Encoder Feedback Devices: Basics for Motion Control Engineers

From Battlebots to the Mars Rover, large construction vehicles to semiconduc-

tor manufacturing equipment, wherever there is motion, an encoder is sure to be
found. Encoders may not be as glamorous as the products they go in, but they are
critical in today’s complex and sophisticated motion control systems. The number
of applications that require encoders are very large and can be quite different from
each other, and as a result, the types and styles of encoders available are just as
numerous. It can be quite daunting for the design engineer to choose just the right
encoder for their application. This paper looks at some of the features that can be
specified for an encoder and how that will determine where it can best be used.

Mechanical Configuration
Encoders can be split into one of two mechanical Motor feedback encoders are typically presented in
configurations: linear or rotary. As the name suggests, two further distinctions, encoders with bearings, and
the linear encoder measures linear motion or speed. modular encoders which use an existing bearing set
Typical linear applications are the control of linear like is found on the tail shaft of a servo motor. The
motors or X-Y tables such as those found on vertical determining factor of which motor feedback encoder
CNC mills. They are also commonly found in measure- to utilize is a function of the stability of the shaft/
ment instruments such as metrology instruments or bearings to which the encoder is attached. Motor
digital calipers. A rotary encoder, on the other hand, feedback encoders with bearings are typically utilized
is used to measure rotation; some typical examples when the application shaft has a significant amount
are on a motor shaft for speed control or on a PV of axial or radial run out. This style of encoder will in-
panel for solar tracking. This paper will focus on the corporate some sort of flexible member, either a flex-
more common type: the rotary encoder. ible shaft coupling or flexible body mount member,
to allow mechanical compliance with the application
Within rotary encoders, there are further basic dif-
shaft irregularities and run out.
ferentiations on mechanical configurations, which
are primarily a function of the end application of the Modular encoders rely on a mechanically stable ap-
encoder. There are two primary categories: industrial plication shaft as the shaft is responsible for holding
and motor feedback automation. Industrial encoders a rotating code wheel in a precise location relative
are used for rugged and extreme working conditions to the encoder’s sensing element, as is the case
found in heavy manufacturing conditions. Extreme with a tail shaft of a high quality servo motor. Motor
temperatures, significant vibration, dirt and debris manufacturers put a considerable amount of effort
are some of the challenges that these encoders in designing a motor with a very stable shaft/bearing
have to survive on a daily basis. As a result, industrial assembly which in turn allows them to provide a high
encoders tend to be big and solidly built to withstand performance motor. Modular encoders take advan-
the tough environment. tage of the stable motor tail shaft assembly provided
by manufacturers and are assembled from compo-
nents supplied by the encoder manufacturer to the
tail shaft and motor end bell.

The primary difference between the two motor

feedback encoders is that the modular encoder
requires the user to supply labor for assembly to the
application shaft. However, the modular design does
not add the expense of the mechanical components
US Digital EC35 Commutational Kit Encoder associated with an extra set of bearings in a motor usdigital.com
360.260.2468
 2

feedback encoder. If the user is able to apply a modu-

lar encoder they will have the most cost effective mo-
tor feedback solution. In both cases, motor feedback
encoders are generally less environmentally sealed
when compared to the industrial style encoder.

Optical and Magnetic Encoders

Linear and rotary encoders may use optical or mag-
netic to sense movement. Optical encoders are more
common and provide higher accuracy and resolu-
tion. Magnetic encoders have an advantage in that
they do not need to be in as tightly controlled of an
environment as optical encoders, and can be used
in certain environments that have higher humidity,
dust, et cetera. Magnetic encoders may also operate 5 US Digital S4 Miniature Optical Shaft Encoder (left) and
E4P OEM Miniature Optical Kit Encoder (right)
in various fluid environments.
An example of when a magnetic encoder is applied in
Optical encoders use an optical sensor to detect light
place of an optical encoder is when there is a chance
that is transmitted through or reflected from a disk
for an optical disk to become “fogged” as moisture
(also known as a code wheel) whose pattern has both
condenses on the code wheel. Consider an applica-
transparent and nontransparent lines. When the light
tion where the encoder is held at a very low tempera-
is received by the sensor, the encoder puts out a high
ture and then the ambient temperature quickly in-
signal and conversely when the light is blocked by a
creases. In most cases this quick temperature change
line on the code wheel, the sensor puts out a low sig-
causes condensation on all surfaces of the encoder,
nal. So with a known pattern on the disk, the distance
including the optical code wheel. When the code
moved or speed of movement can be measured by
wheel surface collects droplets of moisture, the light
the encoder.
transfer of the code wheel image to the optical sen-
Magnetic encoders use magnetic code wheels where sor becomes disrupted and a false or missing signal
magnetic poles are separated according to the may occur on the output. With a magnetic encoder
resolution required. A magnetic sensor detects the design, condensation of moisture is not an issue with
change in the magnetic field and produces digital the rotating magnet and magnetic sensor.
pulse trains from the encoder. Magnetic encoders 6 US Digital HD25 Industrial Rugged
use less power than their optical counterparts but Metal Optical Encoder
struggle to provide the same resolution or positional
accuracy of an optical encoder due to inherent non-
linearities in the magnetic field.

US Digital E5 Optical Kit Encoder

(cover removed) 6

5
US Digital MA3 Miniature Absolute Magnetic Shaft Encoder (foreground)
usdigital.com
US Digital MAE3 Absolute Magnetic Kit Encoder (background) 360.260.2468
 3

Absolute versus Incremental

One of the basic classifications used for encoders is
whether their architecture is absolute or incremen-
tal in design. This refers to the type of output the dress or bit of resolution over the rotation of the code
encoder emits. An incremental encoder uses the wheel. A multi-turn absolute encoder is a variant that
lines on the code wheel to output a digital pulse train also has the ability to provide not only positional
that corresponds to the sensor detecting light and information over one rotation of the encoder shaft,
dark regions. Typically incremental optical sensors but also has the ability to count turns to give absolute
use multiple sensing elements separated by various positional information over multiple turns of the
mechanical degrees and simple analog to digital encoder shaft. Once again the absolute encoder turns
electronics to ultimately produce two phase shifted are defined as a number of bits: 10 bits, 12 bits, et
output pulse trains (commonly referred to as channel cetera. The total available resolution of a multi-turn
A and channel B). The speed of rotation can be de- absolute encoder is the sum of the number of single
duced from the frequency of these pulse trains while turn bits plus the turns counting bits. For instance, a
direction of rotation is derived from the phase differ- 10 bit single turn resolution with 12 bit turns counting
ence between A and B. Many times an encoder offers resolution absolute encoder will yield 22 total bits of
a third output called the index. The code wheel has a resolution over 4096 turns of the shaft.
second optical channel with a specific optical pattern
designed to give a once per rotation pulse, index, As mentioned previously, an absolute encoder com-
which is often used as a known absolute position dur- municates a word that defines its position commonly
ing the rotation of the encoder code wheel. over a serial bus. There are a number of industrial
serial communication protocols that are used today
Incremental encoders are very useful for both posi- with absolute encoders such as Modbus, CANOpen,
tion and speed control, as the pulses are very simple and Profibus. The design engineer will have to choose
to process. The drawback to incremental encoders what is best for the application. As a result of the
is that anytime power is lost, the true position of the added complexities, absolute encoders are more
mechanical system will be lost as well. In this case the expensive than incremental, but they are also much
mechanical system will need to be reset to known better suited for critical positional control situations.
initial location and restarted. This process is com-
monly called homing the system. In some applications,
this type of recalibration process is not possible, or is Encoder Accuracy and Resolution
very expensive. An example would be semiconductor Ultimately, the encoder gives the motion controller
manufacturing equipment, where the machines have information about the velocity and/or position of a
very tight positional tolerances. After a power loss rotating shaft or linear motion. The encoder must
event, the machines require the ability to continue provide position information plus or minus a specific
from the exact at the moment power was lost, with- accuracy. Defining the accuracy of an encoder is
out performing a recalibration procedure. In such the most commonly misunderstood aspect of an
situations, absolute encoders are necessary. encoder. Often an encoder with higher resolution
is specified thinking that more lines on a disk will
An absolute encoder differs from an incremental
provide more positional accuracy. This is incorrect, as
encoder in that it outputs a position word that
accuracy and resolution of an encoder are not con-
corresponds to a specific angle of shaft rotation, as
nected in any way.
opposed to the incremental encoder series of pulses
that must be counted to define position. The absolute The accuracy of an encoder is primarily defined
position word is defined as a number of bits (10 bits, by the precision at which the code information is
12 bits, et cetera) that will determine the resolution placed on the disk and how concentric the disk pat-
granularity of the encoder over its rotation. The code tern rotates with respect to the encoder’s sensing
wheel of an optical absolute encoder consists of an element. Encoder accuracy is specified in units of arc-
usdigital.com
optical pattern that has a specific pattern for each ad- minutes or arc-seconds. For example: a low resolution
360.260.2468
 4

incremental encoder of 100 lines per revolution will may be different from the amount of rotary motion at About US Digital
just as accurately report 180 degrees as an encoder another point, simply because the radii may be differ- US Digital designs and
with 10,000 lines per revolution if they both share ent. The typical result is that on half the disk (the half manufactures motion
control components for
the same positional accuracy specification. The high where the radius is small) the increments of motion
OEM manufacturers as well
resolution encoder can just break the steps between corresponding to each count of change are too large, as end users.
0 and 180 degrees into finer increments. while on the other half of the disk (the half with the
Motion control building
longer radius) the increments of motion are too small. blocks include absolute and
Another way to understand the difference between
This gives rise to the above-mentioned sine-shaped incremental magnetic and
accuracy and resolution is to think of encoders as a optical encoders, inclinom-
error curve.
form of analog to digital converter. They convert the eters, interfaces, drives, and
more.
analog value of mechanical shaft position to a digital Fortunately, there is an equation that describes the
form, and like their electronic A to D cousins, it isn’t worst case integral non-linearity error in terms of With complete manufactur-
ing capabilities in-house,
enough to simply specify resolution. The accuracy optical radius and eccentricity: standard products are
of a traditional electronic A to D is usually specified delivered within 1-2
with two terms: integral nonlinearity, and differential TIR working days and custom
INL (worst case) = 60 sin-1 OR orders that exceed industry
nonlinearity.
expectations.

Differential nonlinearity is the amount by which adja- Where INL is maximum integral nonlinearity in arc
US Digital is located in Van-
couver, Washington, and
cent codes differ from each other and is directly tied minutes, TIR is the total indicated run out of the pat- recently celebrated their
to resolution; it has to be less than one bit, or else the tern on the code wheel, and OR is the nominal optical 30th anniversary in the mo-
resolution has been over-specified (a 12 bit encoder radius (the radial distance at which the encoder tion control industry.
with a differential non-linearity of 2 bits is really an 11 senses the code wheel pattern). Notice that this
bit encoder). With encoders, differential nonlinearity calculation is completely independent of the chosen
corresponds to the monotonicity of the encoder out- disk resolution; therefore, selecting a higher resolu-
put codes. For an incremental encoder this is almost tion disk simply does not increase accuracy. The only
never a problem: if the encoder has 1000 lines on the means to improve the accuracy error is to reduce TIR
disk you will get 1000 pulses out of the encoder. run out and/or increase the optical radius.

Integral non-linearity is a measure of how much the

actual response of the converter departs from the Conclusion
ideal response of a perfect converter; most electronic
Encoders are available in many mechanical and
A to D’s have a bow-shaped error curve relative to a
functional versions. Understanding some of the basic
line drawn from zero to maximum input value. How-
considerations of an encoder will allow a system de-
ever, integral nonlinearity is more elusive to pin down
signer to properly select an encoder that will provide
in the world of encoders. It won’t be found on most
accurate information and performance over the life of For more info please contact:
data sheets, though it still exists, usually as a sine
their application. US Digital Sales
wave shaped error curve. The major cause of integral
sales@usdigital.com
nonlinearity in encoders is the concentricity error be-
tween the code wheel and encoder sensing element.
There often is a set screw that secures the code wheel
hub to the shaft; this, combined with the clearance
necessary to slip the hub onto the shaft, gives rise
to a very slight eccentricity of motion relative to the
encoder’s disk pattern detection system. This ec-
centricity of motion means that the distance of the
sensor from the axis of rotation is not constant: it is at
a maximum at one position on the disk, and a mini-
mum 180 degrees distant. Thus the amount of rotary
motion necessary to produce a one-count change
usdigital.com
at one angular position of an eccentric code wheel
360.260.2468