Drives and Controls: Comparisons between the Various Technologies

The following discussion is basically a comparison between magnetic/inductive technologies and optical technologies. In many cases, the cost and performance issues make the technology selection self-evident. E.g., it would probably not be a good idea to use an Inductosyn for an application that could get by with a Hall device. In the area of servo applications for resolvers and optical encoders, the issues are more gray. Cost and performance issues overlap significantly for these devices, so this comparison focuses more on this issue than on others. Finally, since resolvers, synchros, and Inductosyns all operate in much the same way, they are generically discussed as resolvers unless specific details warrant.

.1 Power

Magnetic sensors enjoy the most favorable position in this respect. The magnetic field is free and lasts forever. As a result, these devices consume as little as 125 mW maximum. Encoders come in second place. Encoders with commutation included consume approximately 950 to 1030 mW, depending on the manufacturer and the temperature rating of the encoder. Encoders without commutation run less than this, averaging around 650 mW. A resolver with its associated converter comes in last place. A typical RDC chip alone typically consumes 300 mW, and the resolver consumes more, depending on its design, reference frequency, etc. Typically, resolvers require approximately 1000 mW (Clifton Precision).

.2 Noise Control

For all the sensors discussed, there are methods that can or should be used to improve noise immunity in any given application. Noise can be minimized by the following methods:

Increasing signal amplitude or signal to noise (S/N) ratio Using signal filtering Using hysteresis Using shielding With the exception of shielding, all of these methods have performance implications.

Signal Amplitude or Signal to Noise (S/N) Ratio. In magnetic encoders, this can be optimized through gap, magnetic material, and sensor geometry tradeoffs. How-ever, stronger magnetic materials cannot produce short pole pairs, and so resolution is lost.

In optical encoders, signal strength can be increased through using phototransistors, increasing illumination output, increasing sensor area, and reducing optical gap.

Using of phototransistors reduces frequency response and causes problems at higher temperatures. Increasing illumination by raising LED current or using an incandescent light source reduces overall life and increases power requirements.

Increasing sensor area works only if illumination is available, and reducing gap can result in mechanical stability problems or increased cost.

In resolvers, not much can be done to modify the output signal without modifying the power supply. This may be feasible in an analog system, within limits, but it's constrained by the digital converter in all other cases.

Signal Filtering. Adding filters in the signal-conditioning circuit can help to eliminate transducer errors, but it can add phase lag into the system transfer function, and can reduce the slew capability of the device. Resolvers include significant filtering by virtue of the converter electronics. For encoders with TTL outputs, digital filtering methods must be used at the receiver.

Hysteresis. Using of hysteresis in the signal conditioning circuitry can help to eliminate errors caused by low-speed operation in the region of a transition point, or when the transducer is stopped at a transition. Hysteresis is created in the input signal-conditioning circuitry (ill. 10.25) in the following way.

FIG 25 Hysteresis in the signal-conditioning circuitry.

The signal input current at V1 can be from a voltage or current source, and must result in a voltage across R3 large enough to exceed V2, ma x R2/(R1  R2). Switching won't occur again unless the voltage across R3 goes lower than  V1. This eliminates the possibility of small oscillations about a switching point causing spurious counts.

A disadvantage of hysteresis is that it induces phase shift in the output signal, and it affects absolute accuracy. All position-sensing sensor technologies make use of hysteresis to condition their switching circuitry. Optical encoders design this into the amplifier, Hall devices utilize the Schmitt trigger, and RDC circuits implement it in the up/down count logic.

Shielding. No one technology has intrinsically better noise immunity for all environments than another, although for a specific application one technology may be preferred. Magnetic encoders must be shielded from magnetic fields in order to minimize the possibility of demagnetization. Optical encoders must be shielded from EMI in order to minimize switching errors. Resolvers and Inductosyns rely heavily upon proper shielding and grounding techniques to ensure accuracy.

.3 Performance

For lower-accuracy systems, performance is based upon cost. In this area the Hall sensor is currently the low-cost leader, followed closely by simple optical products, and then MR sensors. As accuracy requirements increase, resolvers and optical encoders compete. The next performance level has optical products competing with Inductosyns. At these levels of performance, system requirements are so specialized that costs become secondary, and performance is everything. The application-specific needs will determine the appropriate technology.

Performance must also be derived through proper design, and the difficulties of successful implementation must be considered. E.g., resolvers are infinitely tunable devices. The user must carefully consider a number of parameters and select fixed components in order to successfully utilize a resolver in any given application.

The minimum resolver application must consider reference frequency, bandwidth, maximum tracking rate, number of bits in the conversion result, input filtering, ac coupling of the reference, phase compensation of the signal and reference, and off-set adjustment. All of these issues affect the overall accuracy and performance of the installed device. In contrast, the designer choosing to utilize an incremental or absolute encoder in his system is faced with only two issues, line count and the maximum rpm to frequency-response rating. Many encoder designs can now provide data at up to 1 MHz. This would allow a 4000-line encoder, equivalent to a 12-bit RDC, to turn at 15,000 rpm. There are also no encoder dynamics to deal with, and the digitization process in an encoder eliminates any analog consideration on the part of the user.

Encoders provide data with guaranteed signal separation and symmetry at up to the rated speed, and that's all there is. The added flexibility of interpolation at low speeds is similar to what is done via a multispeed resolver, but it can be implemented at a much lower cost.

Inductosyns must consider similar issues, but they achieve better overall performance than resolves, due to the greater number of poles and the good signal qualities they obtain. This is especially true for rotary Inductosyns.

.4 Accuracy

In general, optical encoders have an advantage in this area, followed closely by Inductosyns. Angle measurement systems can be as good as 0.1 8” (0.0000 5 deg ). Inductosyns can reach 0.3 2” for minimum resolution. High-resolution self-contained encoders have a minimum resolution of about 1 2” or 1/20 cycle, which for a 2000-count encoder can be as low as 3 2” . Modular incremental rotary encoders usually have an accuracy rating of 2’ or better.

Although specially fabricated resolvers can be obtained which are accurate to  , in general, resolvers have accuracy ratings of 2 to 2  .

Hall switches can achieve a resolution of about  , and are typically very difficult to align. This can be a real problem in a brushless motor commutation application.

In these applications, the switching point should be optimized in order to ensure minimum torque ripple and to maximize motor performance. For these applications, resolver and optical encoder commutation can be good to +/- 1 0’ with some careful planning, and they are much simpler to align.

.5 Resolution

The resolution, or measuring step, of a sensor is the amount of motion the device must experience before a change in output will occur. ill 10.26 shows the range of resolutions each of the major position sensor technologies can accommodate.

.6 Slew Rate

Ill. 10.26 Comparative resolution of various technologies. Position sensors generate data based on the resolution of the sensor, multiplied by the interpolation factor of the interface electronics. The maximum slew, or tracking rate, of an RDC is limited to 1/16 of the resolver reference frequency. E.g., the Analog Devices AD2S80A RDC, using a 400-Hz reference, has a tracking limit of 1500 rpm. This value can be increased to 18,750 rpm by using a 5-kHz reference.

When the tracking rate of the converter is exceeded, the position value cannot be computed fast enough to keep up with the input, and the digital output becomes totally unpredictable. A similar condition exists if the maximum acceleration rate is exceeded.

For encoders, the tracking rate is limited primarily by the frequency response of the sensor. Low- to intermediate-cost encoders generally have a maximum frequency response capability of 200 kHz, and there is no reference frequency dependency.

Newer product offerings are beginning to push the envelope up to approximately 400 kHz, which allows a 4096-cpr encoder to turn at 6000 rpm with-out missing counts. Higher performance is available at a higher cost, with a maxi-mum frequency response capability of approximately 1 MHz.

.7 Dynamic Range

Low-speed operation requires high resolution and accuracy from the sensor.

Encoder systems have dealt with this in a number of ways in the past, with new ideas and approaches being limited by available technology until recently. Many encoder-based BLDC systems have simply compromised by using a resolution based on the maximum rpm and frequency-response capabilities of the encoder, then letting the system designer deal with low-speed operation via drivetrain design. E.g., many drives use a 2000-cycle encoder with a 200-kHz frequency-response capability.

This allows the motor to run at 6000 rpm without missing counts. However, for a digital drive system with a 400- u s sampling interval, the one increment per sample speed would be 18.75 rpm. It is obvious that the system could run slower, but control accuracy degrades from this point. Improved low-speed operation is usually obtained by using interpolation electronics, which use encoder signals that are sinusoidal, as in resolvers. There are many types of interpolation electronics, and interpolation values of up to 4096 times the base resolution can be obtained, allowing encoder users to develop extremely high count-per-revolution values. Table 10.5 shows how this can be obtained. E.g., a 2048-cpr encoder paired with 51 2x interpolation electronics yields > 1 million cpr, and the minimum controllable speed now becomes 0.0006 rpm.

Resolver users have similar capabilities, but not to the same degree. Many RDC circuits allow for programmable resolutions, and tracking rates are resolution dependent. As an example, the AD2S80A has programmable resolutions of 10, 12, 14, and 16 bits. A 16-bit position resolution used with a 400-  s sample interval would allow a minimum speed control of 0.038 rpm.

From the preceding discussion, it can be seen that the two sensor types can utilize system software to realize a substantial dynamic range. For typical resolver systems, a maximum dynamic range of 18,750 to 0.038 rpm (113 dB) can be achieved. For encoder systems, using a 4096-cycle baseline device, a maximum dynamic range of 15,000 to 0.0006 rpm (144 dB) can be obtained. In order to obtain these dynamic ranges, both systems require some gymnastics on the part of the designer. The RDC system requires switching of gain resistors in order to switch resolution, and the encoder must switch to the interpolation outputs for low-speed operation. Both situations require thought on the part of the system designer to ensure that the transfer is "bumpless." As a point of reference, the typical range available from a dc tachometer is approximately 30,000 to 1 rpm  90 dB).

.8 Reliability and Durability

Reliability depends on the application environment, the overall temperature of operation, the shock, vibration, humidity, and so forth. It is common knowledge that resolvers are capable of operating at high temperatures, and in many cases can be operated at up to 15 0 C. The question is, how useful is this? In many high-performance applications, it's the motor winding that's the limiting factor. The feed-back device is usually mounted at the end of the motor, and in many applications is thermally isolated to some extent.

TABLE .5 Effective Resolution of Interpolated Outputs Interpolation factor | Line count (256 512 1024 2048 4096 5120) 256 65,536 131,072 262,144 524,288 1,048,576 1,310,720 512 131,072 262,144 524,288 1,048,576 2,097,152 2,621,440 1024 262,144 524,288 1,048,576 2,097,152 4,194,304 5,242,880 2048 524,288 1,048,576 2,097,152 4,194,304 8,388,608 10,485,760 4096 1,048,576 2,097,152 4,194,304 8,388,608 16,777,216 20,971,520

As a result, it's usually not necessary to require a feedback device to be operated above 10  C, and this temperature can be met by many modern encoders as well as resolvers. No one can discount the sturdiness of the resolver. It is a simple device with a similar makeup to that of the motor, consisting of windings, bearings, and sometimes even brushes. However, for the majority of the environments encountered, an encoder is completely adequate. Typical specifications for vibration, shock, and temperature quoted by a variety of manufacturers can be summarized as shown in Table 10.6.

TABLE 10.6 Environmental Specifications Factor Encoder Resolver Vibration 1 5g, 10 to 2000 Hz, 3 axes Same Shock 5 0g, 11 ms duration Same Temperature  10 to 10  C  55 to 12  C

Because both encoders and resolvers can be purchased with or without bearings, neither can claim an advantage in this respect. However, frameless resolvers may have slightly less sensitivity to axial play than a modular encoder would. With respect to electronics, it's true that the resolver electronics can be mounted remotely from the motor in a less extreme environment. However, they are much more complex than those of the encoder. Typical encoder designs use a small number of very basic components, and can be implemented using commercial-, industrial-, or military-rated devices. It is possible that for extremely high impact shock environments, a resolver could claim some advantage over an encoder. However, if resolutions of less than 1000 counts per revolution are desired, than an encoder with a metal code wheel can compete favorably with resolver designs.