Manufacturing requires increasingly productive and flexible machines, so many machine builders are switching from stepper to servo technology. Although steppers have their advantages, servos often increase productivity and reliability over time.
However, it can be a difficult switch, as servo tuning usually requires special expertise. How can a designer determine whether the switch will be worth it for a particular application? That depends on three things.
- Torque and speed requirements
Can a switch be made without significant changes to the machine design? Is it dimensionally acceptable within the current machine design? The servo system should meet or exceed standard specifications, and servos replacing steppers must operate at both high torques and speeds.
Is it cost effective to replace a machine's steppers with servomotors? How much of a price difference will result, and is that acceptable?
- Tuning and ease of startup
How easy is it to tune and start up a servo system? Stepper systems do not require tuning, so commissioning a traditional servo system often requires a new skill set. The general opinion of machine builders and end users is that servos are difficult to tune to specifications, especially when there is mechanical compliance, large inertial loads, or when fast settling time and high holding torque are required. We'll review how this isn't necessarily so shortly.
Addressing conversion issues
Torque and speed requirements: There's no doubt that servo systems provide high torques at wider speed ranges than stepped systems. Some new designs are particularly compact and efficient — with up to 30% fewer parts than predecessors, to improve reliability. What's more, improved feedback designs have doubled motor ratings for shock and vibration in these systems, and cutting-edge stator and winding designs reduce motor size for given power levels. Such servo systems, capable of high speeds, help increase manufacturing throughput; wider torque ranges make machines flexible enough to handle products requiring different torques.
Often overlooked when sizing step motors is allowing for enough torque margin. Step motor systems are open loop by nature. If torque requirements exceed what the system can deliver, the motor stops turning and stalls. Without position feedback, the controller incorrectly assumes that the motor has reached the target position. On many machines, motor stall conditions and position loss lead to increased scrap rates and downtime — and even physical interference and machine damage. These consequences all increase overall stepper-system cost of ownership. To guard against this, engineers typically must add 50% torque margin to worst-case move scenarios when selecting step motors; unfortunately, this leads to much larger motors than necessary.
Cost: Stepper motors may seem cheaper than servomotors. However, investment in servo systems can pay off in the long run, especially on systems with high mean time between failures, even to 20-plus years in some cases.
Over time, the higher initial cost of a servo system can be also justified by increased productivity. For example, 20-bit encoders installed on some servomotors offer resolution of more than one million counts per revolution — clear indication of how accurately positioning can be executed in a closed-loop format. Such high resolution makes for more accurate positioning and production quality. Some servo systems can also operate at higher speeds and at higher torques, which helps in accelerating process time and increasing plant throughput.
Easy startup and tuning: Tuning a servo — adjusting feedback gains to satisfy output performance specifications such as overshoot limits, settling time, or position error — is considered an art by many. Tuning gets complicated if the load inertia is much higher than the inertia of the motor being used to drive the load. Once inertias are calculated, output specifications can be met by adjusting feedback gains using equations that model the machine's physical and electrical components. This is the most complex and time-consuming part of tuning, and the reason why many designers shy away from the process.
One alternative is newer on-board tuning techniques that solve equations and execute optimization routines internally with three different servo-tuning options.
Tuningless functionality is usually set as the control default. Useful for general-purpose axes not requiring very specific responses, this allows for consistent performance without requiring designers to adjust feedback gains. Adaptive features — performed by varying torque during acceleration and deceleration — continually adjust to provide satisfactory performance even if the load changes 20-fold.
Autotuning is used for stricter performance specifications. This level of tuning is best executed using software tools. Some manufacturers offer free software to help set up, tune, and monitor parameters; here, autotuning is just one of several configuration wizards.
Advanced system identification techniques are employed in this part of the software. Once the moment of inertia calculator is turned on, automated test commands excite all modes of the system used for physical movement. The system response is then captured and analyzed to estimate the attached load's moment of inertia to high accuracy.
Next, autotuning guides designers through a series of steps in which information is collected about the system type and application mode. Based on the mode (position, velocity, or torque) unique reference signals are used to automatically tune the system by driving the servo while feedback is monitored to gauge the level of gain optimization. Various sets of reference signals are used to fully tune the system (through steps such as oscillation level measurement) and adapt feedback gains in real time so that the best possible system responses are obtained.
Also performed by upper-level host controllers, advanced autotuning can be used to place filters (torque reference and notch filters), compensate for friction, control anti-resonance, and suppress vibration. When enabled, vibration suppression functions detect and suppress vibration in the 0 to 100-Hz range commonly generated during particularly fast positioning moves. This reduces settling time while preserving accurate positioning.
One-parameter fine-tuning is the next level of tuning. Here, designers manually adjust gains and filters and fine tune machine response after autotuning. Depending on the tuning level required, several parameters are optimized through adjustment of a single fine-tune parameter.
Tuning modes can be selected based on features like stability, responsiveness, or overshoot suppression. A machine resonance filter selection can be determined according to the actual machine assembly. Notch filters found can effectively avoid areas of frequency resonance that can otherwise cause servo instability and inefficient performance. Adding one-parameter tuning to the tuning process can also cut settling time in half compared to that of autotuning alone.
For more information, call (800) 927-5292 or visit yaskawa.com.
Comparable size, comparing designs
This screen shows a scope trace of motor performance in response to a step velocity command. It is used to compare actual results to test moves during tuning. The commanded profile goes immediately from zero to some set velocity — and instantaneous acceleration is the toughest move profile possible. Actual motor response traces the commanded profile extremely well — indicating a well-tuned system with fast settling time and little or no ring out. In addition, the torque command (in purple) is stable, showing high response without associated buzz.
A power density comparison between our two similarly sized motors illustrates the differences in their speed vs. torque curves. The servo provides a wide torque range at high speeds, despite its slightly smaller package. The solid blue curve on the servo plot is the instantaneous torque curve. The dashed blue curve on the servo plot denotes the continuous torque provided by the motor (available infinitely). The servo system delivers high torques at speeds up to 6,000 rpm. Such speeds are not even achievable using steppers, which top out at approximately 2,000 rpm, bleeding available torque along the way.
Pick a model, any model
Compare the reference velocity plots (in purple) and feedback velocity plots (in green.) To go from the response in the first plot to the much-improved response in the second takes just a few minutes of interactive work with a tuningless software tool, and one mouseclick to enable tuningless functions. In the latter, feedback velocity plots right over reference velocity. To achieve the quick, sophisticated response in the third plot can take a day or so of an expert tuner's time — or just a few minutes for a novice designer with tuning software. This particular plot is actual output from a system after autotuning.