So a lot of this post includes events that actually occurred a while back... I've ridden 1000 miles on this guy... by the end of this post everything will be up to speed.
First up, battery packs weigh 19.5 pounds each. Here's the right one temporarily mounted for fit.
Here's a rear view. The duck tape is technically a temporary, but nothing works like a temporary. I've noted that the internal cardboard has a tendancy to work its way towards the front of the bike.
The batteries are attached with 15' nylon straps:
Wiring harness is well hidden. Most people don't know this is electric until I blow past them, or they notice the wires... if this bike weren't unique by being 8' long, I don't think anyone would notice...
The DeWalt charger sucks... I profiled the 'balancing' feature and found that 'balanced' means "within 300mV"... which is the entire useful range of the battery... that's why my cells kept being the wrong voltage when charged.
Here's the finished bike. As I mentioned, it's very hard to tell this is an electric bicycle. The bike weighs 89 pounds, with most of the weight (~65 pounds) in the back. I placed the front and rear wheels on a scale and found the weight distribution with my rider weight to be about 55% rear/45% front. This is much more than normal for a front wheel, but so far the front wheel has held up very well.
Since the DeWalt charger sucks and took 25 hours to charge crappily all the way, I went shopping for a nice power supply. At work we use Lambda supplies and I really like them, so I set this as my reference standard. After browsing eBay for several hours, I found a just-listed auction for a 60V/25A Lambda supply with GPIB, Analog and digital control, etc. This is a 2K supply and the buy it now price was somewhere north of $500. Needless to say, I immediately bought it now. It came, I picked it up on the bike and short of a long ramble, it rocks!
Here we are charging at 54 Volts, ~22 Amps (58.4V, 26.2A under full charge). Since this is a CC/CV charger, I simultaneously set the maximum voltage and current and whichever one the supply hits first is the monitored until the other one is hit... so right here I've limited the current to 21.18A and the voltage is climbing up to the maximum 58.4V. It's pretty automatic, but I cautiously set the limits during test. The charger will work with LabVIEW, but I'm not there yet. Eventually I might use some old HW to set up a voltage monitoring system and GPIB to prevent battery damage... but not now... as it is this charger can pump 1520W and that's pretty solid. FYI this is the same amount of energy that a big plug-in space heater uses.
All I can say about the above is don't plug the wrong leads into each other. I swear I'm not colorblind :). Instant melt.
The battery design is working great thus far, but I've noticed one particular bank is always empty first, so I'll need to monitor that in the long run. It's empty at 1.17KWh, which is way under the pack design and the empirical data from the other cells. I imagine one of the cells is dead. During my testing around mile 900 I accidently overcharged that bank, but the problem already existed beforehand, so worst case I made it worse :). I've come up with a zener diode/darlington circuit that will offer some protection. Here's the curve:
This is actual data I took from a circuit I conceived while riding a bike. Note that the zener diode I used has a zener breakdown voltage of 5.1 volts... obviously, I'll need one with a breakdown of 3.6 volts to prevent overcharge conditions; a 3.6V Zener simply shifts the data curves to the left. The red dots are a Darlington transistor setup; the blue dots are for single transistor. This is a crude BMS and all the excess current (i.e. 4A@6V=24W) is turned directly into heat on the shunting 15A transistor... but that's ok assuming a big enough heat sink...
...a better solution would be to use LabVIEW to switch out channels as they become full and then lower the power supply voltage 3.65V per disconnected cell. Since switching systems are what I do, it's ironic that I don't use such a system... money is the key issue here.
I mentioned previously I overheated my front brake while biking. Here's the solution to that problem:
Yup, that's the largest rotor you can buy for a mountain bike. Note the burn warp marks on the smaller rotor. The new rotor hasn't shown signs of overheating in 1000 miles, whereas the old rotor overheated in one afternoon of driving.
Next up, I rigged up a front hub onto the back of the Xtracycle, creating a so-called Extra-Xtracycle, which is about 14 feet long and is amazingly awkward to drive. A guy named Trever -- who is traveling across the USA and happened to be staying at our place for a week -- helped me brainstorm a mounting bracket to connect the hub to the frame. Adding any bike onto the back is as easy as removing the front wheel and setting the fork onto the mounted hub... some people think I'm actually carrying a 'spare hub'... cute ;).
So Trever and I made the mounting bracket out of the I-beam from a level I found in the road. The level is still functional, but a foot shorter. I was in a rush to create the bracket because later that day I had a mountain bike race at Lance's ranch that I needed to get to...
...42 miles away. I also stopped by my parents and met some cool people on the way home. This trip validated my trust in the bike.
I mentioned previously that I'm having problems with the rear wheel. The first build (that came with the wheel) lasted well under 100 miles. The second wheel build was actually a borrowed wheel that self destructed gracefully during SXSW... with three people on the bike. The third wheel used better spokes, but still the shitty A119 rim that the original wheel came with... and that lasted about 900 miles before breaking during an unintended bump in the road. Note that the spokes were 2mm too long on this wheel build so I just cranked them down... which caused an immense pressure as the threads were mechanically increased... this probably didn't help; all failures occurred at the nipple end right where they were overtightened.
So while I was rebuilding the wheel with new spokes -- 244mm, which I found to be too long and have since moved to 240mm, which are perfect size -- I dropped the damn hub and it fell wire-side down onto the concrete floor:
Note the damage to the upper 10 gauge power phase, the cut red encoder wire, and the nearly cut blue and yellow encoder wires. I could have fixed this without actually replacing the wires, but I'm going for a reliable vehicle and this would fail unexpectedly... so the only choice was to replace the wires again. In my last post I mentioned that this would take 5 hours... in fact it took exactly this much time, and it flew by since I knew what I was doing. It's so fun to tinker with things; all the theory they teach in college is meaningless if you never get your hands dirty. On that note, WD-40 dissolves the glue in electrical tape, which makes a lubricated mess.
Making the wire harness that inserts into the motor requires three single strand 10 gauge enamel coated magnet wires that must be meticulously straightened out. Five encoder wires are also needed, but since they don't carry any considerable current, they can be quite small. I used 22 gauge single strand wire this time, but 20 gauge multistrand also works... but it's a bit tight. All the encoder wires should go around the three 10 gauge phase wires... none in between the three in the middle. So then you wrap everything together tightly with electrical tape, making sure to overlap the tape such that there are two layers in all places. If it's thicker than this, it likely won't fit into the motor.
And we're done:
Close up of the straightened wires:
Now all you have to do is stick this harness into the axle until you hit the end of the spindle. Now comes the fun, tedious part where you have to bend three 10 gauge copper wires (and some change 22s) ninety degrees. Don't worry about the enamel on the ends... once we get the wires through, we'll cut off the first 2 inches or so; trust me, they'll be beat up and exposed. Once you jimmy the first bit through, you simply bend the wire down, pull up, then bend up and down, then pull up again... about 40 times and then enough of the wire is exposed.
One good thing about reworking this motor after 1000 miles of use is that I got to see that the design modifications I made are valid. The wires I installed last time are in great shape and no showstopper issues exist. One thing to note is that one of the three phase wires got hot enough to melt the solder and cause it to reflow:
This is the only one of the three phases that mates in a wishbone fashion:
the other two are soldered together something like a lashed flag pole:
I'm entertaining the idea that the electrons in the wishbone centralized about the bend and caused a large magnetic flux at the connection and that intensified the heat at this junction... this is definitely behavior I've seen of PCBs at work... who knows. Speaking of work, here's a picture of a test I ran that involved some loud RF relays... note the frivolous use of bubble wrappers to suppress noise :)
That's all for now...
...it's raining in Austin, so no riding for now...
...in other news, I got a Brooks B17 saddle and it's amazing. People complain they aren't comfortable for the first thousand miles... from mile 0 it's been more comfortable than the old saddle I used...
