Blushless 3 Phase DC electronic speed controller shield

[Adam 'Weirdarms' Wiggins]

This might be a bit of a stretch but for a while now I've been looking for an open hardware design for a 3-phase DC blushless motor electronic speed controller (ESC) for hobby engines used for multicopters. The reason is you can get these things cheap but they are the PWM servo interface style controllers (some can be hacked) with no telemetry information, and no I2C. Most of these controllers are Atmel chips and more then capable of providing speed information and in some cases power usage, all of which would be handy information to have feed back to your flight computer.

Starting with a smaller motor rating (say 10A) it should be possible to make a shield (might require a headsink or to be the top/only shield, my wish list would be:

* Battery draw current sensing.
* Temperature sensing.
* 3-phases driven by N/P mosfet pair, and voltage sensing for cross-over voltage detection, and PWM output to drive the mosfets.
* Voltage sensing of battery input.

Atmel have a lot of design notes, so it shouldn't be too hard. I'd be happy to provide some ideas regarding design and software. I think you could do a range of shields to provide kit based development for various parts of a multicopter, R/C plane, or otherwise for things like the flight computer with IMU (3-axis gyro/accel/compus), wireless communication, ESCs, etc.

Let me know if it sounds interesting or not, I have seen on kickstarter some interest in the IMU sheeld which I personally would be very keen on.

Cheers, Adam

[DaveChambers]

Sounds great... I'm wondering if some of our ideas may not be completely confined by the size if we used components off the form factor. For example anything requiring a heatsink could have a header off the board to be mounted as part of the enclosure. We still get the 'quarter stack' for the main logic and I/O and have the ability to place MOSFETs and the such as part of our enclosure.

[TinyCircuits]

Adam,

A BLDC controller is a great idea!  Size will obviously be one of the limiting factors about what could be possible with this.  10Amps is probably a bit high in this form factor.

One of my thoughts for certain things like this would be to allow for remote "smart" modules, such as sensors, displays and motor controllers.  The interface back to the main stack would just be something like I2C so the cabling would be very cheap and simple, and the module itself could be different form factor to allow for heatsinking or different packaging. 

I definitely would like to build up the product offerings to exactly what you had mentioned, to have numerous easy to use building blocks, to allow you to put together exactly the functions that you need for your project, whether it's RC, a copter, data loggers, sensor reader, etc.

I'm definitely in favor of proceeding with this, although this is a bit of a higher end module that some of the others planned for the stretch goals, so it probably would not be part of the Kickstarter campaign.  I'm definitely interested in your thoughts on what the design of this should be (as well as the IMU). 

Thanks!

Ken

[Adam 'Weirdarms' Wiggins]

A varient which might cover the points both of you have raised is to have a very simple shield that then provides a connector (outside of the stacking profile) for a power module that has the MOSFETs, etc. That way you can have one shield that can you can plug different rated (10A, 30A, 100A) power modules into and down the track probably simpler AVR (or other microcontroller) boards that the power modules can plug into when you don't need the stacks. The power modules could be 'side-cars' to the shield

The thing I don't like about this is that the shield is very basic, I don't think it would really be much beyond a connectors/tracks and I think in most cases you are probably going to have a single ESC made out of one power module and one micro. In that case it just adds cost and probably space for little gain. For example there are not enough ADCs and timers in general to do more then one ESC per AVR.

If you look at commercial ESC they are pretty small, for example I have a 30A one with all the power fets on one side which nearly fix in the square profile of the TinyDuino and it uses 5 fets per phase rather then just two which clearly fit inside the profile. The other side has the AVR, etc.

I think we could for 10A probably do a single shield (no stacking the MOSFET packages are too high and probably need a heatsink), in the standard footprint. I might look at how feasible such a design is. I think both a straight (top of the stack) shield and power module side-car shield have their usage in different domains or stages of development at any rate.

[Adam 'Weirdarms' Wiggins]

As for the IMU, someone suggested in the kickstarter comments for the project an all in one 9-axis chip (http://invensense.com/mems/gyro/mpu9150.html) which looks ideal, it seems to be about $17. A simple shield with that (maybe some other telemetry like a I2C temp chip and/or barrometer) and a connector for an off-shield GPS module would be ideal.

Maybe we should split the IMU into a new topic?

[knickers]

@WeirdarmsWiggins.
I think your power module is an awesome idea and should probably be its own topic. I was thinking the same thing about being able to remotely mount just the MOSFETs.

Because I want to use the motor controller for large motors, things like scooters, bikes, motorcycles, and cars(?). This would be really easy with different power shields.

[Adam 'Weirdarms' Wiggins]

I had not thought about really high powered motors, that's a good point maybe looking at the modular power boards idea first is the best way to start. I'm currently looking for some good info on algorithms and general software design for 3 phase BLDC motors, so if anyone has any pointers please pipe up.

[amnipar]

I vote for this also, I have been looking for a very small ESC for use on a tiny quadcopter. Ordinary DC motors seem to be not powerful enough for copters, so brushless would be better.