Sinewave Commutation
The SSt servo drives use a sinewave-based, electronic commutation method for driving brushless motors. This technique ensures that the static torque produced by the motor (for a given torque command) does not vary based upon the shaft's position. Let's describe exactly what this means: If you were to put a torque wrench on an SSt servo system's shaft and command a constant torque, the reading on the torque wrench would be the same no matter where you turned the wrench. This is because the torque produced in a three phase brushless motor (with a sinusoidal back-EMF) is defined by the following equation:
where:
ø is the electrical angle of the shaft,
Kt is the torque constant of the motor and
Ir, Is and It are the phase currents.
Notice that if the phase currents are sinusoidal:
this reduces to:
a constant independent of the shaft angle.
All sinewave commutated amplifiers will have constant torque (provided they can produce accurate currents) at low speed. Physically, what is occurring when the currents in the motor are forced to be sinusoidal (as defined above), is that the magnetic vector created by the stator is held at ninety degrees to the magnetic vector of the rotor. This ninety degree angle is where optimum torque is produced and losses are at a minimum for a given amount of current. Unfortunately, sinewave commutation alone has no method of enforcing that this magnetic angle will be maintained under dynamic conditions (when the motor is moving, accelerating, braking, etc.). This is discussed further under Closed-loop Vector Torque Control.
The SSt servo drive's sinewave commutation helps produce ultra-smooth motion free of acoustical noise and premature mechanical wear generated by torque ripple.
So, what is 6-step (hall-sensor) commutation?
Let's contrast the SSt servo drive's sinewave-commutation with the more common, 6-step commutated brushless amplifier (also know as hall-commutated, DC brushless or trapezoidal amps). In these amplifiers, the current is not controlled in a sinusoidal manner. Instead, current is forced into one phase and removed from another with the third phase left open-circuit. When the motor rotates sixty electrical degrees, based upon the rather crude physical positioning of Hall-effect commutation sensors, the phases are abruptly switched (i.e., current is then forced into the second phase and removed from the third, while the second phase connection is abruptly disconnected). So, the magnetic field is jumping in approximately sixty degree increments as opposed to the smooth, continuous rotation of the magnetic field that occurs in a sinewave commutated amplifier. What are the effects of the magnetic field jumping within the motor? Let's take a look: First, static torque ripple is produced. The vector diagrams below illustrate why it is important to keep the magnetic field at 90 degrees to the rotor:
-
- Figure 1: Torque ripple produced by a 6-step brushless servo amplifier
Torque ripple is produced by a 6-step brushless servo amplifier because the permanent magnet and electromagnetic vectors are misaligned at all but 6 points in the electrical cycle.
As you can see, the electromagnetic vector creates a force that "pulls" on the rotor. In the sinewave commutated case, this force is smoothly rotated along with the rotor rotation so that it will remain at 90 degrees to the rotor. In the 6-step diagram, the magnetic vector is almost always misaligned, leading to a reduction in torque at all but 6 points in the electrical rotation. So the torque (as a function of angle) is not constant, as shown in the diagrams below:
-
- Figure 2: Theoretical torque ripple produced by a 6-step amplifier
-
- Figure 3: Actual torque ripple produced by a 6-step amplifier
-
- Figure 4: Actual torque ripple produced by SSt servo system (sinewave commutation)
Notice, that the torque ripple for a perfectly 6-step commutated motor is 13%. However, due to several factors—Hall-sensor-to-Hall-sensor misalignment (Hall-device-to-motor package, motor package-to-Hall circuit board), hysteresis in the Hall-sensors (required to quell oscillation when the motor lands on commutation points) and localized magnetic variations in the magnetic field of the rotor near the Hall-sensor array—the actual torque variation is typically 17% to 20%. This torque ripple (inaccuracy) causes vibration, noise, mechanical wear, and greatly reduced servo performance.
A secondary effect of 6-step misalignment is a reduction in continuous torque capability due to useless heating of the motor windings. The heating of the motor coils is proportional to the square of the current, independent of whether it produces torque or not. So the portion of current in the windings that produces a magnetic field that is not ninety degrees to the rotor's magnetic field, produces harmful heat. This heating, along with the poor utilization of the copper in the motor (caused by only using two phases of the motor instead of three), significantly reduces the continuous torque capability of the motor. This means you'll need to use a larger, more expensive motor (which often also means using larger and more expensive mechanical components, such as gearboxes.) Typical experimental results show that a motor will have 15% more continuous torque capability when driven by the SSt servo drive rather than a 6-step servo amplifier.
The above analysis of the debilitating effects of 6-step commutation leaves out one important, yet seldom mentioned, effect. What happens when a phase that has current flowing through it, is disconnected (unhooked and left "open-circuit")? Ideally, the current in this phase would immediately drop to zero. Unfortunately, for systems employing 6-step amplifiers, this is physically impossible because of the inductance in the coil. As we know, the laws of physics won't allow the current in a coil to drop to zero immediately. It will continue to flow for several times the inductive time-constant of the motor windings. (This happens without arcing because the winding is not truly left open circuit—it "flies" into a catch diode in the output stage.) This effect leads to sharp torque pulsations at the commutation points, which the current loop can't fully correct. As you can imagine, this is hardly what you want inside your servo loop. The ripple and pulsing are a major source of disturbance, which typically leads to de-tuning the servo system control loops and affects settling time, tracking, acoustical noise and obviously, smoothness. In addition, for reasons that are beyond the scope of this discussion, 6-step drives also have very poor dynamic response, further reducing the fidelity of control at speed.