Encoder vs.

Resolver-Based Servo Systems

This application note describes the basic between the primary and the quency, is induced in the primary
operation of resolver and encoder secondary windings varies de- winding, both output signals of
transducers, their related specifications, pending on the angle of the shaft. the stator will have this same
and issues to consider when applying This arrangement is called a frequency but will be out of
systems based on either technology. rotary transformer. The rotary phase by 90 degrees because of
transformer is wound with two the mechanical placement of the
Motion control requires precise secondary windings oriented 90 windings (Figure 1). One output
and repeatable indication of degrees to one another which can be called sine and the other
where the moving parts of the are stationary and mounted in a will be cosine since the cosine is
machine are located. Modern housing. These are called sta- 90 degrees out of phase with the
motion control methods provide tors. The primary winding is sine.
accurate and quick movement in located on the shaft of the The important characteristic
a machine process, allowing the resolver and is called the rotor. of the resolver is the peak voltage
manufacture of products at a When a reference AC input (amplitude) of each secondary
faster rate, for less cost, and with voltage, set at a constant fre- winding. The peak voltage will
greater flexibility than is possible Maximum influence
with older control methods. on the SINE winding
REFERENCE
These new methods require SINE SINE
devices that can accurately Minimum influence COSINE

measure and indicate the posi- on the COSINE winding

tion of all moving parts of a
machine. These devices help 0 degree
COSINE ROTOR shift
organize, synchronize, and keep
all parts of the machine in step at
a given pace.
Electronic motion controllers REFERENCE
use devices called transducers to Equal influence on SINE COSINE
both windings SINE
accurately resolve and encode
the position of a motor shaft,
other rotating shaft, or linear
45 degrees
movement on a machine. COSINE shift
R
TO

Through the history of machine

RO

development the common names

adopted for these devices are the
‘resolver’ and the ‘encoder’. Less of an influence
Knowledge about their strengths on the SINE winding COSINE

and weaknesses will help prop- SINE REFERENCE

erly apply these devices for the More of an influence

on the COSINE winding SINE
best motion control solution.
67.5 degrees
TOR

Resolvers COSINE shift

The resolver is like a small
RO

transformer with one primary and

two secondary windings. The
effective turns ratio and polarity Figure 1 The affects of shaft angle on the sine and cosine peak voltages
ORMEC Application Note -1- Encoder vs. Resolver-Based Systems
 13 12 11 0 light, the other will not sense a
14 bit digital conversion change or transition. Because of
16,384 counts per revolution
this physical arrangement, two
detectors give four transitions per
Angle within quadrant division on the disk and each
from 0 to 90 degrees
transition occurs at a unique
11 10 9 0
angular position on the shaft. By
12 bit digital conversion counting the transitions, it
4096 counts per revolution
effectively multiplies the line
count by four, hence the name
Angle within quadrant
from 0 to 90 degrees quadrature (X4) multiplication.
The relative position of the
90O two detectors is also important
for sensing the direction of
I : 00
Quadrant Select II I
rotation. While one detector is
180O 0O
II : 01 on steady and the other detector
Two most significant bits
III IV III : 10 senses a transition, it is the type
of transition that determines
IV : 11
270O direction. Light turning off to
light turning on indicates one
Figure 2 Digital format of resolver converter
direction, and light turning on to
vary as the shaft of the resolver and the remaining bits represent light turning off senses the other
rotates. This happens because the angle the shaft is at from the direction. Typically, the output of
the effective "turns-ratio" of the edge of the quadrant. On power one detector is labeled channel A
rotary transformer varies as the up, the conversion process will and the output of the other is
rotor (primary winding) is turned. track and find the current shaft labeled channel B. When an
The amplitudes are also propor- position for immediate represen- encoder specification tells you
tional to the angle of the shaft. As tation to the controller. This that channel A leads channel B
shown at the top of Figure 1, the process is explained in more in the clockwise direction, it
shaft angle is 0o . This position detail in the resolution section of means, when the encoder shaft is
puts the SINE amplitude at its this note. rotated in the clockwise direc-
maximum or 100% of the REFER- tion, detector A will turn on
ENCE voltage and the COSINE at Encoders before detector B turns on, and
its minimum or 0% of the REFER- In its basic form the incre- detector A will turn off before
ENCE voltage. In the middle mental encoder generates two detector B turns off. Transitions
view of Figure 1, the shaft angle is output signals using evenly on both channels are transmitted
45o, which puts the SINE and the etched divisions on a glass disk. back to the quadrature interface
COSINE at 70.7% of maximum. On one side of the disk is a light and counted with a digital
(sine(45)=cosine(45)=0.707) source. On the opposite side of counter circuit. The Z channel
The polarities of the voltage the disk are two light detectors. which detects the once per
sine and cosine outputs are used The disk is mounted on the shaft revolution index mark is included
to determine in which quadrant and light source and detectors as a reference point if required.
the shaft is located. As the shaft are stationary. As the disk
rotates beyond 90o the polarity of rotates, the detectors record Regarding Absolute Position
the SINE reverses. when a division interrupts the A fundamental difference
Information extracted from path of the light. The number of between resolvers and incremen-
the SINE and COSINE is trans- divisions that are etched on a tal encoders is that resolvers
lated into a digital format using disk, forms the encoder line determine absolute shaft posi-
analog to digital conversion count (Figure 3). tion directly.
techniques (Figure 2). The most The position of the two With encoders, only incre-
significant two bits represent detectors is important. As one mental shaft movement is
which quadrant the shaft is in, senses a change from dark to tracked directly, and absolute
ORMEC Application Note -2- Encoder vs. Resolver-Based Systems
 decrementing the digital counter
in an effort to keep the value in
the counter equal the angle of the
resolver shaft. This continuous
DETECTORS
A
tracking process allows the
A
B B
system to obtain the angle at
Z Z
anytime. Therefore, the amount
Started here
of usable resolution depends on
Index mark
DISK
LIGHT SOURCE
how precise the circuits, of the
on resolver-to-digital converter that
Detector A
off combine the signals to produce
on the angle, are designed and
Detector B
off manufactured.
on The resolver-to-digital con-
Detector Z
off version system, like the encoder
Time system, will limit the rotational
Figure 3 CCW rotation for 2 3/4 line counts or 11 quadrature counts
speed of the resolver shaft. The
‘tracking rate’ is finite. Generally,
shaft position is determined by a Regarding Resolution: like encoders, the greater a
third photosensor which reads a Resolution in the encoder is resolver's resolution is, the lower
“zero reference track” sensing a based on evenly spaced divisions the tracking rate will be. But
unique position once per revolu- or line counts on a glass or metal unlike encoders, the resolver's
tion of the device. disk. Using integrated semicon- method of tracking has an inher-
The result of this is that ductor techniques for etching the ent time lag of up to 15 millisec-
when power is first applied, an glass disk and aligning the onds when it encounters velocity
incremental encoder based detectors one can produce line changes during hard accelera-
system does not know absolute counts of over 10,000 with excep- tions. Performance is affected in
shaft position until the encoder is tional precision. applications such feed-and-cut
moved far enough to pass the On the other hand, resolution when cut rates reach 1000 per
first zero reference position. This of a resolver system is solely minute. The cycle time at this
poses a problem for brushless based on the techniques of the rate is 60 milliseconds and if the
servo designers, since the analog-to-digital conversion feed time is half of the cycle, the
servodrive must know absolute method used by the resolver-to- resolver's lag can be 50% of the
shaft position in order to commu- digital converter electronics. The move time. Clearly this is not an
tate the servomotor allowing it to angle of the shaft within the 90 application for a resolver-based
generate torque and move. degree quadrant has to be re- system. Hard accelerations will
This problem is commonly solved into a digital format. The not affect the encoder's ability to
overcome by adding additional popular conversion method is a give the true position of the shaft,
encoder tracks which provide 'tracking' process. First the two as long as its tracking rate is not
absolute shaft position informa- resolver signals (sine and cosine, see exceeded.
tion for the servodrive. Figure 1) are combined to recon-
Independent of whether an struct the reference signal. Then Regarding Accuracy:
encoder or a resolver is used for a digital counter holding a value What is accuracy? It is a
determining motor position, it representing the shaft angle, is combination of component errors
has become a de-facto standard modulated with the original and uncertainty.
for brushless servodrives to reference signal. The modulated Uncertainty defines how far
provide position information for signal is subtracted from the the actual position could be from
the user in the form of three reconstructed signal forming a the ideal point. Uncertainty is a
signals which correspond to difference signal. The difference function of resolution. Resolu-
encoder quadrature and a once- signal becomes the 'error' value tion defines the number of ideal
per revolution reference. that is constantly forced to a zero points a parameter can have. The
amount by either incrementing or space between the points is the
ORMEC Application Note -3- Encoder vs. Resolver-Based Systems
uncertainty. In other words, when device are determined by the This affects the resolver's ability
a system is presented with a quality of construction and the to repeatedly be at the same
value of the current point, it does ability to be stable over tempera- position for the same point over
not know how close it is to the ture ranges and vibration levels. the resolver's temperature range.
points on either side of it. Now, let's compare encoder Resolver error is also cyclic,
For an example; take one and resolver error. In a resolver meaning it will vary by the full
revolution of a shaft and divide it system, each component contrib- range (+14 to -14 arc-min for this
into 4096 parts. Since there are utes to the error from the resolver example) for one complete
60 minutes in one degree and 360 shaft back to the conversion revolution of the shaft. The cyclic
degrees in one revolution, there process. These components error will not accumulate as the
are 21,600 minutes (60 x 360) in a include construction consisten- shaft turns, but it may affect a
shaft revolution. Each division is cies in the windings, signal control system's ability to pro-
spaced by 5.273 minutes (21,600/ scaling by the converter, signal duce torque and speed control
4096). If the device were perfect, mixing technology, reference with necessary quality for some
the error would be +/- 0.00 but phasing, and tracking methods. applications.
the accuracy would be +/- 5.273 Typically signal, phasing, and Error in a encoder accounts
minutes. Even though the tracking errors are lumped into for how well the divisions on the
system is presented with the the converter specification, and disk are etched along with the
ideal value, it could physically then the resolver's error is added placement of the light detectors.
move from that point by 5.273 for the total error. Today's photographic and semi-
arc-minutes before it would get Typical error, without includ- conductor techniques, in a cost
an indication that it is on the ing uncertainty, for a general effective manner, can produce
next ideal point (see Figure 4). resolver system is 14 arc-minutes. disks and detectors with errors of
Component error defines how Resolver error will exhibit charac- approximately 0.5 arc-minutes.
close the value of a parameter is teristics that change the actual The affect due to uncertainty
to the ideal point that the param- error at its resolution points and error on accuracy, can now be
eter represents. Errors in a depending on the temperature. determined. For the resolver in

Stated position Resolver error (14.0 arc-min)

Ideal point

Uncertainty(4096)
5.273 arc-min Encoder error (0.5 arc-min)

Uncertainty(8192)
2.637 arc-min

Uncertainty(16384)
1.318 arc-min

Encoder 5.773 arc-min accuracy

4096 counts
Resolver 19.273 arc-min accuracy
Encoder 3.137 arc-min accuracy
8192 counts
Resolver 16.637 arc-min accuracy
Encoder 1.818 arc-min accuracy
16384 counts
Resolver 15.318 arc-min accuracy

Figure 4 Accuracy as a function of error and uncertainty

ORMEC Application Note -4- Encoder vs. Resolver-Based Systems
this example, accuracy would be
14 arc-minutes + 5.273 arc- Specifications: Encoders vs. Resolvers
minutes for a total for 19.273. For
Encoders Resolvers
the encoder with the same
resolution, it would be 0.5 arc-
Standard resolution 24,000 16,384
minutes + 5.273 arc-minutes for a
(counts per revolution)
total of 5.773 . When we increase
the resolution, which lowers the Maximum Speed 1,000,000 500,000
uncertainty, its affect on accuracy (counts per second)
is better for the encoder than for
the resolver. Typical/Best Accuracy 2.5 / 1.0 15.0 / 7.0
Typical accuracy for a 6000 (arc-minutes)
line count (24000 counts) device
is 1.40 arc minutes or 0.023 Typical Tracking Response Time <1 15
degrees. For a simple compari- (milliseconds)
son, let's drive a 6 in. radius roll
which has a 37.7 in. circumfer- Tolerable Shock Level 5 50
ence and look at how accurately (G's)
the surface of the roll can be
Operating Temperature Range 0 to 100o -55 to 175o
positioned. Using a 4096 line
(degrees C)
count (16384 counts) resolver
system with 15 arc-min accuracy
is (15/21600) * 37.7 inches or cable. The shield wire will drain frozen food processing and
0.026 inches. Using the 6000 line off the noise into the ground packaging, the resolver based
count encoder system with 1.40 circuit before it can affect the system could be specified.
arc-min accuracy is (1.40/21600) * transducer and produce addi-
37.7 inches or 0.0024 inches. tional error. Summary
Clearly, this example shows that When the environment is hot,
encoders can be 10 times more Environmental Issues humid, oily, or dusty beyond
accurate than resolvers with When comparing the specifi- normal industrial conditions the
approximately the same line cations of either transducer type resolver-based system is the
count. one finds that standard resolvers preferred choice. When precise
will survive in an extreme range positioning, smooth torque and
Electrical Interface of temperature, humidity environ- stable velocity control are top
Most controllers in today's ments, and will take higher shock priority for the application, the
market use the quadrature levels. This difference is evident encoder-based system is the
interface scheme. To accommo- in the specification of a resolver preferred choice. The choice is at
date these controllers, the based servomotor. The continu- best, application dependent, but
resolver based drive will convert ous duty zone is typically limited other factors can play a role. All
it's digital format (see Figure 2) by the thermal environment of factors must be balanced for the
into signal transitions emulating the motor's case. For a high right design fit.■
channel A, B and Z of the en- TENV limit, internal motor case
coder. Therefore the resolver temperatures over 140oC are
based drive system, for all intents normal. Clearly, this is an un-
and purposes, will look as an qualified place for an encoder For further reading:
encoder based drive system. unless you lower the rated 1.) ILC Data Device Corp. ,
Both transducer types use differ- continuous duty zone enough to Synchro Conversion Handbook,
ential driver and receiver tech- keep case temperatures below Third Edition.
niques in their electrical interface the safe operating specification 2.)Dynamics Research Corp.,
schemes. These differential of the encoder. In applications Techniques for digitizing rotary and
techniques are designed to reject where environments are below linear motion, Encoder Division,
electrical noise induced into the freezing point of water such as Fourth Printing.
ORMEC Application Note -5- Encoder vs. Resolver-Based Systems