With the semester winding down (or, perhaps, finally ramping up!), many of the 2.00gokarts are in the process of being wired up and tested. The final product is due next week, and our competition (last year’s video) is on May 5th!
Some of the students have been industrious and scheduled their checkoffs and inspections early. Here’s a preview of the action that will unfold in a much larger space next week:
Because conventional controls and riding postures are for wussies, apparently. I’m both amused and somewhat terrified at the prospect of there being three (out of eight) karts in which you ride head first. As it was my stated mission to not interfere much with the design and construction of the karts to let students experience as much of the design process, I might have to start padding BurnoutChibi and run interception for wayward karts.
Speaking of which…
Here’s a picture of the aftermath of BurnoutChibi’s motor detonation. As I would later find out, the sparks seen in the video were not the magnets grinding on the can, but rather them cutting up the phase wires.
Here’s a better picture of the ownage. The red wire, in particular, was cut almost all the way through. The annoying thing about this is that the wires were so close to the stator. If they were further out, patching would be a simpler job. I’d have to loosen the epoxy holding the wire stubs in place and also trim the heat shrink selectively.
While I await better motors, i decided to try and repair these. First step was to pop them open. There is a front retaining ring that comes out, then 2 set screws loosen up to free the shaft from the rotor. Then it’s a matter of pushing the shaft out to the right in the picture – this step was done on an arbor press.
Ouch. In total, five magnets broke loose. I figure this must have been a chain reaction where one magnet ditched first, and the resultant imbalance caused can deformations which broke the rest loose.
This is why I recommend motors that have “rotor bearings” or “skirt bearings” to everyone who asks me about them for vehicle apps. Even though it adds a little drag, the distal end of the can is properly supported on its own bearing. The only exception is if the motor is very short, like a more “pancake” style design.
I mixed up a generous dose of long-cure epoxy with glass microsphere (microballoon) filler, to slightly under nutella-like consistency. The offending magnets were pried out, the mating surfaces cleaned, then this epoxy smeared into the new joint. I replaced the magnets and used as much of the remaining epoxy as possible to completely fill in the gaps between them.
Evidently, I didn’t add enough microballoons, as the mixture did sag a little. To keep the cure symmetric, I actually chucked this thing into Tinylathe and ran it on a very low speed for several hours.
After the mixture was firm (but not cured), I set it on a radiator to cure with heat. Luckily for me, the radiators in the building were still on; they were switched off successively as recently as 2 days ago!
I didn’t get a good picture of the wiring repair before, but it basically involved exactly what I described before – carefully scraping away the heat shrink tubing to expose as much wire as possible. The wire was actually all magnet wire, so it would have been difficult to solder. To combat this, I “frayed” each lead as much as possible to expose the maximum amount of magnet wire surface area. Then I cranked the 80W soldering iron up all the way to 850 fahrenheit and literally burned away the enamel by embedding the frayed ends in a big ball of solder for heat transfer.
I think I managed to get back 75% of the red lead. The rest were patched similarly, but did not need as drastic soldering measures.
After the real epoxy fully cured, I reassembled the motor and crammed it back into the left side transmission.
I have yet to ditch a single magnet. Though I figure it’s only a matter of time before the right side lets go…
And with that, BurnoutChibi is ready to lasso its rogue… brethren? Bastard children? Offspring conceived via assistive reproduction technologies? Something. The only thing it does not do very well, sadly, is burnouts! Because the rider weight is basically square in the frame, and is up so high, it really just like to drag the front wheels along even if I’m holding the brakes. The same reason contributes to its severe power understeer (and associated lift-off oversteer!) behavior. Oh well…
In the previous week of work on BurnoutChibi, I’ve fully completed the vehicle but have yet to get it out to really test. This thing really is too damned fast for our indoor.. uhh, test track. A motor quality issue also prevented me from blasting it around in our usual outdoor venue (for very long, anyway). These issues have since been addressed, so it’s almost time for more test video!
As previously discussed, BurnoutChibi is a refit of the derelict Chibikart1 frame into something a little more hair-raising, as if Chibikart 1 wasn’t bad enough already. Since the last update where I had just finished reconnecting the steering, I’ve finished mounting the braking system, the transmission shifter cables and linkages, and also completed electrical hookup. At the behest of some of my students, I completed it in time for CPW last weekend, though the aforementioned motor problem meant it was not out scaring parents and wide-eyed potential freshmen.
Here’s the story in the pictorial form.
I began with a little aside in order to solve the problem of how to mount the two “Sand Castle” controllers. They have no mounting flanges and both sides are made of heat sinks, so just gluing it to a plate would make for some pretty poor thermal design. I decided to come up with a “cradle” that held the two controllers right under a fan for some forced convection cooling. The fan I selected was out of my plentiful stock of 80mm LED case fans.
This design was an exercise in designing a snap fit for 3d printing. While I could have made the base a little wider and added some through-holes to hold the two halves together, I decided to get creative and dovetail each corner post together. The angle is extremely steep – about 85 degrees – so the whole assembly could be pulled out with force, but otherwise snaps into place cleanly.
…and it’s printed out of PLA.
Yeah, so what if it’s going to melt at about 60 celsius? It’ll just smell like delicious waffles while the ESCs burn.
I decided to try the “translucent light blue” PLA which is sold commonly, and I must say it’s my favorite PLA color so far. It’s not the vaguely jaundiced-rainwater color of natural PLA, and I also don’t like solid color PLA. A tinge of blue helps, but is not overwhelming and makes me think it’s some real plastic.
Putting together some of the electrical deck and testing the fit of the ESCs. Result: pretty perfect!
I set aside the e-deck for a while to return to the transmission and drivetrain.
First order of business is to attach the sprockets to the wheels. This basically entailed making four standoffs which acted as the lug nuts (M6 thread) on one side, and regular 1/4″-20 on the other side. The standoffs hold the sprockets a set distance from the wheel so the chain clears the tires, and also holds them concentric.
Or so I hoped.
There is practically nothing concentric or wobble-free about these shitty caster wheels. I had picked them up since they’re $10 each, but I swear not even Harbor Freight wheels are this bad. While the sprocket seemed to have minimal runout (radial misalignment), the wobble from the poorly stamped wheel rims was incredible.
I literally had to take a dial indicator to the sprocket and hammer on the wheel rims to bend them around. I got most of the axial wobble out of the sprocket this way, but this meant it all ended up in the wheels themselves, which now are a bit “googly-eyed” as a result. It will look hilarious when running.
With all wheels mounted, the frame could finally support weight. It’s definitely lost the Chibikart look a little since it’s so far off the ground (in comparison…). I have an incredible 2.5″ of ground clearance now.
The brake pedal hookup was the exact same as for DPRC. This pedal design doesn’t have a spring return on the pedal side since it is handled by the built-in spring elements in the brakes themselves.
Which, as it turned out, weren’t quite strong enough, so the pedal felt quite mushy and also did not return all the way. I added a long compression spring on each side between the cable stops and the brake lever, and this made the pedal feel much more positive. The brake cables sit in barrel adjusters so the balance could be finely tuned.
Shifting to the back again, I’ve appended the Vex sprockets to the Vex transmission’s VHex output shafts. The Vex sprockets didn’t come with any set screws or other means of axial retention, so for a quick fix, I drilled and threaded three #10-32 screws 120 degrees apart. The three set screws will offer way more retaining power than just one. I decided to forego any other spacers and shaft end-tap screws for now.
Here’s a view of the shifter linkage. The mechanism is a spring-balanced cable setup where I provide the pull to shift into 2nd gear, and the spring pushes the shifter back into first.
This was simple enough, but I chose springs which were way too strong initially. I figured “10 pounds of force” at max deflection was enough, but that translated through the cable into the shift lever, times two, meant it was just too hard to throw!
I went to a hardware store and bought several sizes of springs in roughly the same length that were much ‘softer’. The replacement spring is about half the spring rate, and was also too long in that it could not compress enough. The solution to that was to really quickly dremel a few loops off the spring, just like a good ricer. The shifter now has a positive click as the ball detents lock into place.
Once that affair was taken care of, I routed the chain and moved the gearbox up to tension it (the “goalposts” having slotted mounting holes for this reason). To lock the gearbox in place, I simply tightened the…
… Oh, I can’t reach those bottom socket screws.
Must have bought those hex headed screws for a reason! I was wondering briefly where they were supposed to go on this thing. With the hex heads accessible with a regular wrench, now I could actually tighten the drive up.
With both transmissions hooked up, I spent some time getting pushed around synchronizing the cables. I put another set of barrel adjusters on the shifter cables so they could be adjusted as needed.
What I (not surprisingly) discovered during this push testing is that the brake shimmy is pretty severe. This is caused by combination of factors, two of which include my “kinematically suboptimal” rotor retention method (two screws across a diameter) as well as the complete non-concentricity of the wheels. To reduce the severity of the effect, I had to dial the cables to different tensions. The braking is still effective, but it definitely feels like it’s trying to jerk all over the place.
Ultimately, I’m likely to ditch these drums and go to a disk brake setup with its own guide bearing on the front spindles to maintain concentricity. But for now…
…back to the electronics deck. Here’s the wiring mostly in place with batteries mounted. The batteries are my old 5Ah, 10S sticks. Two of them.
The batteries are secured by Velcro ties and sandwiched between two rigid plastic panels (the baseplate on one side, a 1/4″ thick polycarbonate strip on the other). A 1/8″ silicone rubber pad sits below each battery for shock absorption and more impact protection. Combined, this ought to ensure the batteries don’t move anywhere.
The ESC power leads directly into a 150A fuse junction, and ground has its own big brass distribution block also. Overall, this is the beefiest power system I’ve built since probably LOLrioKart.
At the point, the frame was flipped over for installation of the power electronics deck. The rest of the wiring, including connections to the motors and to the main switch, happened in-place after the installation.
The long run to the power switch is doubled-up 12 gauge wire in each direction.
The only other power side wiring was to make one motor extension cable. With main power wiring completed, I quickly hooked up a HV BEC to provide 5V and a servo tester to convert the foot pedal’s analog 1 to 4 volt output to servo pulses. These two components were heat shrunk and sealed, then attached with Velcro to the top of one of the battery pack plates. The signal electronics for this thing are extremely basic – no fancy signal processing is occurring. One thing that could happen with this system in the future is converting to electronic shifting, such as with solenoids, upon which I think a system which cuts throttle before the shift and slowly brings it back in would be helpful.
After confirming the functionality of the ESCs and calibrating the controllers, the whole rig is put together.
Here is BurnoutChibi posed next to DPRC! The wheelbases for both vehicles are the same, but BC has a slightly wider track because of the pneumatic wheels. Otherwise, they handle alike and are mututally just as difficult to sit in.
testing
The first few test runs of BurnoutChibi were done indoors, in our Conveniently Circular Building hallway. Due to the extreme acceleration ability of the vehicle, I couldn’t really test it any faster than DPRC or original Chibikart, so we decided to not take video. More testing commenced in an underground garage, then our usual spiral parking garage haunting ground. Unfortunately, I really only got a minute or two of hard driving in before the left motor threw several magnets.
The high speed of the motor caused some serious sparking as the loose magnets scraped the stator and also cut up the motor leads. Unfortunately, the only video that was taken was not focused properly…
Since that test, I’ve reglued the magnets and repaired the wiring, and BC is currently operational. I am currently waiting for a day in Boston / Cambridge when all hellis notbreaking loose (in fact, as I write this) to test in the garage again. These pictures and videos will be uploaded when they are taken.
