Considerations for Specifying, Installing and Interfacing Rotary Incremental Optical Encoders

Scott Hewitt, President
Dayton, OH

The vast majority of encoder users have a solid understanding of the type of encoder they need to accomplish their objective. What they don’t always know, at least without painful experience, are the major factors that need to be considered when specifying, installing, and interfacing with the encoders they select for their machine.

This article will address those issues, primarily as they relate to rotary incremental optical encoders, though most of the concepts that follow apply to several encoder types.

Encoder resolution is often described in terms of lines, pulses, counts. Incremental encoders provide a fixed number of square pulses on each channel per shaft rotation. Each unique number of pulses required demands a separate code disk. Each pulse has a nominal 50% duty cycle – high half the time and low half the time.

The number of lines usually refers to the number of lines on the code disk. Common line counts include binary numbers (1024 is one of the most common line count used in incremental encoders), multiples of 360, or even hundreds/thousands. For position control, the line count is normally selected to give some meaningful number of pulses per degree of rotation, or unit of linear travel (such as in ball screw or rack and pinion applications). Incremental encoders can be obtained with single channel output (A) or dual channel output (A/B), with or without complementary signals (A-not and/or B-not), and with or without an index/marker pulse (M and/or M-not).

Unless the encoder electronics multiply or divide the basic signals, the number of lines will equal the PPR. Avoid “counts per turn” or “counts per revolution”. Counts may be used to refer to pulses, but not always. If counts refer to voltage transitions, “counts” can be two times the pulses (for single channel) or four times the number of pulses (for dual channels).

Make sure you know the number of pulses per revolution, and make sure to communicate what is truly needed to the encoder supplier.

It’s not always possible to select a PPR that is already available. Several manufacturers offer selectable resolution, via DIP switches or external software programs. These encoders may be more expensive than standard encoders. However, if the PPR one needs is not available, these encoders are less expensive with faster delivery.

Older systems use encoders with open-collector outputs, and the encoder signals must be connected to a supply voltage, or to ground, inside the controls, depending on whether the controls can source or sink current. Open-collector outputs are generally limited to cable lengths of 30 feet or less.

The most common type of encoder output today is the differential line driver (DLD) that has an integrated circuit to drive various ranges of voltage and levels of current. These line drivers can drive the signals over hundreds of feet of cable. They also provide complementary signals for noise immunity, provided that cables with twisted wire pairs and shielding are used. The most common line driver outputs provided today include the 3487 (5V), the 7272 (5/30V) and the 4469 (5/15 V, with high current capability).

The 3487 chip is the most economical solution, and should be used if 5V is used all the time. The 7272 would appear to be the “universal” choice, but it comes at a cost premium to the 3487. Also, not all 7272 chips are created equal. Some have high voltage drops, and can result in less than satisfactory performance with 5V supply voltages. In any event, the 3487 and 7272 are generally limited to 40 mA or so of current. The 4469 should only be used for applications with the potential for mis-wiring, or situations that may draw high current. They are nearly indestructible, but they come at a significant cost premium. As with anything, the lowest cost solution that meets the requirement should be selected. Most encoder manufacturers state in their literature what chips they use. If not, ask them.

Cable costs can be a significant part of the expense associated with encoder installation. It’s tempting to cut costs in this area, but cable and cable/connector problems cause a significant number of field-service problems. It’s relatively easy to inspect and troubleshoot cable assembly quality issues, but there are several more subtle things that are much more difficult to detect.

Cable length, application bandwidth requirements (function of resolution and rotation speed), and driver type must be considered together when selecting a cable. Higher bandwidths mean shorter allowable cables. Differential line drivers can support longer cables than open collectors. For applications requiring differential signals (e.g. A and A-not) for noise immunity, the signals must be transmitted on a twisted pair of wires. Also, it is recommended that at least an overall shielded cable be used. Normally, shielding over each twisted pair is not necessary, except in certain applications.

In the United States it is typically recommended to connect shielding at the power supply only, to avoid ground loops. In Europe, it is normally recommended to connect shielding at both the machine and the power supply to avoid any “antenna” effect that can affect the encoder, or the power supply level. It is very seldom recommended that shielding be connected at the machine only, but there have been applications where precisely this arrangement has resolved noise issues. In any event, avoid running signal cables in the same conduits as motor power cables, or power cables for other machinery.

Encoder connectors are another consideration. Most US-standard encoders use mil-spec (MS) or similar connectors, with 6-pin, 7-pin and 10-pin being most common for incremental encoders, and 14, 17, or 19 pin connectors most common for absolute encoders. Most European encoders use M23 (23 mm) connectors, with either 12 pins (incremental and SSI absolute) or 21 pins (parallel absolute).

Such connectors require labor for assembly with cables, and each solder joint is an opportunity for a loose connection, a solder bridge, nicked wire insulation, and other problems. Contact materials, size and tolerance, and plating/flashing thickness (such as for gold contacts) can all affect performance as well. For example, in high-vibration applications, contact flashing can wear through, in some cases resulting in a very difficult to detect problem with data transmission.

Many sensors use relatively low-cost M12 (12 mm) connectors. Sensor users have eagerly noted that mating cable/connector assemblies for these sensors are very economical. Such cables are produced on automated equipment, resulting in very reliable connections, and very cost effective over-molded construction. Encoder users looking for these same cost savings on their encoders have demanded M12 connectors on their encoders, and many manufacturers have obliged them. Care should be taken to select the cable assembly that is correct for the encoder – and this may not always be the low-cost off-the-shelf solution. Most M12 cable assemblies use 5-pin connectors, which cannot be used for 6-channel, line driver output incremental encoders that need 8 pins. To be sure, M12 8-pin connectors and cable assemblies are available. However, do they use twisted pairs? Do they have the necessary shielding?

Encoder enclosure ratings generally follow applicable NEMA or IEC specifications. The most common ratings follow the IEC IP – or ingress protection – ratings. For industrial applications, encoders follow IP65, IP66 or IP67 requirements. This means that they are suitable for applications involving liquids – either dripped, splashed or sprayed. The temptation is to pick the encoder with the highest rating, as it’s difficult to know what is really required in some applications. It’s important to know what you are really getting. Keep in mind that rotary encoders have a shaft and shaft seal. Generally, they will not allow liquids to enter the encoder, unless there is a direct high-pressure spray on the shaft, or the encoder is installed “shaft up” and liquids can pool on the shaft/seal area. As a result, for the vast majority of applications and IP65 or IP66 rating is suitable. Beware the IP67 rating, as it may be based on static testing (i.e. the shaft isn’t rotating). You may be paying extra for something you don’t need, or you may not really be getting what you are expecting.

If you follow the NEMA ratings, ask the encoder manufacturer for the NEMA rating. If you follow IEC IP ratings, ask for that. There are numerous cross-reference charts floating around to convert NEMA to IP and vice versa. There is no direct correlation between the ratings, so take care when following this practice.

Bearing Loads and Ratings
Shaft Loads. The actual axial and radial loads on and encoder shaft is rarely known. Ideally, it should be avoided to mount belt shieves directly on encoder shafts, or otherwise intentionally impose a load on an encoder shaft, unless it is specifically designed for such a situation. For most applications, it is painfully easy to install the encoder such that the actual loads on the shaft are very low. Correct mounting, shaft/coupling alignment, and use of flexible couplings instead of hard connections will go a long way toward minimizing loads on encoder shafts. It also a good idea not to use the encoder as a step or lifting point for the machine!

For applications with measuring wheels or belt shieves directly mounted on the shaft, it is important to know if the encoder can tolerate the load, or if some type of mechanical isolation is needed. Encoder data sheets will usually state allowable radial and axial loads. For radial loads, you may need to ask the manufacturer where on the shaft these loads can be applied. The ratings are not always based on the end of the shaft. Sometimes it’s the middle of the shaft; sometimes it’s a specific distance from the bearing.

Bearing Ratings. Bearing ratings (usually lifetime in cycles) will also be shown on data sheets. Are these bearing lives based on the maximum allowable shaft loads? Not always. Again, ask the manufacturer if there is a doubt.

Mounting Shaft or Hollow Shaft. Rotary encoder shaft connections are available in one of two basic ways. The most common way is to couple the encoder shaft to the machine shaft with a suitable coupling (see below). The other way is to mount and encoder with a blind or through hollow shaft over the end of the machine shaft. Coupled shafts can effectively isolate the encoder from the runout, shocks and vibrations from the machine shaft. But they require a mounting bracket for the encoder, and a suitable coupling, both of which have associated costs. They also require more space be available for mounting.

Hollow shaft mounting requires no bracket or coupling – only a suitable point for the anti-rotation arm to be mounted. However, hollow shaft encoders are often more expensive that shafted encoders. Also, all machine shaft runout, shock and vibration are transmitted directly into the encoder. If the machine shaft diameter is larger than the shaft bore of the encoder, adaptor shafts can be used. However, it is important that the tolerances of the adaptor are suitable so that additional runout is not created. For applications with high loads, and for which space constraints do not allow for shaft mounting, there are hollow shaft encoders with internal flexible connections to isolate the encoder bearings and optics from the loads. However, these designs are significantly more expensive.

Coupling Selection. There are almost as many coupling manufacturers and models as there are for encoders! A suitable connector for mounting encoders should be flexible enough to allow for some minor angular and parallel mis-alignment of the encoder and machine shaft. However, they must be torsionally stiff so that there can be no relative movement between the encoder and machine shaft.

It is tempting to simply follow the coupling data sheet for allowable angular and parallel mis-alignment. Many times the coupling can tolerate much more than the encoder can. Any mis-alignment will impart loads on the encoder shaft, and these loads can be significant. Such loads can lead to premature bearing failure, while be comfortably within the load rating of the coupling itself.

Most encoder shafts are 10 mm/ 0.375 inch diameter and smaller, and many machine shafts are significantly larger than this. Generally, if possible the machine shaft diameter should be reduced to something similar to the encoder, or an adaptor shaft can be used. Avoid a large mis-match between the shaft diameters if a coupling will be used. The coupling size needs to accommodate the larger shaft, and this will result in a very stiff coupling. Almost any shaft mis-alignment will create a large load that will prematurely fail the encoder bearings.