DIYson

Firmware for driving single pole motors used in dyson vacuums with arduino

Things you will need:

1. Dyson vacuum running a single phase motor
2. Any high current H bridge, exact inputs doesnt matter as long as you can get current flowing in both directions
3. Any Atmega 168/328
4. Hall effect sensor salvaged from the old motherboard

A single phase motor only has 2 methods of control: amplitude and phase. Here, we keep amplitude at max, then the phase shift is tuned until max speed and efficiency is achieved. Speed control could be achieved by varying the input voltage with a variable buck OR very high frequency PWM minimum 160khz or rather depending on motor coil inductance (unachievable by atmega MCUs and most hobby H bridge modules)

Design considerations:

1. hardswitching at 62khz causes alot of switching losses espescially with typical off the shelf h bridge with slow transistors designed for typical 20khz switching (just to get it out of hearing range)
2. I havent found a way to make the pwm peripheral flip directions immediately upon timer 1 interrupt, usually it waits until next overflow before the direction variable is loaded and acted on
3. So I skipped PWM peripheral to remove the delay before the it fires and simply fullsend the motor ie it is stuck at max speed
4. This would significantly reduce mosfet switching losses by 3x from 63khz to 15-20khz(theoretical max toggling frequency depending on input voltage/motor speed)
5. I found that at a certain phase shift (coincidentally where the motor is at max speed) the mosfets run alot cooler which leads me to assume some zero voltage/current switching is occuring
6. assuming 20khz input frequency (at 120,000 rpm) theres 800 clock cycles between each INT0 & TIM1 interrupt which really doesnt leave much room for interrupt handling

Taking reference from traction motor waveforms, I could implement a 1 step pwm by simply letting the current freewheel by setting COMPB vect to trigger and switch everything off/disable drivers earlier before COMPA switches the direction. (PWM_decimation to be tuned) TODO: add rudimentary PWM by using timer1 compB to disable the drivers earlier to let it freewheel and tune how early to let the motor freewheel

//in setup, //enable COMPB interrupt //set digital pin to input_pullup //in main loop, digitalread an additional pin and then set PWM_decimation according to HIGH/LOW/med //in setoneshot, set compB trigger point not too small but not too big either OCR1B = durationin*PWM;//where 2 < PWM_decimation < phase_advance //in compb_vect, disable/ensure all drivers are in the same state ie freewheeling/open motor PORTC = 0;//everything off //DO NOT DISABLE TIMER

WARNING: the whole of PORTC is switched together to save cycles since those extra pins arent being used

step 1: poke motor a few times by alternating polarity until the hall sensor catches step 2: once it starts rotating, switch pulse exactly when the hall sensor interrupts, spins to around 40000rpm at 12v step 3: to speed it up, phase shift the drive waveform a bit forward, this is accomplished by settomg timer1 to fire a little earlier than the previous edge

Hacks: reset function and lazy virtual ground/supplies