Hi everybody! I trying to design a motor controller for a Motenergy motor with the following characteristics:

PMAC motor, with sinusoidal back-EMF.
- 24V - 102V DC applications allowed with BLDC motor controller.
- Rated Continuous Current 140A.
- Peak current 400A.
- Speed: 0 - 5000rpm
- Sin/cos speed enconder

Application would be an electric motorcycle.

I'm trying to decide which method is better to drive this motor:
I've discarded trapezoidal control because of the ripple that affects motor torque (smooth movement desired). Now, i'm not sure if sinusoidal control is good enough or I should think about FOC control. I know that sinusoidal control has some problems with high rotation speeds, but, what is considered a high rotation speed?

I would be very grateful if someone can throw some light on this.

High rotational speed is when the RL lag of the stator windings skews the current waveform too far from the ideal phasing. Your controller pay be able to phase advance the commutation to overcome this.

Since you have a sensor available, FOC is straight forward and the way to go.

You speak as though sinusoidal and FOC are mutually exclusive - they are not. Perhaps you mean phase controller vs FOC. Almost any sinus drive system will work fine for you - you are going to operate in torque mode and the largest torque ripple that you will see is from the current sensors that you use.

Thank you everybody for your answers. Bill, looking at your answer, yes, I've been thinking that sinusoidal control and FOC are mutually exlusive, in the way that sinusoidal control is based in compute the initial phase and then source the motor with tree sinusoidal signals with a phase separation of 120º between then and FOC implies controlling flux and torque vectors in a direct - quadrature space... maybe I'm mistaken about this concept... In addition I've seen that FOC approach seems a little bit more complicated to implement.
Is there something wrong in my reasoning?

No, Santiago - your about right except that we call your first description a phase controller - 3 independent current loops. These can be sinus or trapazoidal (or anything else -6 step). The FOC uses PARK/CLARK and usually SVM to do DC math in a reference frame. I don't know of anyone who makes an FOC that is not sinusoidal. So, both implementations can product sinusoidal waveforms to the motor and this is why I said that Sine wave is not mutually exclusive with FOC.

Ok, I understand. I was naming it in the wrong way. So, my question should be, if I can correctly drive my motor with a phase controller (with sinusoidal waves) or I need to use FOC for any reason.
For implementation I was thinking about using a dsPIC (I have found some starter kits, like DM183021, that looks like a good point to start) or maybe something specific from Atmel. This brings me to another doubt. I've read several recommendations about using a DSC instead of a general purpose microcontroller for motor driving applications. I can see this logical for a method requiring great computational power (maybe FOC), but I think it is unnecessary for a phase controller (trapezoidal or sinusoidal, taking into account that sinusoidal waves can be generated using a look-up table). What do you think about this?

For your application a FOC doesn't bring much to the table.... maybe a little more speed for SVM for a given battery voltage. You're application is not demanding. (THat said, FOC is easy to implement so why not do it...).

dsPICs are well supported and documented AN957 (Great doc / software with bugs) and others describe FOC and SVM well. You not going to get much out of a DSP over a micro controller. Why? The only place it makes a difference is computing the PI math and your application is not demmnding - you don't need to process data very fast because you are not going to need big current loop bandwidths. Again, big deal - cost isn't that much different (You may think otherwise). dsPIC30F is great for what you want.

Particular microcontrollers (like TI C2000 series) also have richer peripherals, that can make them more suitable to motor/power control.
However, for a standard FOC at a reasonable updating frequency (for a traction application), as Bill said, a microcontroller, if application notes describe FOC implementated on it, should be OK.

FOC can get the best from your motor in terms of torque dynamics, smoothness and torque vs. speed curve. Moreover, implementing it can be a good "investment" for future applications...

Thank you to all of you for your answers. I've found them very helpful.