Back-phasing damping reduces settling time in stepper motors by over 60%. And, it’s easy to implement.
Getting step motors up to speed has never been a problem. Getting them to stop is a different matter.
Step motors (because of their magnetodynamics) are naturally springy, so they tend to vibrate at the end of each move cycle after the final pulse. In applications requiring high throughput, the relatively long settling times take a big chunk out of the motion cycle bandwidth.
The obvious solution to this problem is to make step motors less springy; that is, find a way to reduce the vibrations at the end of the move cycle. Several techniques have been developed, from mechanical dampers to electronic switching schemes such as back-phasing damping and delayed-last-step damping. Control methods such as S-curve acceleration and deceleration are also used, especially when it comes to generating ramp up/down motion profiles.
While back-phasing damping is not the only method available to reduce settling time, it has the advantage of not causing large overshoot. As a result, vibrations and motor noise are stifled more quickly due to faster motor stopping.
In the back-phasing method, a delayed pulse opposes the rotor, causing it to accelerate from the equilibrium point of the first phase to that of the second phase. To avoid overshoot, a second delayed pulse is applied (this time in the first phase) when the rotor enters the vicinity of the second phase equilibrium point. In this method, “n+2” pulse commands are required when controlling a motor with “n-step” motion.
Learn by example
The following sections outline an example demonstration in which back-phasing damping significantly improves settling times and reduces the amount of variance between times. Because the results obtained in these demonstrations differ from usage conditions in actual motion environments (such as driver, motor, frequency, and load), back-phasing damping in other instances will vary substantially. However, under specific circumstances, namely iterative, uniform motion with homogenous load conditions, back-phasing damping can reduce settling times and therefore improve system speed and productivity.
In this demo, experimental testing was done using the back-phasing damping method to see how much settling time could be reduced compared with normal operation (no damping). Motor behavior was measured using a system that samples encoder output (the same pulses driving the motor) every 0.1 msec. For example, if the motor were run at 500 pps, the theoretical number of encoder output pulses would be.
This number of pulses represents the sampling resolution for each sampling period.
When the motor is run at 500 pps and 50 pulses are fed, the movement completion time (output finishing time) will be.
To determine the method of analysis, the normal operating (no damping) response was compared to the response obtained when back-phasing damping was performed. The method of analysis was then determined based on the degree to which settling time had been reduced. The settling time was defined as the time required for the motor to stop all vibration, beginning from the end of the last output pulse (100 msec), until the vibration had been damped to less than ±5% of the motor’s step angle (0.72°) specification.
In this test, the pulse width used for motor movement was set at 200 μsec. The back-phasing procedure:
1. Last CW pulse
2. “Timer 1” actuation
3. Output of one CCW pulse (50 μsec)
4. “Timer 2” actuation
5. Output of one CW pulse (50 μsec)
6. Output actuation has been completed
Continue on page 2
Won’t settle for less
In the normal operating (no damping) response waveform, settling time is 32.8 msec (132.8-100 = 32.8). Because the bipolar type driver drives the five-phase stepper motor with full step mode (4-4 ex), there are ten different excitation sequences generated (current flows in five coils in both directions) producing ten different settling times. Settling time was measured for five excitation sequences (instead of ten) because current characteristics are the same in either direction. Each excitation sequence shows that excitation proceeded sequenceby- sequence, beginning from the initial status of excitation at the time of power activation (initial stage).
When the back-phasing damping method is used, pulse input timing is depicted for input pulses. As previously described, because there are five “stopping sequences,” pulse timing must be varied according to each sequence (excitation phase). To find the proper pulse timing for each phase, we have checked the delay of the rotor movement and then converted the phase difference of each phase. The phase differences appear as variances in settling times as the motor stops.
The pulse timing is set for each phase of the experiment as follows: The settling time obtained was 110 msec, beginning from the end of the last output pulse (100 msec). Additionally, the response waveform indicates that the motor slowed down smoothly until stopping, with no overshoot.
In order to perform the backphasing damping under the same conditions of the normal operating (no damping) response, timer conditions were fixed and the excitation consequence was varied consecutively. The average settling time obtained was 12.64 msec, beginning from the end of the last output pulse (100 msec). Considering that when the back-phasing damping was not used the average settling time was 36.32 msec, this new result yielded a 65.2% reduction in settling time.
Ideally, all summed values match, regardless of the excitation sequence or control condition, and there is no variation (all graph bars should be the same length) due to the number of times measured. Therefore, when any one of the five types of control conditions are used, the settling time is still reduced and variance between settling times is minimized.
In addition, by using the most appropriate control condition (timer 1: 0.90 msec, timer 2 = 0.45 msec) an average settling time of 12.24 msec was achieved, with a difference of only 3.7 msec between minimum and maximum settling times.
As a result of these experiments, 142.2 msec passed between the starting and the stopping of the motor (back-phasing damping method was not used). When the back-phasing damping was used, the time period was reduced to 117.7 msec. The difference between the 142.2 and the 117.7 results in a reduction in settling time of 24.5 msec. This value is the settling time reduction for one single motion. If this reduction were calculated over the workload of an entire day, time savings would be significant.
Kazuya Endo is a chief engineer and Fumihiro Kobayashi is an engineer for Nippon Pulse Motor Co. Ltd., Tokyo, Japan.