The insides. Getting the stator onto the internal hub was a bit of an adventure, and it got a bit mangled in the process. If I were to do this again with appropriate funding, the stator will be a professionally made part. I bet that even if it works, the efficiency will be horrendous since the stator laminations are “insulated” with wood lacquer and made of hot rolled sheet steel, the stuff usually used for odd sculptures, not the cores of high performance electric motors.

With a wheel. The only thing that wobbles now is this wheel, which is most likely a quality issue. The wobble isn’t much at all – a hundredth or two at most. What sort of annoys me is that the tire profile is so tall, which means the motor has to have a relatively small diameter compared to the whole wheel. If I could go completely balls-out custom on this, the motor would be much larger and the tire custom-cast from urethane with a lower profile.

The downside to a lower profile tire is less shock absorbtion, which could impact motor lifetime. That’s a matter of IRL testing.

The magnets and magnet wire are still back in my dorm room, so I’ll finish this motor up when I return.

I also took the opportunity to trim down this gear for TB4.5′s arm geartrain. It will drive the rear link through a pin.

Since I’ll be back on campus on Wednesday, I’m going to order some materials that will have parts cut out of them with various large machines. Some half-inch aluminum for the arm and frame bits, some quater-inch aluminum sheet for the wedges, Garolite for the top and bottom plates, and polycarbonate for the internal EBay components.

More random project updates

Jan 11, 2008 in Bots, Project Build Reports, Project RazEr, Test Bot 4.5 SP1

I tried to finish the wheelmotor side plates at the Almighty Dale’s yesterday. Keyword is tried – I thought I would be able to get both done in about three hours. The details don’t matter much, but in the end, I barely got one carved out and that one might need redoing.

The side plates have a little taper on the outer edges to mate with the tapered inner rim of the scooter wheel. It’s a tire bead, if you may call it that. The problem is, it’s a somewhat odd angle (about 40 degrees). So the toolpost on the lathe had to be readjusted to cut it.

Well, after cutting the taper, something was either missed on the realignment or something wasn’t tightened down enough, but the cutoff tool ended up cutting at an angle. That’s weird, it went in straight enough!

Either way, the side plate is slightly concave. The ring structure also indicates “oops, something got loose”.

There’s a 55% chance I might scrap this part too. The 45% “don’t scrap” comes from the fact that most of the dimensions are extremely close to the design – by about .002 or so.

It sucks that I might have to redo something that got this close. Next time, there will be a change of strategy regarding the final cutoff – I’ll probably part it a bit long, turn it around, face cut to the proper length, then cut the taper. Even if I have to do it with a file to not disturb the alignment of everything else!

Lesson? Don’t trust the machine to be perfectly aligned even if it looks like it. It’s a fallacy of luxury that something which looks and runs better requires less brainwork to operate.

Same can be said for soccer moms in plushy SUVs. Oh well.

In TB4.5MCESP1 news, more stuff came.

Drive motors! Since I was going to be running an elevated voltage this time around, I went looking for some slower motors. There are plenty of moderately hot-wound 550 motors on the surplus channels, but not as many with a Kv of around 1000 to 1500 – that is, 12000 to 18000 RPM at 12 volts. That was the “sweet spot” for TB this time around. Fortunately, AllElectronics came to the rescue with some surplus Power Wheels motors. You can never go wrong with Power Wheels motors.

What’s more fun, though? NEW CHARGER! Up until now I haven’t ever had a “real” charger – one that actually follows battery chemistry charging curves and wasn’t a dumb wall wart. The move to Lipolies sort of necessitates my upgrade, since you can’t plug a Li battery into a wall without some fun explosions.

The “good charger” did not get struck off the list, since I can use it for so much more than just TB. This should cover most of my electrochemical needs – 27 Ni cells, 8 Li of various subtypes, and even lead-acid.

The battery in question. This should fit snugly between the Ebay plates of TB. Some additional standoffs make sure it doesn’t fly around inside the bot. TB should never need the full 25C of this pack – if it is drawing 80 amps for any reason, something has gone terribly wrong.

I intend to make an “integrated charge connector” which combines the balancing port with the main battery lines. This way I don’t have to have 2 cables coming out of the bot.

What I particularly love about United Hobbies / Hobbycity is that they make absolutely no attempt to hide the fact that yes, everything on the site is Chinese in some way. This includes full broken English manuals, which I find quite funny. Often, they express things in such a manner that you get what is being conveyed, but wouldn’t ever find it in a real English manual.

Check out my awesome quad-reflection. I mean, come on, what charger besides the most kickass and awesome has a WASTE TIME function?!

Anyways, it’s currently working on the 4-cell Lipoly and I have the new cutting fence set up on the miter saw, so I should get some chassis work done before I have to fly back to the Northern Wastelands.

Oh, and Jesus welcomes you.