Nick Kirkby

3 Phase Bridge Of Learning

IRS2186 4A gate drivers – independently controlled high side + low side.  8-DIP.  nice!

Generic TO-220 N-Channel MOSFET.  doesn’t matter much, the purpose of the current rev is to figure out control, not splurge on fancy MOSFETS.

BIG OLE Allegro hall effect current sensor – senses current through the whole bridge.

PCI-e X4 connectors – taking a leaf out of Jume’s book.  Cheaper than most other edge connectors as they are produced in absurd quantities for computers.  I measured the resistance per pin to be very low – a couple milliohms.

Why edge connectors?  This Three Phase Bridge Of Learning (TM) has not-quite-hot-swappable phase legs to make for easy replacement of a blown phase.  So far, I’ve configured the delay on the gate driver such that no combination of logic inputs will cause it to shoot through.

I don’t yet have the current sensor set up for integrating current over time.

treephasor_screenshot daughterboard_screenshot


I wrote some very awful block commutation code that uses conditional logic to figure out which FETs to turn on.  That was enough to get the motor spinning:

It draws a rather absurd amount of no-load current due to delays in my block commutation code.  Bayley showed me a more MCU-friendly way of making a state machine that runs much faster.  After picking out some errors in the commutation table, the motor runs happily with 30v on the bus, only drawing about half an amp at no load.

The motor I’m using here is one that Ben crufted, a beautiful specimen of a brushless inrunner, probably meant for servoing something.

IMG_1720 IMG_1718