[sdiy] OT BLDC motor controller - a MOSFET question

[Tom Adam]

Op 22-12-2010 21:36, ASSI schreef:
> On Wednesday 22 December 2010, Tom Adam wrote:
>> A nephew of mine is working on his thesis and is having a problem.
>> He's building a brushless DC motor controller with an arduino. He's got
>> a working prototype, so far so good.
>> For his prototype he's using a 25W BLDC motor, but in the end he needs
>> to drive an 800W BLDC motor (36V, +/-22A).
> At that voltage/power point things will be fairly interesting, at least if
> you want good efficiency (you will likely need better than 90% just for
> thermal reasons).  If the driver and control stage are not optimized
> together, the driver would have to be overdimensioned quite a bit just to
> survive the switching stress.  What's the application, an E-Bike or
> something?
>
Yes, this is actually for a rather high speed robot. Just for moving the 
robot around. So you could call it an E-bike;-)
This is not the main part of his thesis, but he'd like to make one, not 
buy one. Although he needs to make a decision now as time is running out..
>> Currently  he's using the IRF1205(N)&  IRF5305 (P) MOSFET. And we're not
>> sure these MOSFETS can handle the 800W motor.
> It is somewhat unusual to use a PMOS in such an application, as the expense
> of a high-side driving circuit (and gate boost supply) is usually outweighed
> by the much better performance of an NMOS.  Also, since you're driving an
> inductive load, you need to be really careful at how exactly you switch
> things off (that is much trickier than you might think and you will learn a
> lot about parasitics you didn't even know before).  At this point your
> controller should know something about the driver and provide suitable dead-
> and overlap-times to reduce stress.  High power drivers are usually designed
> to take at least some of the inductive energy away in a controlled breakdown
> (that's what the avalanche and repetitive avalanche energies are for in the
> data sheets), so you're really close to what physics allow you to do with
> the device and the margin of error is just tiny.  At 800W that'd literally
> mean fireworks, so you don't really want to cross the edge.
>
Actually, I'd like a good firework ;-)

>> So how does one select a MOSFET? Any pointers appreciated...
> A good starting point would be the BLDC development and evaluation kits that
> are out there from practically all manufacturers -- especially if the driver
> circuitry is not explicitly a part of the thesis, I'd not really want to try
> to design and debug an almost 1kW power stage.  If you want to learn it
> anyway, there have been numerous articles on BLDC control in EE Times and
> their sister publications over the past years, some of which also discuss
> the various aspects of how to drive and select high-power MOSFET.  There's a
> ton of application notes that make a good reading, too.  Just don't expect
> to go straight from 25W to 800W and have it all working without a hitch.
>
DO you happen to know a good development board that does handle 800W? I 
was looking on the web a bit, but couldn't find one fast. I'll keep on 
looking...
I did find a lot of app notes that I will read in the next days. I'm 
really getting interested in this technology. Now to find a way to 
incorporate an BLDC motor in my modular ;-)
Thanks for the help.

Cheers,
Tom
> Achim.