filtering servo motor- reducing velocity tracking error

copyright 2011 controltrix corp www. controltrix.com

www.controltrix.com

Servo Motor Drive Velocity Tracking

Reducing tracking error


Framework

PMSM vector control drive

Velocity Feedback:

• Encoder with finite pulses/rev (e.g. 10000)

Current feedback:

• LEM sensor + onchip 12 bit ADC +/-15A full range

Voltage feedback:

• DC Voltage sense + onchip 12 bit ADC 800 V full range


Framework…

Application:

• Machine tool industry

Objective:

• Control motor velocity as precisely (from 0 to 100 % or more

rated speed)

• Under external disturbances


Challenges at low speed operation

• Limit abs. err. @ Slow velocities (small value large % err)

• Limit Rel. err. @ faster velocities (small % large err)

• Rel. and Abs velocity tracking error

• Abs err. required approx 10 rev/day


Limit factors1. ADC current resolution2. Velocity feedback resolution3. Dead time4. Inductance5. Back emf distortion6. Residual encoder offset


ADC Current feedback res limits torque impulse res.

• Kt = torque constant = 0.75 Nm/A

• Ts = sampling time = 1 ms

• J = inertia = 0.00017 Kgm2

• dI = ADC current resolution = 10 mA

dT = Kt. dI

Min abs velocity resolution = Ts. dT / J approx 20 -50 rev/day

Limit factors.


Limit factors..

Mitigation :

• Oversampling and averaging to reduce quantization noise

Disadvantage :

• ADC sampling frequency and triggering limitations


Velocity feedback resolution

• Fixed time sampling (# pulses in 1 sample period)

• 1 pulse/1ms = 6 rev per min

• @ High velocities less troublesome

• @ Low velocities major problem

Limit factors…


Limit factors….

Mitigation :• Fixed angular distance/measured time • Limited by processor timer resolution • @ High velocities more trouble some • @ Low velocity is OK

Disadvantage :• Non linear / non time invariant dynamics

(LTI analysis not valid)


Deadtime

• 2 - 5% duty is wasted in deadtime

• Leads to 6 step voltage to be applied @ electrical freq.

• Non linear dynamics

• Distorts dynamics @ 6X electrical frequency + harmonics

( open loop ripple torque)

Limit factors…..


Inductance

• Inductance varies with current due to saturation



Limit factors……


Back emf distortion

• Back emf waveform is never ideal sine wave.

• Some distortion always present



Limit factors…….


Residual encoder offset

• Leads to direction dependent performance

• The angular offset between encoder 0 and motor 0 position is

always an estimate. Some residual always remain

Limit factors……..


Static friction

• Non linear direction dependent component near zero velocity

Note: 3 – 4 – 5 – 6 – 7 are not easy to mitigate unless using

repetitive control techniques

Limit factors………


Other considerations:

• Non linear dynamics with fixed distance variable time sampling

• Dynamics change with velocity


• Kalman filter based data fusion

• Motor Current and encoder data combined to estimate

instantaneous velocity

• Estimate velocity @ current sampling freq. approx (20KHz)

• No gain changes

Proposed method


• No change in sampling freq. based on velocity

(only pulses per sampling period measurement used)

• Zero velocity control

• Abs velocity resolution limited by current feedback

• Relative resolution can be arbitrarily reduced (subject to

abs limit) trade off against dynamic performance

• Other limit factors automatically taken care

Proposed method.


Simulation results

• Typ. velocity ripple @ 300 rad/s

• Abs err = 0.004 rad/s = 57 rev/day

• Rel. err. = 13ppm


• @ low rpm (6 rpm)

(abs error/ripple estimate based on current quantization)

= 0.0002 rad/s = 3 rev/day

• Rel. err. = 100 ppm

Extrapolated estimated results (simulation time is very long at slow speeds)


Thank [email protected]

filtering servo motor- reducing velocity tracking error

Technology