Crap

We got three out of four hub motors calibrated, which is awesome. The current problem is that the fourth hub motor's sensors decided to die...no idea why or how, because they sometimes work, and sometimes not. It can't be a magnetic field issue because a. they are mounted in the exact same locations on all motors. b. they do change state...except sometimes to the WRONG state. WTF, that shouldn't even be possible. I even tested them with a small magnet and an oscilloscope before assembling the motor, and they worked fine. And we determined it wasn't a problem with the MC board because we ran a different hub motor on it just fine. Oh well. I've ripped the sensors out and am going to put three new ones in at some point. Then reassemble and test again. Then put NEW HUBCAPS on! Wohoo.

Lessons learned over the past month or so:
1. Sensors can fail even when taking every precaution
2. Don't use steel alignment pins (see earlier posts) until the final motor integration onto the board. Pulling those things out sucks
3. Make your hub motor rotors with a water-jet, like this.
4. I should have wound the motors hotter (or just made smaller motors). At 10A, I'll have more torque than I could ever use/stay on the board with. In fact, this board would have no trouble at all running on two motors. Something I will consider if weight ends up being an issue.

Mid-November

I'll know I'm doing too much stuff when these posts start coming at more than 1 month intervals.

Anyways, time for another ELB update!
I finished cutting and rough sanding the last 40 polycarbonate inserts for my new hubcaps.

It's INCREDIBLY tedious work to polish these things. The above is after cutting, 220 grit, and 400 grit. There's still 600, 1200, 1500, and 3000 (lapping paste) to go. Anyways, this is the point where I got bored and decided to work on something else.

This is my charger: Turnigy 4x6S . It can charge four 6S packs independently. I figured I could use it to charge my 12S ELB pack (just link two of the ports together). But...
Each charging port is NOT independent, despite what you may have heard about it, or may assume about it, or read about in the description ("All ports work independently of each other."). From the outside, all you see is a single input for 12Vish DC and 4 charging ports. Internally, there are 4 separate chargers. Sounds ok, right? However, all of the chargers' inputs (+ and - ) are linked by traces that go all the way around the board. This wouldn't be a problem if they were high quality (isolated input/output) chargers. But this is a Chinese Cheap Charger (CCC). The negative/ground wire is connected straight through from the input to the output (battery ground). This means that all of the outputs are at the same potential, which in turn means you can't link ports and charge anything over 6S. I need 12S. There goes $100. OR:

Doctorbass to the rescue. Check out his hack for this charger...genius. He made it into a single 24S charger by isolating the four chargers by cutting the traces that link them, and then buying 4 small (cheap) switching power supplies, 1 for each charger. His application is a little different than mine, so he did some other cool modifications, but I'll just show how I did it.

Take it apart

Each charger has two holes like this for wires to go into. It's like they were originally going to manufacture it to run off of four independent power supplies (like it should have been), but they instead just added a trace all they way around to save money.

Cut traces on each side of each set of input holes. Dremel worked well.

Cut the bottom traces, too. Be careful not to cut into other traces, just the power traces.

Back side was easier. Could just slit the board without hitting any components.

The four 12V, 5A power supplies. You can get these on eBay for <$40.

Cut the DC plug off the power supplies and string the wires through the wire hole in top plate of the charger. (De-solder the existing input wires first.)

Solder the new inputs to the circuit board.

Yep, definitely isolated now.

Finish soldering all the input wires. Route them as you see fit.

Semi-finished. They all work!

Removed excess wire. That's better.

Double-stick tape/duct tape the power supplies together. I velcroed the charger to the power supplies, too.

Now I have one big block of a 24S, 200W charger. Not too bad for $130 and 2 hours.


In addition to these things, Shane's been helping me calibrate the motor controllers to my motors (well, he's been doing most of it).

Note the old hubcaps with chunks of missing polycarbonate....why I'm making new ones.

I've got updated data for the motors now, too. Before, using a Turnigy controller and running the motors sensorless, we calculated a top speed of about 30mph. Now, using sensored sinusoidal control and about 15 degree advanced timing, the top speed is around 20mph. Not too bad. The really impressive thing is the torque. I'm looking at getting 50lbs of force at the ground..assuming 200lbs of mass (rider+board), that translates into 1/4G of acceleration, which is nuts...especially for something without a seat or handlebars. I probably should have wound the motors a little less for torque and a little more for speed (less windings, or thinner stator, etc), but I think 20mph is plenty. 



That's all for now.

The other Electric Longboard

Well, bad news for the other electric longboard (the one made during the Edgerton Center Summer 2010 Engineering and Design Class).

Remember the eXKateDB (drill battery) hack?:

I shoulda seen it coming, but see those little connectors? They can be plugged in backwards...yeah...bad stuff happened. Turns out the Exkate controller doesn't have reverse polarity protection. I actually found this out the hard way at the end of the class this summer. I plugged the battery in backwards and it fried some components. Luckily, Shane was there to help me fix it. This time (wasn't me) however, more things fried:

Blown up transistor and resistor. Another transistor is fried on the board.

You can see the diode we replaced before. It was fine, but the transistor below it wasn't too happy.

Assuming more stuff wasn't burnt, we probably could have fixed it, but the whole thing was turning into a hack at this point. So....new controller! Brushed Kelly Controller to be specific. It has better specs than the Exkate controller, for a lot less. However, it doesn't have a built in radio system. So....new radio system! A Hobbyking 2 channel 2.4GHz radio was purchased ($17). We'll have to figure out some way to translate the PWM servo signal to a 0-5VDC signal for the Kelly throttle input, but that shouldn't be too difficult. Hmm...what to do with the extra channel.

A new set of Lipos are here (I need to find time to install them :/ ). To prevent what ultimately caused all of this trouble in the first place, a keyed battery selector will also be added. After that, the longboard should pretty much be set and idiot-proof resistant. 

About the Edgerton Center at MIT

http://techtv.mit.edu/videos/9e150e3e855438bb92984ae2e87081cecc19684a/private

Check it out.

Polycarbonate Inserts

Time to suck it up and polish some polycarbonate.
I bought 220, 400, 600, 1200, and 1500 grit sandpaper, as well as some 3000 grit lapping compound, a felt bob, and a cotton bob. With my polishing kit ready, I cut ten pieces of the 1/2" dia. PC and ten of the 1/4" dia. PC with a band-saw. Then I started with the 220 and progressed to the finer grits. To remove the burr raised from sanding, I put each stud in a drill press and filed down the burr. Then I flipped the stud over, and did it again.
For the buffing, I put the felt bob (I found it worked better than the cotton one) in a drill press, applied some of the lapping compound, turned it on, and buffed the ends of the studs. They came out much clearer than before, but still kinda hazy.
So then I took the heat gun and carefully heated up the each end, making sure to take the heat off before the PC bubbled. The final result were ~.45" long PC studs that you can read text through. Then I pressed them into a hubcap with an arbor press. Check it out:

Random sheet of paper I found. You can see the circuit diagram through the PC.

However, it wasn't nearly as easy as I just made it sound. The whole process took 4+ hours (yes, I'm blowing off work). That's just for 1...I still need to do 3 more. Ahhh, not enough time.

October ELB

Despite a crazy work load, I was able to get a lot done this month. The machining on the new AL hubcaps is complete. They're made out of 2024 alloy, which is very tough as aluminum alloys go. Check it out:

(Ignore the chuck changes...these images are from different days.)

The apparent warping is from the thick coat of oil covering the part.

TA-DA:

Lathe work done

Sorry, fuzzy.

Done

Scale: these are 3.25" in diameter.

24ish hours total spent in the machine shop. Polycarbonate plugs will be press-fit into the big holes to maintain waterproof-ness (remember, one of the goals is to have a waterproof board I can ride in the rain/snow). You can sorta see the ring I left inside the bearing pocket- that's for spacing out the bearing so the inner bearing race doesn't rub on the outer wall (learned that from the first set of hubcaps). The large holes in the inner side hubcaps (the ones that don't look like swiss cheese), are for the massive 2" bearings I bought. See previous posts for why I'm using 2" bearings. Anyways, they came out really great.

The next step is to figure out how to make the polycarbonate plugs. I bought some 1/4" and 1/2" PC rods and plan to cut them into 80 plugs. The big problem is figuring out how to polish the ends so that they are see-through. There are a number of ways to polish polycarbonate:

Potentially the best way is methylene chloride vapor polishing. This method basically eats the top surface (and thus the scratches) of the PC part, leaving a perfectly smooth (and thus clear) finish. The problem with this is that methylene chloride is nasty stuff, and I don't have access to a vent-hood. So that won't work for me.

Another method is dipping PC parts (or any clear plastic parts apparently) in certain types of floor was (like Future) and letting the wax dry on the part. Apparently, this results in a perfectly clear part (I guess the wax fills the scratches?). I know R/C plane hobbyists use this method to refinish their cockpits. My problem is that my motors will give off heat, possibly melting the wax. 

Another way is to just sand the ends with steadily finer and finer sand paper, and then use polish (like jewelers rouge) on a buffer. I tried this, but the ends were still very hazy. I'm going to try again with finer grit sandpaper (like, 800 if I can get it local).

Yet another way is to heat the surface of the PC part to the point where the surface melts. Unfortunately, you have to make sure you are evenly heating the surface, and only the surface. If you overheat parts of the PC, it'll bubble. I tried this method, too, with a heat gun, but the edge got a lot hotter than the center, causing bubbling. However, the parts that weren't bubbled were significantly clearer than the polished ones, so if I can figure out some other heat source, I'll try this method again. Results:

The outer two are sanded and polished (note haze). The middle one was heated (note bubbles).

Anyone have a good way to polish polycarbonate?



Here's a sneak peak:

Beginning the electronics layout.

Note the old hubcaps.

My activities this term

Dear readers,

You may have noticed that the Electric Longboard project pace has slowed down significantly since the summer. This is due to the following activities:

1. Fall term Classes. I'm taking the Aero/Astro Core class(es) called Unified, which is basically one GIANT 48 unit bootcamp course covering thermo, fluids, structures, and signals/systems. This is in addition to other classes, including an advanced composites course (16.202) and German 2.

2. Research. I'm starting a UROP (undergraduate research opportunities program) in the Space Propulsion Lab. I'm working with a senior on a new kind of L.E.O. ion thruster. VERY cool.

3. Rocket Team. The MIT Rocket Team is doing NASA's University Student Launch Initiative (USLI) this year.

4. Design/Build/Fly. I'm doing DBF again. This is always a huge time commitment. Last year, while it only counted as a 6 unit UROP, I spent probably 18 units worth of time or more on it.


5. House Manager for my fraternity. Takes LOTs of time and effort.

Yeah....I'm busy. So I'll fit the ELB in when I can.

9/20/10 Update

ELB: The motors are in the process of getting internal Hall effect sensors. The method seems to be working well, so I plan on doing it to all of them, reassembling temporarily with the semi-busted polycarbonate hubcaps, testing/calibrating, then replacing the hubcaps with aluminum ones. The aluminum stock is in, so I should be able to start machining those soon.

eXKateCD: It is now the eXKateDB (exkate drill batteries). Let me explain. Someone (no idea who) came into the lab, saw it under the desk, probably tried to play with it, and left the Lipos plugged in. This drained the ENTIRE pack down to about 1V...yeah, Lipos don't like that; they're dead.

Wall of Shame.

So while we wait for $160 worth of new batteries to come, I rigged this up:

 2 18V drill batteries in series. Not pretty, but no noticeable power loss, which is nice. Plus, we don't have to worry about these being drained all the way down (NiCD's actually like that). Our plan to prevent this from happening again is to have a keyed battery selector and hiding the key. I never wanted to have to do the 4 hours worth of soldering, etc. that it took to put the original Lipos in again, so I definitely don't want to have to do it a third time.


That's all for now.

9/5/10 Motor Update

Good news: I took the sensor board out and tested it with a small magnet. The sensors were fine, they just weren't close enough to the magnets/coils. So I'm going to stick them in the slots of the stator and get rid of the sensor board. Even though they won't have the same angular separation as on the board, it will be a combination of slots that works, such as A(sensor)aAB(sensor)bBC(sensor)cCAaABbBCcC. I'm pretty sure I did a previous post on calculating hall effect sensor positions, or it's in my design review for this project, so look that up if you want to know where to place hall effect sensors. Anyways, this means that the sensor board cannot be adjusted, which is nice to have if you want to mess with timing. Oh well.

picture added 8/20/11

picture added 8/20/11

Bad news: Chips of the polycarbonate hub caps/spacers fell out. However, it wasn't the total disintegration I thought it would be. In fact, it's salvageable, so I may just put them back together the way they were. (I'll at least do one to test the sensors in the slots idea.) On the other hand, sidewalk cracks and other bumps might finish the disintegration process, which would be really bad while riding.

Another option is to make new hubcaps/spacers out of aluminum, but it wouldn't be see-through anymore. :(
I could waterjet new polycarb spacers and hubcaps, but I really don't want to deal with that again. Especially because it'll be impossible to get rid of every little molecule of loctite.

Another option is to shove the rotor into an aluminum pipe and make aluminum hub caps for the pipe, eliminating the spacers. Something like this:

I got this idea from Charles if you can't tell.

Cool. The cylinder in the background is the tire.

The holes are filled with glued or pressed-in polycarbonate disks. I really like the way this looks, and I think it'll be much stronger than the current design (though the current design is plenty strong...). The larger diameter will give a slightly higher topspeed. One draw back with this is that the outer edge of the hub cap is now an important surface and has to be a very close fit for weather-proofing. In fact, the hubcap-outer tube interfaces would probably have to be sealed with silicon. Drilling radial holes is no easy task, and getting it apart would also be a pain.

Here's a combination of the above two options:

 Aluminum hubcaps with polycarbonate inserts in the outer-side one. Keeps the current bolt pattern. Scraps the useless o-ring seal (explained in an earlier post). This allows partial see-through without the draw backs of polycarbonate (can't loctite, not as stiff).

I'm leaning towards the last one.

Sigh...

Bad news. A bunch of my sensors aren't functioning. This means I'll have to break the motors open and either a. fix broken wire connections, b. reposition the sensors, or c. both.

That doesn't seem that bad, except the polycarbonate hubcaps have spiderweb cracks from the loctite and parts of them might disintegrate...which would be very bad. It would pretty much mean a redesign of the motor casings.

Since I don't really have time for the next few weeks to work on it directly, I've started designing custom LED lights for it.

That's all for now

eXKateCD

The build log is this Instructable. Feel free to ask questions about it in comments here or there.




Other projects included:

• Overhaulin' this custom built Segway
• Making a Quadrotor from scratch (an instructable should be up soon, so please don't ask me about it)
• Pong from scratch (and by scratch, I mean wires, LED's, and microcontrollers. I didn't work on it, so again, please don't ask me about it)

Here are some pics of those projects and of the Edgerton Center Summer 2010 Engineering and Design Class in general:

Parts of the quadcopter

Believe it or not, the quadcopter actually flew.

Pong from scratch. Tilting the controllers causes the paddle to move.

The old segway.

Clear deck

Oh yeah, LED's

The eXKateCD team

Class photo

Shane's summary is here.