• State Resolved
  • Locked Locked
  • Replies 1 reply
  • Subscribers 82 subscribers
  • Views 3792 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Using Hall-Sensor for startup and switch to InstaSPIN's sensorless FOC when running

Philipp L.
Intellectual 795 points
Other Parts Discussed in Thread: MOTORWARE, DRV8312

Hello everybody,

We have prototype electric car and want to change our inverter control to InstaSPIN. This produces some questions, as the hardware already exists and we have to look for ways how it can fit into the InstaSPIN-environment.

The first problem is, that in control of electric cars, sensorless algorithms are not common, because they are not considered safe and reliable enough. Our motor (PM, ca. 10 kW) has a built-in hall-sensor (60 degree-signals), and we would like to use this sensor. As far as I have read in the User-Guide, only incremental encoders are supported.

Now we have some alternatives: 

- Implement our own algorithm for startup and switch then to InstaSPIN-FOC, and use the Hall-Signals to supervise InstaSPIN's angle estimation

- Directly use InstaSPIN's algorithms (also for startup) and only use the Hall-Signals for supervision

- Not use InstaSPIN for this application

What would you recommend us to do?

Would it be possible to use InstaSPIN BLDC for startup, and then switch to FOC? Or is this a good idea? It should be possible to use InstaSPIN BLDC with a Hall-Sensor, right?

Is the InstaSPIN-Startup-procedure still only an open-loop-algorithm, or is there a closed-loop (signal injection) method in the newest version?

Thanks for your answers!

Philipp

  • 0
    TI__Guru** 119380 points

    Philipp L. said:
    sensorless algorithms are not common, because they are not considered safe and reliable enough.

    This is not true. I would argue that GOOD ones are many times more reliable than sensors with much lower chance of "breaking".  And they are being used in vehicles.

    The challenge however is in the initial start-up from 0 speed of the rotor until a software algorithm / observer / estimator can take over. With the FAST observer in InstaSPIN-FOC & -MOTION we can take over at a VERY low frequency (best in class), especially with high flux motors typically used in high torque applications.

    But the initial start-up with the InstaSPIN solutions today is still not acceptable for some applications, mobility/traction being the one where a smooth start is critical.  I've posted several times how we are producing additional algorithms that will detect the rotor position at zero and very low frequencies, allowing perfect start-up of a fully sensorless system.  We plan to roll-out first example in Oct/Nov MotorWare release and then expand this in 2015.  We call these IPD (Initial Position Detection) algorithms.

    BUT, considering that you have Hall sensors in your system I agree that a great first step is to use them to start the motor in a consistent manner before allowing FAST to start providing the feedback.  We do this today in applications like e-Scooters.  In fact we are using the hall sensors to create an angle estimate and still running the InstaSPIN-FOC control system as-is.

    Unfortunately we have not released these example projects into MotorWare.  It's on the to-do list and I agree it would make the transition to sensorless or sensorless + IPD easier and more flexible.

    Several months ago I did however post a .zip that was the basis for this Hall sensor FOC work and it uses DRV8312 + InstaSPIN-MOTION proj_lab12 as the basis.  Several customers have used that - unsupported - to integrate into their own project.

    Philipp L. said:
    Would it be possible to use InstaSPIN BLDC for startup, and then switch to FOC? Or is this a good idea? It should be possible to use InstaSPIN BLDC with a Hall-Sensor, right?

    No, this will not work. Sensorless -BLDC is poorer at start-up.

    Philipp L. said:
    Is the InstaSPIN-Startup-procedure still only an open-loop-algorithm

    It's not quite open loop, but the ForceAngle start-up feature does simple give a "dummy" angle estimate feedback in an attempt to enable the control system to get the motor to start moving.  In larger flux machines it's amazing how quickly FAST can lock on to the angle estimate with just a TINY bit of movement.  But this ForceAngle start-up can be "harsher" with more of a "kick". It basically makes the system keep fighting to start-up until it does, which can be unpleasant in an application like an e-Scooter.

    Philipp L. said:
    or is there a closed-loop (signal injection) method in the newest version?

    This is one of the methods for IPD that we will release. Injection techniques rely on the motor design to have high saliency (Ls_d >> Ls_q), typically created by embedding permanent magnet material below the surface of the rotor to give an assymetric reflection of the injected current.  These techniques ONLY work at very low frequency, so they must be paired with a good observer (like FAST) to operate at higher speeds.  There is effort in tuning the injection technique and then effort in the transition to FAST.

    Philipp L. said:
    What would you recommend us to do?

    I would use InstaSPIN-FOC and add logic for the Hall Sensors to provide the angle and speed feedback at zero and low frequency until you can switch to FAST (this may be at 1 Hz, 5 Hz, 10Hz, 20 Hz, etc. will have to test).  This is the best solution, we know it will work, and gives a path to try the IPD solutions as we release.