Archive for April, 2008

 

Jet-powered doohickey? You bet.

Apr 30, 2008 in Project Build Reports, Stuff

I have found that I’m nearing the limit of what electric fast spinny things can provide. Five years of standing arenaside while watching fast electric spinny bits smash things, and some times even operating one myself to do the same, has kinetically jaded me some. I will spin up, stand next to, and touch many things which my peers would not. This is not always a good thing, of course.

But anyway, the spooling of brushless motors can only give you so much excitement. What I have found far more exciting is the spooling of turbojet engines. I always pick the rear seat on a short flight not because I’m cheap (I really am not), but because the “precisely engineered array of pointy things at extremely high speeds” sound is relaxing and spiritually refreshing. Most people would say otherwise.

Okay, so I lied. The cheap seats are usually in the back.

Anyways, stumbling upon home-made jet engine & gas turbine videos on Youtube like this and this and this and this and… uh, this have convinced me that it is my destiny to build a simple jet-powered implement. I’ve seen the videos before, but that was back when electric spinny-things were more exciting. At the very least I’ll investigate a turbocharger-based ghettojet, since pulsejets, despite being loud and obnoxious, just don’t have that high-speed spinny effect I’m going for.

I’ll probably make one for static testing, then consider mounting it on something. So that begs the question to be asked, WHAT DO I STICK IT ON?!

Scooter?

Bicycle?

…Unicycle?

Shopping cart? I already have plans for a 50HP electric shopping cart in the works, so why not have a companion?

Go-kart?

Blender?

Hmm.

A flat motor.

Apr 27, 2008 in Bots, Pop Quiz 2, Project Build Reports

I got some flat motors the other day.

They’re seriously flat motors.

For a really flat robot.

It’s been a while since I’ve updated Pop Quiz, which has been in much the same form since the end of 2005. That makes it my longest-lived bot ever!

For revision 2, the blade height will be dropped from 0.8″ to a bit over 0.6″. The chassis height will be lowered from 0.5″ to 0.375″.

This requires extensive customization on almost all levels. The only stock gearmotors I can use are Sanyo micromotors, which are about .380″ tall, but unless combined with an additional gearup on the output, will not offer satisfactory speed.

So I’m going a different way.

This is a custom 3:1 gearbox design for a Mabuchi slot car motor, the square kind. I got a bag of 30 of these surplus a little while back.

3:1 straight off these motors at 14 volts will make the bot travel over 9000 miles per hour. However, I plan to inject some big N50 magnets in place of the ceramic ones in there now, greatly increasing the magnetic flux within the core and hopefully dropping the voltage constant enough such that 3:1 is actually sort of manageable, especially with wheels 0.5″ in diameter. The target speed is about 4 feet a second, which is incredibly zippy, but only if it works out. Extensive testing of this setup will occur.

The gears are metric module 0.4 things from SDP-SI, which is about 64 pitch. Aligning them will be such a royal pain in the ass that I don’t want to think about it at the moment.

The housing is a chunk of aluminum channel or rectangular tube, machined to taste. Four modules will be made for 4 wheel drive. Each module should be around 2/3 ounce according to Inventor.

Wepon motor in place. Oddly enough, the version of PQ’s motor that worked unquestionably the best was the original hacked hard drive motor with its 2504 stator and ball bearings! The later custom motors just never quite matched up to it. So I’ll be using a HDD stator again in a custom housing, with ball bearings – no more of this bushing crap that plagued the current weapon motor.

The axial flux one had ball bearings, but didn’t fare any better because its torque output was so low.

The flat motor stators in the picture are 28mm diameter x 3mm tall, but only 9-poled. I have another set of HDD motors which have 28mm x 4mm stators with 12 poles, allowing the use of a dLRK winding, which have been designed into the above motor.

This is, of course, the fun part. Ground-to-blade height of a hair over .625″. It could be less, but running the blade to close to the body results in self-eating.

The top of the frame is exactly half an inch from the ground. The ground clearance around the wheels is a bit less than 1/8″.

Overall, the parts rundown is:

  • Banebots 3-9 controllers for drive. The Scorpion Minis inside Pop Quiz at the moment are workable, but the BBs are more compact and lighter. I’m not too concerned about thermal issues with the Banebots controllers since PQ shouldn’t be doing any pushing anyway. I might revert the decision if they can’t handle the…
  • Frankenmotors, former slot car motors with giant magnets attached to them. I wish they had carbon brushes.
  • A GWS Pico receiver. The Cirrus 4 channel micro Rx that I use in PQ at the moment is actually too big to fit inside the frame! Spektrum makes a 2.4ghz receiver that I can use, but it is of the airplane type and does not return signals to neutral in the event of transmitter signal loss. Even though PQ is pretty close to takeoff when running, I’d prefer it stay on the ground.
  • 14.8 volts of 430mAh lithium polymer cells, rated for 10-15C. I want to run high voltage, low current this time around, so 6 amps should be plenty (14.8v and 6 amps is still a solid 80+ watts!)
  • A random 10 amp controller from United Hobbies. The ESC I have in PQ at the moment is 10 amps, but this one can run 4 cells. I bet it’s not as plushy as they make it out to be.
  • A 12 inch waterjet-cut titanium blade.
  • Lots of carbon fiber! Instead of using heavy 1/16″ fiberglass (Garolite) plates, I wanted to see if the lighter, stiffer carbon fiber material can be used in slightly thinner section (.041″). The weapon motor is supported from both sides this time, so I can probably run .021″ material on the bottom to increase clearance.

More to come! This is a summer project, since my attention is focused on the scooter and … classes (yeah, I think) at the moment.

Bot on.

Re(Snuffles Reloaded: Update)

Apr 25, 2008 in Project Build Reports, Project RazEr

There’s plenty in the imaginary part of the update, but you can’t see it anyway.

Hey, this thing looks kind of familiar… It’s the extend-o-pack, with 4 4000mAh lithium polymer cells in place. No, they’re not shorting on eachother, despite the precarious appearance of the tabs.

The Deans connector cutout is a snug fit and should hold a female Deans in place firmly with some CA glue. I need to remember to undersize slots and holes for the LASER cutter by a few thousandths to account for its kerf (since it cuts on the line).

Here’s how it was built.

This is sacrificial vehicle #2, another junked scooter that was sitting around MITERS. I’m not sure if it’s an older generation or what, but there are quite a few structural differences compared to the new A3 model I’m converting. It’s certainly beefier in the folding joint (0.1″ formed steel plate!) and brake area, and there is more material in the chassis.

Efficient re-engineering or corner-cutting?! The world may never know.

The plan is to cut a 12 inch segment out of the chassis and use it as the extend-o-pack body.

After a trip to the bandsaw, this was what remained. It was much like partitioning a fish for cooking – remove the tail, remove the head…

…and clean the middle. Oddly enough, with the parts that remain, I could make a very innovative vehicle.

Here, the side flanges that used to form the upper deck have been milled off, and in a previous unpictured operation, the mounting ear holes drilled.

It might have been better to mill with each flange facing upwards, since the cutting head is always at the same height as the vise above the table, but the vise might not be aligned with the table axis. I tried centering it in as well as I could, but across 12 inches of travel there was still .003-.005 of deviation, enough to have a flush-cut flange at one end but a very light remnant of it at the other.

Oh well, I’m not that good… yet.

After trimming the flanges off, I milled the remaining channel down to the design height of .606 inches, which is just enough to clear the two cells with some breathing space above. I did this in order to minimize the ground clearance hit – these things aren’t known for their great terrain ability, and I was only going to make it worse by sticking batteries under it.

Here’s one of the LASER cut acrylic endcaps installed. They are retained by some drops of CA wicked into the cracks between acrylic and aluminum and one screw on each side.

The waterjet-cut mounting ears have also been installed. It turns out I was off by exactly 2mm on the width of the channel, so dumping all 4 blocks in the same vise and running an endmill through at 1mm solved the problem and made it a slip fit onto the bottom of the chassis.

The other endcap with connector cutouts! On the right side is the last leg of the height-trimming cut where the milling cutter went Z-axis Tokyo Drift on me and ended up slipping lower. Oops. Crank the drawbar a bit harder next time?

And here it is installed. It actually looks quite elegant, with the exception of the other side where there is a small gap from the milling cutter slipping.

This whole assembly slides onto the chassis tube and the mounting screws grab the small flange on the underside to hold it in place. There’s lots of potential for “slide-on accessories”, actually.

So, with my Maxamps order on the way (two more cells to fill in the insides!) I need to get going and design the internal mounting structure. I suspect that it will also be a “slide-in” thing, but from the front – mounting via the four holes at the front.

Snuffles Reloaded: Update 2F1(a;b;c;z)

Apr 17, 2008 in Project Build Reports, Project RazEr

It’s getting EVEN CLOSER! Holy crap, it might actually MOVE soon! I took the opportunity of a convenient gap in the schedule to get some more work done. The motor and drive (okay, so it’s one unit) is mounted, and pretty much now the only thing missing is electronics. And the battery bay – because despite my best machining abilities, things still take a while.

Pics!

The motor mounting & structural & electronics bay mounting rails. These are waterjet-cut out of a sheet of 2024 aluminum. In fact, it’s the same sheet I had TB4.5SP1‘s wedges made from.

There are either 2 or 6 mounting points for the assembly, depending on how ambitious I feel. The quarter inch hole in the center of each beam clear 1/4″ screws to mount the motor between two of them (I cut four, in case I somehow fuck up twice). The 3/16″ slot at the front of each beam latch under the bolt which holds the brake fender in place. Overall, the assembly should only need those two (four) mounts, but I also have 4-40-tappable holes in a rectangular arrangement if I feel the need.

The small chunks are the mounting ears I designed for the supplemental battery pack.

Additionally, I GIANT LAZER’d some acrylic endcaps for the supplemental battery pack.

WHAT? CHARLES USE ACRYLIC?! WHAT HAS COLLEGE DONE TO HIM!?

Well, it’s not a particularly structural application, and plus the Media Lab laser cutter does not have a proper ventilation system for handling the cyanide gas that burning polycarbonate can produce. Also, the waterjet would have left a nasty draft and clouded the edges.

So why not. It’ll glow better.

Additional machining was require for each frame rail thing. First, the mounting hole had to be counterbored 1/8″ deep at 1/2″ diameter to accomodate the motor. I could have done without this, but extending the motor shaft gave me a critical few more millimeters to run the wires out without squeezing them. Here, on the right side, the counterbore is “extended” outwards away from the wheel. The three motor wires will sit in the hollow until they pass in front of the wheel.

Compare to the left, which only needs the dimple to seat the motor shaft.

Next, each rail had to get a chamfer machined on its outside-facing edges. The interior of the Razor scooter’s chassis extrusion has a radius of about 2mm. I cannot duplicate this precisely, so a messy one-size-fits-all chamfer takes care of it.

Because I had no dovetail or  real chamfer cutters, I milled these chamfers with a countersink.

That was exciting.

Next, the back end of the chassis extrusion had to be opened up to ~42mm wide, from the 33mm stock. The width of the motor is 40mm, and I did want some play on each side.

And hey, it’s mounted. I had to shave down the waterjet edge draft on the rails before they fit properly, but that was trivial to do on the mill. The motor slips into place and bolts down with no fuss. Best of all, it’s smooth.

Really smooth. I took the thing for a quick test cruise, unpowered of course, down the hallway. It makes the same obnoxious 3-phase PMDC motor sound, amplified n-fold by the waveguide-shaped chassis.

Here, the brake has been removed along with its retaining bolt, which would normally cross the gap and mate with the latch shape in the frame rails.

While the machines were warm, I turned a new standoff & wheelie bar for the back end. I could have just shaved down the stock one a bit, but I left it back at my dorm room.

It took about as long to find a piece of 1/2″ aluminum, shave it down to 11.5mm, drill and thread both ends, and throw it on there, as it would have for me to go back to campus, dig it up, and return, only to throw it on the lathe anyway.

Mine’s shinier!

So here it is, sitting amongst my random stuff pile. I need to design and build the electronics bay (internal) and the supplemental battery pack (external), then wire stuff up. Oh, and order two more 4AH cells from Maxamps.

I did ride it the whole way back (yes, with no brake… who needs ‘em anyway?), and threw some durability tests at it in the form of bunny hopping off sidewalks, wheelieing, skid-stops, and purposefully riding over stone-paved paths and cracks. Nothing has let loose yet.

At this point, I can make a duct tape test rig similar to what I did with Beta 1. Maybe I should do that just so I don’t have to get my hopes up…

Snuffles Reloaded: Update x[n]+2x[n-1]-x[n-2]

Apr 12, 2008 in Project Build Reports, Project RazEr

It’s getting closer.

In the last episode, the can was filled with magnets and the stator was test wound. I settled on 25 turns per tooth, distributed LRK style, as a starting point, and proceeded to wind the whole thing.

And here it is. This took quite a bit of time, as I had to make about 300 little wire loops in total and make sure they’re all aligned and not overlapping as to have the maximum space efficiency.

I initially bolted the stator to the end of a metal flat clamped to a table so I could tug a bit harder. After it took an hour to do half of one phase and after I discovered I could go much faster and pack as well by hand, I ditched it.

One quirk about the LRK scheme is that one phase is wound from the opposite side of the stator from the other two. I neglected that fact, and while it doesn’t impact the performance of the motor in any way, makes my termination slightly inconsistent, resulting in the need to fudge a bit to link the common ends of the windings. So… yeah, start the bB phase from the other side.

From the other side. The terminations exit from 3 holes drilled in the core mount thing, and then through the space between the motor bearing and the shaft flat. I found that this was a better way to do it than running everything through the center of the shaft.

So after the pigtails were soldered on and everything secured by a good helping of epoxy, it was time to drop the stator in. This was not trivial, because I could not hold onto the thing while they were being brought together since the magnetic pull was so strong.

In the end, a giant vise and the quill of a drill press came to the rescue. It may pull hard, but not hard enough to overcome a 50 pound vise.

Completed motor, side one…

Completed motor, side 2.

So now that it was put together, it was time for a test run. Of course, I neglected to bring my stator mounting pole back with me from MITERS, so I bolted it to the nearest object that had a 1/4″ hole in it.

Conveniently, it was my Institvte-provided bedframe. This brings up great visions of motorized furniture, but that’s another project for another day.

Here’s a video of the first firing of the motor. It’s a bit choppy – One of the side plates is not a good press fit, which, along with the poor molding tolerances of the wheel, caused the side place to mount a bit off-axis.

Using a drill motor with a wheel, I was able to calculate how fast the wheelmotor was rotating at a given drill motor voltage, and correlate it with the AC voltage it was producing at the leads. This gave me a working value for the voltage vonstant Kv, which tells how fast the motor turns given a certain voltage and an unlimited power source.

I worked out the number to be around 65, since it should have been rotating about 306 RPM given my 900RPM drill motor. It was producing 4.7 AC volts at that speed. An actual tachometer will give a more precise reading (and can do one better by factoring in system losses since it’s measuring the motor RPM while driven). Most likely, it’s higher than this, but not by much.

Anyone have a tach I can borrow?

It’s time to address the electrical system. Here’s some 4,000mAh LiPo flat cells from Maxamps, courtesy of the folks at the Media Lab. According to Maxamps, they’re good for at least 80 amps continuously. If I ever draw that much current continuously, I’m making a large smoke cloud with it. I currently (PUN!) have 6 cells for 22.2 volts.

Curiously, Maxamps doesn’t quite explain how 80 amps could be drawn through a little aluminum flashing tab. Yes, one of the tabs is actually aluminum with a copper bit spot-welded to it. Whatever, if it has worked for everyone else…..

Trouble alert.

The 3 cell stack doesn’t fit by about .5 millimeters. I wish I were kidding. However, I knew this would be risky business.

Most LiPo cells of this capacity are around 5.5mm to 6.5mm thick, with the thicker being the higher discharge-rated ones. Maxamps advertised their cells as 6mm thick, which, in a 3-stack, left me less than 1 millimeter of play inside the scooter’s 19mm tall internal channel. The original battery mount was going to be “bottomless”, with the supports for the cells coming only from the sides and the bottom capped by a wide tape, such as fiberglass strapping tape or Kapton tape. My 3S pack with 6mm thick cells (which is not high discharge, unfortunately!) did fit in a test.

However, these cells are actually the 6.5mm type. I won’t blame Maxamps despite the fact that it’s printed right on the cell: “6545135” indicates a 6.5 x 45 x 135mm cell; beecause designing a battery mount to such tight specs is not very wise anyway, considering wires and heat shrink and tolerances have to be accounted for.

The result is that 3 * 6.5 = 19.5 which is more than what a snug fit would allow.

So naturally it’s time to deploy the backup plan.

Here’s a sample section of a Razor scooter chassis. It’s a one-piece extrusion, which is great for mounting stuff to. The part above the flange (which is actually the underside, if you can’t tell from the scuffs) is 63mm side with a 57mm inner width.

Two of the cells stacked give ample clearance for a proper internal mount. Conveniently enough, a regular Razor scooter has enough material between the wheel mounts to hold 4 cells, 2 stacks of 2.

Here’s a rendering of the hypothetical underbody battery pack. The original plan was to use 6 cells mounted inside the body with an optional 6 added on in series on the underside to more run time. However, I have also considered just running one 8S pack split between the internal channel and the external mount.

So the plan is to take the long section of chassis, mill off the flanges and one long side, then make mounting ears that slip onto the underside of my frame and screw in place. I lose about 0.6″ of ground clearance by doing this, which is as low as I want to go anyway. Going this route with only 6 cells isn’t worth the effort, and so I will buy 2 more cells from Maxamps.

The charging port and balancer will probably be located in this secondary channel, since I’m free to machine it however I want.

But wait! Why do all this work to the chassis of another scooter when I could just use plain aluminum channel?!

…because they don’t make 63mm channel, sadly enough, at least not where I can get it readily. The only channel close in dimensions to what I need (that I have found so far) is far enough such that I can’t machine the walls down and maintain structural integrity.

Besides, there’s a junked scooter that’s exactly the size and shape I need.

Hey kids, this is how NOT to wire up lithium batteries!

The first test run of the motor was on 36 volts, off Snuffles I’s pack. I tried a test using only 22.2 volts, and the motor was rather sluggish, and rightfully so. 8S will be advantageous in this situation anyway.

Now that things are actually wired up and moving, it’s time to make the mechanical mounts. I hope that I can get a test drive in before summer…

Bawt on?