Monday, June 26, 2017

DISASSEMBLING MOTORS, TINY BIKES, AND A SHAFT


Hello again! Project directions have shifted quite drastically since last post. While compiling the bill of materials for Tee-mobile, it became clear that this was going to be an expensive endeavor.

Cost of raw materials sans motor, wheels, controller, batteries..

Luckily, the quest for a fast thing didn't end there. Last summer, I wound up with one of these babies:


[Vrooming Intensifies]

A kid's pocketbike complete with things like brake (singular), belt drive, chassis mounts for electronics, and bodywork ripe for plastering with obnoxious stickers.

Stock, they come with a 250W brushed DC motor and controller, and some SLAs. However, this was soon going to change.


Womp @_____@

Courtesy of MITERS I was given one of these things. A copy of a Hacker motor, and mechanically very similar to Hobbyking's 150cc Rotomax offerings (same mount pattern and external dimensions), this little guy (dubbed the Tiramisu) is good for ~10kW with a nice and low kV (100RPM/V) for vehicular operation.

Running the show is a 72V 100A Kelly sourced from no less than Dane.

But Noel, the motor has the shaft on the wrong end!

Have no fear, italicized text: we thought of that already. We have the technology.

In search of the one true shaft, I began by removing the end-bell shaft. Unfortunately, these motors had gone swimming once upon a time, and the four fasteners keeping the shaft attached had suffered from the softening. 
Ripe for stripping

Initial attempts involved using the correctly sized hex heads, but that resulted in the inevitable stripping of the heads, so I resorted to gently hammering in a slightly larger torx bit.

Success!

In hindsight, I should've left the end-bell shaft on for initial disassembly, however, it led to the discovery of the bolt keeping the real shaft secured to the rotor.



After the liberal application of WD-40, the shaft bolt was removed and I undid the motor mount-side screws keeping the can attached. The rotor and stator assemblies then happily came apart after some pulling. Undocumented is the use of the mill to pull things apart: I clamped the motor mount in the vise and used the drill chuck attachment to grab hold of the reinstalled end-bell shaft.

Afterwards, I took the rotor assembly, removed the end-bell shaft and the bolt securing the one-true shaft and clamped the can in the vise and pulled yet again.


Pliers unrelated


Tape was applied to the magnets promptly after removal to prevent crap from accumulating in the air gap.

With the shaft extracted, I took some measurements and turned a new, longer one that would protrude out of the mount side. The taper was completely unnecessary and may as well just be a step to make machining easier.

Left: Original. Right: New and improved.
In reality, this process took a few tries to get right because I tried to jam things together with a hydraulic press after machining the initial shaft slightly too large and wound of destroying a bearing that had to be angle-ground off of it  I suck at turning things and was too lazy to do it right the first time.

The tolerances on the motor are such that the bearing interfaces with the shaft are slip fits, while the interface between can and shaft is a press fit.

Reassembled, things looked pretty good!


Whew

The motor mounting bracket was also made out of a piece of U-Channel left over from last season's car.

U-Channel evolved into ... angle bracket ಠ_ಠ
In the back of my mind this whole time was also the fact that I had to mount a pulley to this thing, so afterwards, I went ahead and also filed down some flats on the shaft for the pulley that would mount on the shaft. Thankfully, the bike came with a standard 5M HTD belt, so I could buy a pulley off of ebay.

Almost a thing
Also sent out was a hall sensor board to play with the Kelly. The motor is a 10 pole pair affair and using some math, I established the mechanical spacing between sensors. That is:

1 rotation = 360 mechanical degrees = (360 * N pole pairs) electrical degrees.

For this motor, we get 10 times the number of electrical cycles per mechanical cycle.

Going further, we get a conversion factor of

0.1 mechanical degrees = 1 electrical degree

So for 120 degrees of electrical offset, we need the sensors to be mechanically offset by 12 degrees.

The idea of having the hall effect sensors mounted to a board was also attractive, so I went ahead and made a board in Circuitmaker in part to get some more experience in the Altium-like environment, but also because importing funny board shapes was especially convenient.





The design is a blatant knock-off of the boards Charles has made in the past with adjustments made to fit the larger can size and pole count of the Tiramisu and the addition of a JST-XH connector on the back. You can buy Charles' stuff here.

As a Circuitmaker project, all of the files are free to access in their community vault. Just search for "Tiramisu Hall Board" in the public projects bar.

IRL



With board made, I made up a quick mount in Solidworks that I ended up printing and bolting to some holes I drilled in the motor mount.


Rawr
Note: I'll probably redo the design and add some support material between the standoffs to prevent bending and the terribleness that is encoder misalignment.
Motor on a Bench
Unfortunately no video was taken during the encoder tuning process, but it went a little something like this:

Whirrrrrrrrr

>Looks at current draw
>Adjusts Encoder

Whirrrrrrrrr

...

Once the position was settled, I filed down some flats on the motor shaft, mounted the pulley and everything was dumped into the frame.




Next up: batteries and wiring!