Sometime last summer, I crufted a bike frame with working brakes, minus wheels and a bottom bracket. A 700mm road bike wheel fit the front nicely, and a chunky scooter wheel with freewheeling sprocket presented itself for the rear.
I dug through the bin of motors and picked out an old induction motor that Charles had modified, dropping in a custom rotor to make it into a permanent magnet inrunner. I hooked it up to a with a generic 36V sensorless chinese e-bike controller and used the 12S3P A123 pack borrowed from the venerable derpscooter.
Despite the apparent functionality of this sweet ride, the torque produced was not particularly satisfactory. Bench testing of the motor placed it around 50 RPM/Volt. With a reduction of something like 3:1 in the transmission, that puts no load wheel speed at about 660 RPM. 11 RPS with a 25 cm diameter rear wheel yields 8.6 m/s – almost exactly 20 miles/hour. Not an unreasonable no load speed to aim for. I'd estimate it hit around 15, but the torque was pretty dismal. Other MITERS folks suspected a gain could be had by ditching the jasontroller and replacing it with a big R/C plane controller. The throttle was replaced with a servo tester and some connectors were replaced to permit the fitting of the new controller.
This time, the (still sensorless) controller had a tough time locking on to the BEMF of the motor. It required an awkward and difficult to master rolling start routine to get going. Performance was still comparable to that of the jasontroller. Well, guess we need more voltage. The still unnamed abomination cluttered up MITERS for a couple more weeks while I finished up my job up in Maine.
Now I love motors, but I just wasn't digging the three phase. It's unnecessary complication as far as I'm concerned as I'm not shooting for high efficiency or weight savings. Inspired by Mike's wizardly badass revival of the brushed etek LoLrioKart with a huge IGBT as a low side drive, I decided to make my own brushed motor controller.
I've made a few silly attempts at simple motor control over the last couple of years riddled with mistakes and setbacks due to my virtually nonexistent EE know-how. I've used pFETs where I should have used nFETs and been stumped as to why it doesn't work. I've plugged it in backwards, I've smoked my gate drivers without being able to explain why.
The ultimate precipitator was the appearance of a rather large brushed treadmill motor in MITERS. I measured it to be about 10 (yes, 10) RPM/Volt.
A while ago (maybe 2010-11), people at MITERS got really into tesla coils. Someone nabbed a bunch of Infineon FD600R17KF6C_B2 IGBTs at swapfest for coiling use. The stack remains largely undepleted and has been taunting me with its absurd power switching capabilities for much of my MITERS existence. The datasheet claims a Vds of 1700V and a continuous current of 600A.
I figure: a big honking low side drive is about the simplest way to get switching control of a DC motor. Here's what I slapped together to show the treadmill motor some love:
Back in cambridgetown for a few months, I assembled the controller in about a day in the following order:
- get arduino to spit out 2kHz PWM given a throttle input
- run that through an optoisolator, get a clean looking output
- feed that to a gate driver
- connect the IGBT and flyback diode
The vehicle has been dubbed Brems-Chopper after the giant hunk of power silicon that is its heart. Conclusion? Acceptable, but needs more volts! The motor is rated to 90V, the IGBT is rated to 1700V. The LS7812 that feeds the gate driver is limiting, as its absolute max. input voltage is somewhere around 40V. The original pine and steel strap motor mount is on the edge of failure, and needs some TLC.
I have coexisted with MITERS for about three years now. I don't think I'd heard of MOSFETs prior to that, let alone IGBTs (or ig-butts as they're called around here). Solid state switches are pretty freaking rad.