Archive for the 'Project RazEr' Category


Snuffles Reloaded: Update ∫f(t)e^(iωt)dt

Apr 03, 2008 in Project Build Reports, Project RazEr

Oh man, it’s ALMOST THERE! Spring break was too short, or else I partied too much at the Media Lab. Either way, OH MAN, IT’S ALMOST THERE! Build pics from the past few days…

So, to make the motor endcaps, I had to perform some slightly odd machining. The endcaps are dish-shaped, which would have been simple had there not been a lump in the middle to house the axle bearings. In order to make this lump in a dish, I needed to plunge cut into the face of the dish by about 2 millimeters.

This was a simple enough task when I had access to an awesome shop full of precise machinery and tooling out the nose, since some careful boring bar work took care of it the first time.

Unfortunately, I only had my not-boring-bar. Despite my best attempts at grinding the proper angles to let it plunge cut, and trying to align the toolpost to my best ability, it still sucked ass. I decided to wait until the building was vacated to keep working, since although the tool did cut, it made a 9000+ decibel high-pitched squealing despite all attempts to quell it. So there’s a very good reason I waited until 4AM to finish these things…

And so above is a picture of the First Attempt. All I had to do was make a trench big enough to stick a real tool in, or use the Not Boring Bar like how it’s supposed to be used (which is how…?). The outside is decent, but as you can see, the inside surface is horrific.

A quick run-through with a pointy tool solved the problem, however. It turns (hehe) out I cut a bit shallow, but fortunately, the custom-ground HSS bit could face cut a few hundredths and clean the inside corners to boot.

Flipping the thing around to bore the bearing cavity. As usual, it was ghettocentered with additional assistance from a stack of milling parallels jammed between it and the chuck. This insured the stator against being completely off-axis from the can and blowing things up like it did on the previous wheelmotor’s first build attempt.

Done. This is what I call a “Loctite finish”, which is when the finish is so smooth and accurate that whatever is being mounted slips in with a modest push of the thumb. With some green Loctite (I prefer 609 ultra-thin retaining compound) in the mix, the part will never come off again.


(BTW, I lost the above image twice due to WordPress’ curious habit of making “undo” remove everything you’ve done for the past 5 minutes….)

And so with lessons learned from the first part, the second part was much, much smoother. I ground one of the 60 degree threading tools with a rake and side clearance angles so it could take the plunge smoothly. The procedure was to plunge cut, then move the tool in and out radially, making a wide slope-sided trench, feeding in more with each pass. When the sloped sided trench had been cut to the proper depth, the not-boring-bar was used to square off the corners.

The last step was to trim the diameters and cut the odd taper to mate with the scooter wheel. The diameter downsizing was smooth, the taper was not. There’s one downside to machining on low sleep reserve – you also have low common sense reserve. It would have been greatly smoother if I had set the compound slide at an angle and fed it in and out, like how you’re supposed to cut short tapers.

But instead I angled the tool itself and brute-forced it. Chattery results clearly shown. The dimensions and such are acceptable, but the finish is total bullshit. I slid some sandpaper over it to try and redeem myself, but decided it wasn’t worth the effort.

It’s on the inside of the wheel anyway….

Hey, it’s the 100th build picture of Reloaded! I have a habit now of documenting every little process involved with my projects. This is good, I suppose. Too bad the 100th picture can’t be of a finished product.

Instead, it’s of the magnet can. There will be 14 magnet poles (7 pairs), each of which is composed of either 2 or 4 mini-magnets.

Permanent magnets really love to 1) stick to eachother when you don’t want them to and 2) repel eachother when you want them to stick. Hence, I couldn’t back the mini-magnets up against each other. Instead, I had to place 14 “keystone magnets” first, at the proper angular displacements, and make sure they were firmly in place.

Thin CA wicked into the gaps between the magnet and the can held them in place well, and was also more convenient than making a mess with epoxy. I printed a 1:1 template from Gobrushless to help with spacing the magnets. When I’m happy with the magnet arrangement, I’ll lock them in with said epoxy.

The round aluminum piece in the background is the original “failed test piece” which I turned into an axial spacer for keeping the magnets the right distance from the ends of the can.

14 metamagnets composed of 2 mini-magnets are installed.

At this point, I wasn’t too sure whether or not to proceed. 28 magnets around the edges give me a near-perfect filling. Unfortunately, this is actually disadvantageous to an extent, as past a certain field strength, core losses start increasing and efficiency suffers. On the other hand, I stand to drop the voltage constant of the motor even further, making a slower, and hopefully torquier motor. But only 14 magnets gives rather poor filling. If these were, say, 10mm or 15mm wide magnets, I wouldn’t complain.

Intermission! I took the advantage of having my fingertips covered in a thin layer of CA glue to wind the core. Here’s one phase completed. The winding is “distributed LRK” style, with two-stranded 22 gauge wire looped 24 times around each pole.

Unfortunately, even with CA-shell assistance, I still do not have Manly Engineering Fingersâ„¢ and doing this took quite a bit of skin off both of them. The stator was mounted on something solid so I didn’t have to hold it, but pulling the wires tight did me in enough. I need to go find some gloves or something.

So I gave in, and shoved the rest of the 28 magnets remaining into the can. Bryan, you can kill me later. I promise. Just let me get one test run in.

I had 56 magnets exactly and couldn’t stand ruining my circle. How’s that for vanity?

Additionally, I decided to designate one of the endcaps as a permanent mount for the stator. Currently, the piece is on the radiator with the epoxy setting. Why?

If you have ever pulled the can off an outrunner motor (or yanked the parts of any permanent magnet motor apart), you know the magnets are very much attracted to the iron core of the rotor. On largeish outrunners, this force can be significant. On this fucker, it’s insane, and I’m going to make a jig to mount the stator without killing myself. The first time I brought the stator near, it nearly sheared off my finger because it flew so fast into the center of the magnet ring. The only way I could get it back out was pushing on the stator as hard as I can while gripping the can. And then it wouldn’t just fall out, because it kept getting sucked back in…

Hopefully, with some überpoxy holding the can in the Designated Endcap, I can bolt this endcap onto something, say the bottom end of an arbor press, attach the stator to the ram, and slowly drop it in.

This must be why permanent magnet motors aren’t made too large.

Anyways, stay tuned for the last few updates, which should be coming soon, assuming I grow new fingers by the time summer arrives. It’s almost moving!

Snuffles Reloaded: Update n(n-1)!/k^n

Mar 27, 2008 in Project Build Reports, Project RazEr

I forgot which update I’m on again, so pick a value of n and k of your liking which evaluates to the previous update + 1.

Finally got a chance to actually work on the wheelmotor… the days so far have been absorbed by the abrasive waterjet cutter at the Media Lab, among other things. I should be able to get some more work done soon.

Remember the aluminum biscuits I got a while back? Shouldn’t have gotten them supersize. The side plates are .393″ thick at the maximum and I got .75″ slices of aluminum. This meant I had to turn down almost half the length of each disc, which was sketchy enough considering the lathe’s limited spindle speed and feed options.  Worse, I still haven’t figured out how to activate the power feed for the cross slide, so it was hand cranking all the way…

I dug up the reversed jaws for the chuck in order to hold the 4 inch piece. This first facing cut was just slamming one bandsaw-finish side onto the chuck, since I need at least one flat, round surface to start.

The piece was then flipped over and spaced with milling parallels shoved between the chuck and the flat face. This gave me some reasonable accuracy and trueness despite the different setups (and the very frighteningly loose chuck jaws).

And violin, three trimmed biscuits. They’re both .393″ thick with a nonplanarity between the faces of at most .002 across the diameter, which is pretty good I think for ghettocentering. One piece was faced too far because of some strange force of machine nature which causes the tool to not cut until I crank the handle past my original target dimension, at which point it removes all material up to the new dimension at once. Oops.  The short piece is .385″, and has been designated as the practice piece.

Two pictures don’t quite convey the passage of time, but these three discs took two hours.

Because you can’t machine an outer diameter while holding onto the same outer diameter, I had to make my first metapart, a part which helps make parts. Dale clued me in on using a mandrel, or custom spindle mount, for machining the other features of each side plate.

This doohickey is made from 1″ aluminum round and has a .875″ section which the discs mount on. A large custom washer (not shown) fits the bore of the discs from the other side. The whole thing is retained by a 3/8-24 bolt. Like a giant inverse prop adaptor, almost.  Each disc has a center bore of .875 (or close enough to it) to mount snugly on the mandrel. The were, of course, made using the not-boring-bar.

Test cut! I threw the short disc onto the mandrel and tested everything. Seems to work great, and the runout is minimal once the mandrel itself is secured and centered. I called it a night there, as it was somehow 4AM.

Now that the raw outer-dimension parts for the side plates have been made, I can take another evening and actually make the final parts. More pics to come! Everything is actually round this time! Holy Robot Jesus!

I can has shenanigans?

Mar 26, 2008 in MIT & Boston, Project Build Reports, Project RazEr

So I fell back into a phase of “Oops, didn’t bring the camera” syndrome, but here’s some of the goings-on for the past two days. I’m pretty sure that my left arm is going to fall off on its own and beat me to death just by itself soon. There’s been alot more threading and tapping. It builds character….and muscle, since it allows me to try to equalize the arms a bit. Being right handed, my right arm has historically been a bit stronger than my left.

Anyways, onto pictures.

More progress on the steering arm, now with 99% more gears! They are all some ungodly large pitch (10? 8? Module 3?) and 3/8″ in thickness, and all waterjet-cut.

I have watched the pile of 50-pound bags of abrasive sitting outside the waterjet room slowly get smaller over the past week. I wonder how much the machine weighs with a full tank of water and sand?

A closeup of the geartrain. Yes, I know, two of the gears don’t touch and two more have very little contact. This was a design error that was corrected by recutting those gears.

The total ratio is around 40:1.

Random closeup through the Lexan mount. I suggested running the gear teeth in with polishing compound to smooth out the sandblast-like finish of the waterjet, which probably pitched the efficiency out the window. They haven’t taken me up on it….yet.

Switching gears a little bit, I did some more cutting and subsequent assembly work on this parking stand for the scooter. It does two things – allow the front wheel to lock against something for leverage while folding, and also allows the vehicle to rest vertically, leaning on the Wolverine-claw-like things (Which will actually have a plate mounted on top). It’s missing a few parts, but that will be addressed soon.
The big difference between this piece and the steering arm is that this has right angles and hence was easier to slam on the drill press and drill quickly and accurately.

Hmm, so all this fabbing has distracted me a bit from wheelmotor work. My magnets and bearings both arrived today.

Interesting thing about these bearings is that they’re tiny. 15mm bore, same as the previous, but only 24mm diameter and 5mm width. I’m slightly concerned about loading and bearing life. Whatever, I suppose. I should be able to get some work on the side plates done tomorow (today? what week is it!?.

28 magnets test loaded and…. HEY! What’s that giant gap?! It’s supposed to be a perfect fit!

Oh well. Nothing some index cards can’t solve. Did you know that a full circle of magnets like this can actually be detrimental to efficiency becuase of increased hysteresis losses in the core?

Stay tuned for the next episode. Meanwhile, bot on.

More shenanigans… and a Snuffles Reloaded update.

Mar 23, 2008 in MIT & Boston, Project Build Reports, Project RazEr, Stuff

Starting from now, I’m going to post mini-pics in 512 pixel-wide preview mode and then the linked pictures will be at 75% camera resolution. I realized that thumbnails suck, and a 512px preview didn’t do much for a 1024 pixel picture.

Anyways, the box of aluminum puzzle pieces turned into…

One of the steering arms for the 4-wheeler. The wheelmotor pod will sit on the end, and will be able to swing about 200 degrees or so. The steering motor mounts in the circular cutout, and there will be a reduction geartrain running through the arm itself (the Lexan plate is a placeholder, but may actually be used). The whole thing mounts to the chassis through the flat plates on the left.

Mating holes drilled and tapped. I took this one, and both regret it and am glad I did it. Some of the holes didn’t quite line up, and there were enough awkward angles that a drill press couldn’t be used. So it was hand drill, clamps, and alot of patience. The screws are temporary and will be replaced with hardcøre cap screws later on.

So now that I’m twice as buff because of all the hand-tapping, I should do the rest… hey, it’s Spring Break, after all. Gotta look good for the ladies.

Here’s a closeup of the steering gear. This is the waterjet on a good day (and on slow, ultra-fine finish mode). Even without a tilting head, it does great on 1/2″ aluminum.

Update: In fact, I like it SO much that I made part of it into another rotating site banner image. Hit refresh enough and you might get to see it.

And now for an intermission. After the great threading ordeal was over, I decided to head over to MITERS with my 4″ steel pipe and try to turn it into a 3.25″ steel pipe. I was a bit iffy about shoving such a large piece of stock onto the lathe, and tried a few ways. It ended up turning (teehee) out rather well.

First pass. Lots of chatter and earsplitting noise, due to two factors. One, I’m using the wrong bit for the job and the tip is far too angled. This problem was overcome by rotating the bit such that the left face was flush against the end of the pipe, which closer approximates how you’re supposed to do it. This solved the chatter issues.

However, the lathe has only 3 speeds: slow, medium, and fast. And slow is not slow enough to turn a 4″ steel pipe (surface speed-wise), so I was actually cutting almost twice as fast as I was supposed to. But it’s a carbide bit, which is supposed to allow a ~100% increase in machining speed…. so maybe it balances out.

Almost there. There’s still a bit of chatter, but most of it was overcome by a slow power feed, light cuts, and plenty of cutting fluid automatic transmission fluid mixed with differential gear oil. It took around 20 passes in all to reduce the diameter from 4″ to 3.245″. I probably could have done it in less, but wanted to tempt neither fate nor the robot gods.

During this time, I generated so much smoke from the cutting that when I finally turned around to look, the room was hazy. Naturally, not wanting to set off any alarms, I opened all the doors and windows as quickly as possible, and waited for the cloud to dissipate.

So, now that 10 years have been taken off my lifespan, the One Ring is done. Dimensionally, everything is within .003, which is freakin’ good enough. The ID, however, is a bit too large by about a quarter millimetre 0.01″. Which is actually beneficial, since that 0.2mm air gap was a really, really tight goal to hold anyway. I’d rather have some more space.

Now, onto the artifacts. For some reason, I find lathe-curl dynamics rather amusing, and have been in a contest with myself to generate the longest continuous lathe curl. My record with a 3″ aluminum round has so far been 25 feet, enough to stretch across the MITERS room and back a bit. The steel didn’t want to draw out as much, but was still cool to look at.

These were from boring out the ID of the pipe with my not-boring-bar. They would clump together inside the pipe, bob around, then when I pulled the tool back out, would roll out and fall onto the ground. They were nicknamed lathe hairballs by the other MITERers.

Here’s one of the ginorm-o-curls I’ve been looking for. Somehow, they find an island of stability in the clump that gathers by the bit, and then extend straight out and form very wide curves, which then proceed to get tangled in everything. I had to stop this cut because I couldn’t reach the levers any more. Without pushing aside razor-sharp, oil-covered smoking steel slivers, anyway, which doesn’t turn me on at all.

And now I present… the Mass.

This is the conglomeration of all the curls from all the OD-reducing passes from the night. Each pass resulted in a big clump of curls, so I just kept piling them on…. and on…. and on some more. Average 12oz WD-40 can for scale.

It has an enthralling iridescence that the camera flash drowns out. There’s all shades of colors, from purple to blue to blue-silver to plain silver and dull gray, but blue is predominant. Part of the reason, I think, is the sudden cooling of extremely hot steel slivers by the oil I dunked the whole thing in. I effectively heat-treated the curls. A process like this is used industrially (and more carefully…) to make hardened steels.

The only fitting fate for such a clump is, of course, as decoration. It now sits on the tool shelf, monitoring all in the room.

So, tomorrow (today!) is Sunday. Will I rest on Easter Sunday? Nah, not even that’s enough to get me to stop building things. The ML guys will probably be off, but I’ll be working on the side plates, hopefully. Magnets are due to arrive by Monday or Tuesday along with my bearings.

Something better move by the end of the week, or I’m going to be pissed

Wheelmotor Beta 2 Update 2

Mar 18, 2008 in Project Build Reports, Project RazEr, Stuff

In another episode of “Hmm, what is Charles building”, we have…

Gee, some aluminum rounds and a large section of steel tubing.

Hmm, it really makes you wonder, doesn’t it?

I opted to skip the Intro to über-basic EE that today’s 6.01 lecture was covering in order to haunt the Central Machine Shop for metal. The guys there are awesome, and cut me off a few pieces of aluminum round and this Seriously Hardcøre steel tubing. They also remember the MIT Battlebots team.
No, this is not a prototype weapon module, despite the temptations, but rather the ingredients to the wheelmotor. Obviously the motor doesn’t require such a Seriously Hardcøre outer shell, but it was the only thickness of tubing (Can something this thick be called tubing any more?) they had in the diameter I needed. At first I wanted to get the aluminum in one large chunk, but after some discussion, they cut it into slices since 1) They have a giant bandsaw and I don’t and 2) It was on the house.

The tubing, though, is likely Hardcøre-er  than the MITERS lathe can handle, even with the jaws reversed. I will definitely have to get into one of the main student shops with larger machines to turn it down.

In the mean time, I’ll work on winding the sucker. Here’s pics from a few days ago.

  Testing windings. On the right is 30 turns of single-strand #22 wire. On the left is 30 turns of double-strand #22 wire for an effective winding gauge of #19. I haven’t settled on a particular winding yet. That will come after some math, testing, and stretching the R/C hobby builder rules of thumb to the edge of existence. I will probably end up winding this motor alot milder than the last, and run high voltage.

One thing that attracts me to keeping the 30T of dual-strand #22 is that everything works out perfectly in terms of fit. I’m left with a little more than .025 clearance on both sides with the winding ends accounted for. That means I can wind the motor, then stuff the whole thing in a vise and squish down the windings some and not have to worry about wire clearance.

I may be able to stuff on more, but we’ll see.

Stay tuned! Spring break is approaching, which means I get to go party my head off somewhere in Florida lock myself in the shop and not come out until this sucker’s done!