Closed-loop motion systems can only run as fast as their sensors let them. And there are many factors that go into sensor speed. We asked the experts to explain what makes sensors fast, and slow, and how sensors can help or hurt motion system response. Here are the answers that will speed you on your way to a faster, more productive machine.
How do you define speed in a motion system?
Tom/Heidenhain: Speed has to do with reaching a predetermined point efficiently and safely. It is defined by the application and subject to various design criteria. The encoder's role in this formula is extremely important, determining drive and system performance as well as safety.
Bob/Optek: We define speed as the ability to detect the distance an object travels, rotationally or linearly, in a given time. It also implies the ability to precisely control motor speed by monitoring the motor shaft or object being moved.
Howard/Renishaw:: Speed, simply put, is distance over time. It is specified in meters per second (m/sec) for linear motion and revolutions per minute (rpm) for rotary.
Kees/Allied Motion: There are several facets to speed: absolute system speed, speed in obtaining information from a feedback device, and the feedback rate itself. Absolute speed can vary from a few microradians per second, in an optical spectrum analyzer, for example, to tens of thousands of rpm.
In general, to get good dynamic response from a motion system, the sample time should be between 50 and 600 µsec, and the delay between capture and availability should not exceed 25% of the sample time. For minimal dead time, the position and speed signals should be available to the control loop continuously and within a few microseconds.
The high speed end is usually the easy part. A system barreling along at 15,000 rpm with some inertia will be very stable even with a limited amount of information per revolution available; as from the Hall switches in a brushless dc motor based system. Inertia at that speed forgives just about everything in the way of perturbations in the motor leads, coil switching hiccups, and so on. Also, the feedback rate is high, as much as 4.5 kHz in a 6-pole BLDC system.
Things get more interesting at very low speeds. The same inertia that worked so well at 15,000 rpm is now practically useless. The easy way out is to add inertia by installing an inertia wheel or flywheel. In some cases, however, this doesn't work. In a machine tool, for example, a flywheel can totally ruin system response time. Imagine trying to accelerate to 15,000 rpm rapidly with an inertia wheel hanging on the load.
Brian/Avtron: For digital, rotary incremental encoders, speed is defined by three parameters: shaft speed (rpm) output (pulses/rev or ppr) and maximum output frequency (Hz).
Martin/Gurley Precision: For a rotary encoder equipped motion system, speed is measured in revolutions per minute, revolutions per second, degrees per second, radians per second, and so on. For linear encoders, common units of speed measurement are inches, feet, millimeters, or meters per second. Of greatest concern to motion system designers, however, is the behavior of the system in three different speed regimes: very high speed, very low speed, and zero speed.
At very high speed, the rotating or translating components must be well balanced and typically of low mass, shaped for an acceptable moment of inertia, constructed of adequately strong materials, and securely fastened. It is often advantageous to use a noncompliant coupling between rotating encoder components and the motion axis. This may be accomplished by using a kit or modular style encoder, or by tethering a self-contained encoder's frame compliantly.
Electronic circuit performance also sometimes comes into play at very high speeds. Certain rotating applications, especially those employing air bearing spindles, maintain angular velocities well in excess of 10,000 rpm. High linear speeds for machine tools and robotics typically start at 2 m/sec for high resolution scales, though coarse tape-style linears for freight elevators and similar applications can go significantly faster.
At very low speeds, bearing and lubricant friction and stiction combined with shaft coupling windup can result in undesirable motion system behavior. In linear systems, similar considerations apply. These mechanical factors can become more acute when aggravated by low system drive torque. Smooth motion at very low speeds also depends on high encoding resolution and linearity, and may sometimes be assisted by control system filtering algorithms. Precise low speed motion in robotics, optical tracking systems, and other specialized applications often require rotary resolutions above 1,000,000 measuring steps per revolution and linear resolutions in the submicron range, with near zero velocities.
At nominal zero speed, motion systems often depend on mechanical rigidity and high encoder resolution. This helps them come to a controlled halt without over or under-shooting the desired stopping position and it helps eliminate subsequent “hunting” behavior.
Jeremy/Baumer: Speed actually has three definitions in a motion system. Throughput speed is the degree of rapidity with which a motion system operates. This is typically reflected in a specified number of units produced per unit of time, or a similar measurement. Mechanical speed is the speed at which individual components can operate. Motors, encoders, resolvers, and other components each have a maximum speed above which they may fail. Data transmission speed is speed of communication between the various control devices. A system's mechanical speed is limited overall by the maximum communication speed of the slowest component in the system.
What sensor attributes are most tightly linked to speed?
Tom/Heidenhain: The four main attributes of an encoder as it relates to the speed of a motion system are resolution, signal quality, mechanical rigidity of the encoder mount, and the transmission speed of the measured data. Regarding signal quality and mechanics, inadequate scanning, contamination of the measuring standard, and insufficient signal conditioning can cause sensor signals to deviate from their ideal sinusoidal shape. During interpolation, these deviations may cause errors. The cure is to find an encoder with high line count (2,048) and interpolation (1,024); an encoder like this can squeeze over two million signal periods into each revolution, reducing interpolation errors to less than 2%.
We must also consider the mechanical influence of the encoder itself. An encoder's mechanics may cause irregularities in system speed, imposing a limit on overall bandwidth. Command response and control reliability are also limited by the rigidity of the coupling between motor and encoder shafts as well as the natural frequency of the stator coupling. A natural frequency of 2 kHz is desired.
Kees/Allied Motion: For encoders, the maximum frequency of the data channels can have a huge impact. This defines, for a given linecount, how fast the system can run. For example, if the maximum signal frequency is 100 kHz, the maximum speed of an encoder with a resolution of 1,000 counts/rev is 100 rev/sec (100,000/1,000) or 6,000 rpm. This is not quite adequate for modern systems where speeds typically range from 0.01 to 12,000 rpm, a range reflective of smooth contouring to high speed cutting in a machine tool.
At the low end, the resolution of a 1,000-counts/rev encoder is not nearly enough; the position data (x) simply stops flowing and velocity extraction (dx/dt) comes to a halt. The system literally does not know, until the next transition, what the speed is and indeed if it goes forward or backward.
For resolvers, the yardstick is minimum speed. A resolver with 12-bit resolution and a 500 microsecond sample time has a minimum speed of about 30 rpm. Resolvers do not perform well at low speed and thus have poor dynamic range. But they are able to control high speeds well; the same resolver will work well up to 15,000 rpm for a 1:500 dynamic range.
Brian/Avtron: The output frequency of an incremental, digital encoder is the product of its speed in rpm and how many pulses it produces per revolution. This means that an encoder has two fundamental limitations: maximum shaft speed, which is normally a product of bearing and seal design, and maximum output frequency, which is controlled by the line driver output (and internal electronic) capabilities of the encoder.
One aspect of an encoder application that is often overlooked is the length of the wires from the encoder to the variable frequency drive or feedback system. The output of the encoder is a set of square waves or pulses. Pushing current through a length of wire gets harder as the wire gets longer and as frequency (speed × ppr) increases.
Consider an encoder running at 1,024 ppr at 1,800 rpm that needs 1 mA of output current to reach a drive located 10 ft away. Move the drive out 250 ft, and the current demand soars to over 40 mA. Moreover, some cable types can make it even harder to get the square wave to the drive.
Bob/Optek: In optical encoders, speed is generally a measure of the number of useable pulses per second that can be delivered. This counting speed factors heavily in determining the highest possible position resolution per revolution, multiplied by the maximum rpm that the motor is likely to achieve. For example, if it is important to have a fine resolution per revolution, say 500 pulses per revolution, and the encoder is only rated for 15 kHz, then the maximum speed of the motor must be limited to 1,800 rpm.
Howard/Renishaw:: Things attributed to the speed of an encoder include bandwidth, scale pitch, and data transmission system. The analog bandwidth, the maximum frequency content of the raw sine and cosine signals, is equal to speed/scale pitch. It is limited by the quality of the signal from the scale, optical limitations, detector performance, and the speed of the internal processing electronics.
Scale pitch for a linear system, or diameter size and scale pitch for a rotary system, are no less important. The coarser the scale pitch, the faster the top end speed. For a rotary system, speed = (maximum linear speed) (60) / π . diameter. Thus, a smaller diameter encoder system allows for greater rotational speed.
As for data transmission, with analog output systems, we are limited by the output driver and the receiving electronics. Our maximum output rate is 625 kHz at 12.5 m/sec. For digital RS-422 based systems, we are limited by the maximum data rate of the line drivers/receivers and the quality of the cabling. Nonetheless, the maximum is about 20 to 30 MHz.
Jeremy/Baumer: An encoder has a maximum mechanical speed that's a derivation of the mechanical speed limitations of each of the components contained within. In addition, an encoder has a maximum electrical processing speed or “switching frequency.” Though often identical to the mechanical speed, the switching speed can vary. Encoders with very high line counts can generate output frequencies faster than their circuitry can process. This can result in errors in the code output, and subsequent errors in the motion system.
Baud rate also limits the system due to the maximum communication speed permissible between the controlling system and the encoder. Any of these factors can affect the design of the machine due to the limitations they place on its performance.
Chuck/Danaher Industrial Controls: The speed of an absolute encoder is determined by two main elements, the integrated circuits within and the communications card. The ASIC reads the encoder disk position sensors, interpolates the higher bit positions, and then packages the position for the communications card. The communications card then packages the position into its protocol and transmits it to the motor drive or controller.
Martin/Gurley Precision: Physical components most tightly linked to encoder speed include ball bearings or friction pads, electronics, and shaft seals or sealing flaps. Encoder speed is also strongly affected by the chosen measuring step resolution. In general, lower (coarser) resolution allows higher speeds. Higher (finer) resolutions, in turn, can help control lower speeds.
What's the fastest system you've seen, and what special design steps were taken to achieve it?
Bob/Optek: Encoder designs are available that exceed 400 kHz. For a fine resolution of 500 pulses per revolution (ppr) with an encoder rated for 400 kHz, the maximum speed of the motor could be 48,000 rpm. For a typical high-speed motor turning at 10,000 rpm, this same encoder could have a resolution of 2,400 ppr.
Howard/Renishaw:: The maximum system speed which we currently offer is 12.5 m/sec (linear optical) or 4,591 rpm at 52 mm (rotary optical). Our magnetic rotary encoders can achieve 30,000 rpm.
We also are involved in very high end semiconductor motion systems with resolutions in the picometer range. Given the resolution and the “accuracy of the resolution” required, we have to use a parallel interface which converts our analog signals into parallel information, 36 bits wide. This allows the application to move at 1 m/sec with a resolution of 36 pm.
Kees/Allied Motion: Our encoders have operated as high as 16,000 rpm at 512 counts/rev. Beyond that, encoders are not really used as the commutating ‘Hall’ switches are usually more than adequate. The (far more challenging) lowest speed that our encoders have operated is one revolution in 4.5 years (0.0000004 rpm).
Brian/Avtron: We have users putting magnetic encoders on dynamometer applications up to 10,000 rpm. We selected a modular, magnetic encoder, which does not use bearings or seals; this eliminated many of the issues associated with high speed operation. With this modular, two-piece encoder, the key to success at high rpm is perfect concentricity of the rotor that mounts on the machine shaft. We created a proprietary mounting system that meets the exacting tolerances of the application.
Jeremy/Baumer: The highest speed application for an incremental encoder is most likely a printing press. As feedback for a laser copy counter, an encoder tells the counter that a certain gap has occurred between newspapers, and thus, the copy counter should count the next paper. The newspapers are printed at a rate of up to sixty thousand per hour, and the encoder will rotate at up to six thousand revolutions per minute without an error.
For an absolute system, the highest speed application is not one where the mechanical speed is of importance, but rather, where the transmission speed is critical. In such applications, designers should consider using Synchronous Serial Interface (SSI) as the communication protocol. Transmission speeds on SSI can be up to 1 MHz, providing nearly real-time communication between the controlling system and the encoders.
Martin/Gurley Precision: Some of our smaller enclosed encoders are theoretically capable of instantaneous mechanical speeds exceeding 75,000 rpm, while our standard size 25 rotary can exceed 25,000 rpm. The highest speed we've run on a size 11 (1.1-in. diam.) is about 18,000 rpm, and 15,000 rpm on a size 25 shaft encoder.
Visit motionsystemdesign.com in the coming week for more information on the factors that limit system speed.
MEET THE EXPERTS
Danaher Industrial Controls Group
Kees van der Pool
Allied Motion Technologies Inc.
Brian W. Winter
Gurley Precision Instruments
Baumer Electric Ltd.