Hi Nicholas, I suggest you see link and attached files in topic.

SimpleBGC 32bit 3-Axis Software User Manual - BaseCam Electronics

https://www.basecamelectronics.com/.../SimpleBGC_32bit_manual... Best regards

Dear Zeashan Khan ,

Thank you, but as I said, there is no PWM. The field windings are driven by linear analogue power amplifiers, to provide a perfectly smooth sinusoidal waveforms, with very accurate control over the phase angle.

For slow positioning control, the angular velocity is essentially zero, so there is no back-EMF generated by the motor windings, so the back-EMF cannot be used as part of the control algorithm calculation.

Like all BLDC gimbal motors, the motor is designed to operate in the "stalled" region where the negligible angular velocity means the impedance of the windings becomes purely resitive, with zero inductance (and hence no back-EMF is generated).

I'm trying to do extreme positional control to 0.0044 degrees = 15.8 arc seconds, for a surgical tool. Rotational speed is irrelevant.

Dear Mohamed-Mourad Lafifi ,

Thank you for the links.

Whilst this information made for interesting reading, unless I have missed something, none of them seems to actually tell me how to do accurate position control of a brushless DC (BLDC) motor.

The basecam information seems to be instructions on how to use some closed-source software, and it does not reveal what proprietary control algorithm is actually used inside their product.

The ST Microelectronics datasheet provides very little detail regarding the control algorithm that is needed, and no equations.

The Thesis is based on a simple brushed DC motor, so that doesn't tell me anything about how to control a brushless DC motor (with three winding signals).

Dear Aparna Sathya Murthy ,

I'm not quite sure I understand what you mean by " do you have motor commutation circuitry ?? ".

The current in each of the three motor windings are controlled by DACs, under the control of a microcontroller.

There are no hall sensors in the motor.

The output shaft's angular position is measured by an high-precision optical encoder (with 26 bits of angular resolution).

To achieve maximum torque in a BLDC motor the flux angle must lead the rotor angle by 90 degrees. However, as the encoder starts off being mounted at an arbitrary mechanical angular offset to the actual motor windings, one of the first tasks would be to electronically measure where that angle actually is. Once that is done, the optical encoder can be used to do the commutation, but not before.

Having achieved the correct commutation angle, this would get us to the point where a speed controller could be implemented. However, I need a position controller, and I can't find any literature that describes the control algorithm for that with a BLDC motor.

Dear Aparna Sathya Murthy,

Thank you for the link, however unfortunately the information it links to only relates to PWM speed control using back-EMF, none of which is applicable to solving the question I asked.

There is no PWM, there is no back-EMF voltage to measure, and I'm not trying to implement a speed controller.

If the index position of the optical encoder is unknown, you are only able to implement some "relative position control".

On the bright side: you are free to choose your reference position. The usual way to get this reference position is trivial: power one or two windings with enough (DC) current to get the rotor rotated into the corresponding position. This is your reference position. Any relative position can be calculated by subtracting the encoder's reference code from the current encoder's code.

This is a scheme often used for BLDC control (no matter what position feedback is used) to get the rotor into some "start position". The only precondition is that the motor load has to be sufficiently low so the rotor is mechanically capable to reach the start position.

Dear U. Dreher ,

Thank you. Yes, something like that will be needed.

I think I might need to enhance that technique slightly to achieve the extreme angular accuracy I require. I need to overcome any rotor friction, and the fact that when the rotor angle approaches the phase angle, the torque drops to zero, so it never "quite" reaches the correct reference position.

However, I have devised the cunning plan of modulating the field angle with a sinewave, so the rotor position oscillates backwards and forwards over a few degrees around the reference position. If I take an average of the resulting physical angular excursion values (as measured by the optical encoder) then this dynamic technique should give a more precise measure of the reference position than applying DC values to a couple of the coils.

The longer I leave it oscillating and averaging, the more exact the reference angle should become.

Having identified the reference angle, I will then be able to apply the required 90 degrees of phase lead to the flux angle to implement proper commuation.

The next step is to work out how to do the position control loop itself.

dear nicholas

for measure the rotor position, u can try with 3 phase sinewave( or 3 squarewave) with low frequency to high frequency. commonly to measure the bldc position using current probe/ shunt resistant, then give 3 phase square wave. When the motor start the position is same with voltage u give. as for encoder the 360 degree was divide by the resolution and can be the position reference. notes: pliz take notes the min frequency for motor can rotate and make equation.. the more frequency the more fast it rotates *at right position

Dear Dwi dharma Artakusuma ,

Thank you for trying to help. However, I have been unable to fully understand your English. Sorry.

I believe the 12N14P configuration will have 12 stator magnets and 14 poles. The number of pole pairs determines the ratio between mechanical degrees and electrical degrees so in this case with 14 poles there are 7 pole pairs. I believe you will see one mechanical rotation after 7 electrical cycles.

You state “The amount of current fed through the three field windings are independently controlled by 16-bit DACs.“ I assume this means you have some three phase servo amplifier that has three voltage set point inputs and produces three independent and proportional currents, but that it will be up to you to determine how current commutates from one pair of windings to the next. If this were an AC servo motor you would simply generate three sinusoidal reference waveforms that are phased 120 degrees with respect to each other. For a BLDC motor a six step trapezoidal waveform may be good enough. If you need to minimize cogging torque you can optimize the drive waveform by back driving the motor and measuring back emf with an oscilloscope. This will take some extra work and I won’t go into detail.

Because the motor does not have hall sensors, no matter what waveform you use, you will need to use the encoder feedback to determine commutation state. You will most likely want to create a lookup table to store a current setpoint waveform. You will use the encoder feedback to index into the lookup table to determine the relative proportions of current for each winding.

In any case, if you are properly commutating the windings then air gap torque will be directly proportional to current. The relationship between air gap torque and angular position of the motor shaft can be approximated as a second order differential equation. In most cases torque produced is balanced by angular acceleration of rotor and load inertia, viscous losses in motor and load, and static torque.

You will need to implement a position control loop that has a single output and the output of this loop will be used to scale the setpoints obtained for each winding from the lookup table. If the control law output is zero the motor will not produce any torque. More current -> more torque -> more acceleration. Negative current produces torque that will decelerate (or accelerate in reverse direction).

I prefer using state variable feedback for position control. I would implement a second order state space observer and a second order state space control law using pole placement or a similar approach to define closed loop response. You will need to augment the state space system with an integrator to drive steady state error to zero. You will also want to implement an integrator anti-windup feature or the system may become unstable if the control system demands more torque than the motor and/or amplifier can produce.

You could also use PI or PID ... it’s easier and quicker to implement but end result may not be as good and it will be more difficult to prove stability. Because you are interested in auto tuning this may be an important consideration. I would use system identification to estimate parameters for the second order system. With state variable feedback it is straightforward to calculate controller gains once you have estimated model parameters.

To determine motor phasing you can rotate motor at low velocity until you detect encoder index pulse. This works as long as you don’t have limit stops etc. You could combine the phase alignment with excitation needed for auto tuning. You would need to record applied current and measured position for some period of time. Then perform regression analysis to estimate inertia, friction, and static torque.

From the IMU we get position and velocity after sensor funsion. The input variable of the plant seems to be the speed of the gimbal motor (since its driven by 3 sine waves), not the voltage as usual in a brushed dc motor setup. So one can not just implement a torque control for holding the position. When I investigated the Storm32BGC, it looks like the PID does influence the position and the speed at the same time: Fast movements leads to fast speed corrections, but also the deviation from the stationary position leads to a given speed.