I am having some trouble in reversing the direction of a sensored 10KW BLDC motor. Motor has 36 stator teeth and 40 magnets (20 pole pairs)

Apparently the hall sensors position is set for CW direction, and when rotating the motor CCW, the timing is wrong and efficiency is low.

Timing cannot be adjusted in controller (Kelly controller), so I am re-positioning sensors.

The problem is, that while for CW rotation sensors are placed inside stator slots(see image below), CCW sensor position - in slot

...for CCW correct position (measured with an OSC) position never matches a slot. Motor manufacturer tried to place sensors for CCW rotation like this (illustration, ignore zip tie): sensor on top of stator teeth (please ignore zip tie, this was for testing) This looked ok unloaded, but was creating a lot of heat and consuming too much power as I increased load. I assume this is because the sensor is placed halfway between magnets and one coil, and therefore shows zero reading when coil polarity is same and stronger than magnet passing in front of it - Can anyone validate this theory?

The new placement I am experimenting with to solve this is under the magnets on the bottom side of the motor, with the 3 sensors 6 degrees apart on one small plate, facing the magnet narrow end: bottom sensor board sensor board front

Haven't tested this under load yet, but unloaded test shows the following behavior: While testing unloaded at 20v, RPM increases as current increases up to around 2A, then when increasing throttle more, current drops to around 1.8A, and RPM increases as you increase throttle up to the expected max RPM for voltage.

This also happens on the CCW motors with sensors on top of stator.

On CW motors, current gradually increases to 3A for max RPM.

Can anyone tell why this is happening?

I would appreciate any comments on this arrangement and possible solutions.

  • \$\begingroup\$ Have you checked the controller datasheet for options? Can you set advance in the controller instead of the sensors? \$\endgroup\$ Commented Aug 31, 2018 at 13:01
  • \$\begingroup\$ Checked and also contacted the manufacturer, unfortunately these controllers (KHB from Kelly controller) do not have timing adjustment options. \$\endgroup\$
    – K.R.
    Commented Sep 1, 2018 at 10:03
  • \$\begingroup\$ I understand that the sensors are placed so as to create some phase advance in one direction. Depending on what you are doing with the motor, that could make sense. However, because of symmetry it should still be possible to place them in stator gaps. It doesn't make sense that the new correct location would never be in a gap. In my experience, moving the sensors too far away from the stator is not good, because they will be in the fringe of the magnetic field, and you will not have reliable timing for the hall signals. \$\endgroup\$
    – user57037
    Commented Sep 16, 2018 at 15:58

2 Answers 2


Apparently the hall sensors position is set for CW direction

Placement of such sensors are direction agnostic as electrical machines are 4quadrant electromagnetic devices.

If this is sensor related I strongly suspect the sensors are wrongly placed. Maybe intentionally biased to provide phase advance in the CW direction (thus phase retard in the CCW direction)

What is more likely is any offset within the controller or wiring of the sensors to the controller is wrong.

Can you state why you think the sensors are wrong

  • \$\begingroup\$ Why would it be wrongly placed to add phase advance for the CW direction? It is really up to the device manufacturer how they want to implement phase advance. \$\endgroup\$ Commented Aug 31, 2018 at 12:57
  • \$\begingroup\$ It might be wrongly placed fpr the OP usage (if this is sensor related). This might be the design case for this.motor but there is limited information from the OP. \$\endgroup\$
    – user16222
    Commented Aug 31, 2018 at 13:00
  • \$\begingroup\$ While checking phase vs sensor sync with an oscilloscope, if I position sensors to be synced with phase while rotating the motor to a certain direction, and then test again the other rotation direction, a small offset from sync is visible, so I assume position cannot be the same for both directions, even without deliberate advance timing. I will try to post some images tomorrow. \$\endgroup\$
    – K.R.
    Commented Sep 1, 2018 at 10:08

After doing some more measurement on the CW motors, it appears that sensors are placed with a considerable advance from phase.

The correct position for CCW was found to be of similar advance (to the other direction naturally).

However, the solution shown above for placing the sensors below the magnets on a single plate perpendicular to motor axis did not work. Above 1/3 power the motor would start to cut intermittently. I can only assume that at this position /orientation the coil field at high power interferes with reading.

As mentioned in the question, no slot position provided the correct timing, but the solution that I found eventually was to place the sensors in slots with sensor face not tangent to the stator circumference: By rotating each sensor inside a slot, I was able to advance or retard timing and calibrate the desired amount.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.