• State
  • Locked Locked
  • Replies 3 replies
  • Subscribers 31 subscribers
  • Views 215 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

TMAG5170: TMAG5170

Karel Fryd
Prodigy 20 points
Part Number: TMAG5170

Hello, we need to use this sensor with linear motor (linear door application) for sensing position for motor commutation and closed loop position/speed control. Used commutation is vector control(sinusoidal). Up to now, we use 3 pieces of Hall sensors directly connected to microcontroller, plan is to use your sensor instead. Disadvantage of 3 Halls is, that sensors must be mechanically very precise mounted and detection states are discrete. On other side this solution is based on very fast signal delay and we can use it in closed feedback control loop.

So our idea is to use 3D Hall TMAG5170 (SPI) or 5273 (I2C). We see advatages for detection of precise commutation angle and mechanical mount is not bottleneck. But we dont know if signal delay from sensor is suitable for closed feedback control loop.

Is it possible? Do you have examples of this motor commutation with this TMAG sensor?

If you can do some meeting and explain it, it will be super.

Thank you. Karel

  • 0
    TI__Mastermind 21805 points

    Hello Karel,

    Thanks for considering to use Texas Instruments.  Per the tmeasure (conversion time) specification, we can expect all axes to be sampled at frequency 1/(50us +25us +25us)= 10kHz.  If your motor rotation frequency is below 5kHz, I think you should be fine per the nyquist sampling criterion.  As for the communication speed limitation, the SPI clock frequency max is 10MHz.  Assuming the tmag5170 is configured and you are just sending read commands while the device is in continuous conversion mode, I would expect that you could get information from the tmag5170 in as little as (1/10MHz)*(32 bits per frames)=3.2E-6s, albeit the measurement data will not refresh to a new value at that rate.

    As for some resources on using TMAG5170 with a motor, this link might be useful.

  • 0 in reply to Patrick Simmons
    Prodigy 20 points

    Hello Patrick,

    thank you for answer. I thing that we want to use 2 axis, then use TMAG5170 onchip angle computation. For 2 axis this is 13.3kSPS (without averaging, info from datasheet page 17). But I dont know what is angle computation time, can you help me with this. But Im not sure if 2axis are enough for angle computation, but I thing that yes and 3 different combinations are possible (XY,YZ,XZ), am I right? But there is another text like 8.1.3.3 Pseudo-Simultaneous Sampling on page 47, there is XYX sampling, so Im a little bit confused.

    By the way, we want to use SPI for "longer distance". We have main control board, then shielded ehthernet connector, then shielded ethernet cable, then shielded ehthernet connector, then motor board with coils(electromagnets) - on this board is Hall measurement. So its circa 10cm on main control PCB, then 20cm shielded cable, then another 10cm on motor PCB. Is this solution applicable?

    Do you have time for short meeting? We can show you our setup on camera. We want to be sure that if we use TMAG5170, it will work.

    Than you. Karel

  • 0 in reply to Karel Fryd
    TI__Mastermind 21805 points

    Hello Karel,

    If your sensor is a fixed distance from rotor, I think 2-axis may be adequate and you can choose to enable the axis plane that is most appropriate for your device orientation with respect to the motor. I would need to see a diagram to possibly suggest what might be best.

    For the XYX pseudo-simultaneous sampling, this is a compensation technique used to compensate for the same ADC sampling all axes. Since the ADC does require some finite time per measurement and can only sample and convert 1 axis at a time, the field stimulus may not be identical during both conversion periods.  By taking the average of two sample-conversion periods for one axis, you can then approximate what the axis value was when it was not being sampled.  So in XYX you would be getting the converted Y value and averaged X value that most likely would be the X value if there were two separate ADCs simultaneously sampling.

    40 cm does seem a little long for a protocol meant for pcb level communication.  However, I do see some sources on the internet saying you can use SPI over distances longer than that.  Here is an application note, that gives some tips for SPI over long distance.  Alternatively, the only other communication protocol we offer with our devices is I2C, which similarly is recommended for pcb level communication.