Considering feedback sensing for position and speed

09 May 2017

Urs Kafadar discusses encoder properties and their selection for position and speed control applications.

The signals of a digital incremental encoder. Counting the state changes (the signal edges of channels A and B) results in a four-times higher resolution than the number of counts per turn on one encoder channel.
The signals of a digital incremental encoder. Counting the state changes (the signal edges of channels A and B) results in a four-times higher resolution than the number of counts per turn on one encoder channel.

The digital incremental encoder is the feedback sensor of choice for many applications with small motors. The main task for most applications if for position control or speed control. The level of accuracy in speed or position control can be very different so should be defined before encoder selection. Speed control at low speed (below 100 rpm) needs a better feedback than speed control at high speed (1000 rpm and above).

The load may be coupled directly onto the motor or there is a mechanical transformation system. Typically, encoders are mounted on the motor shaft, but can also be on the load itself. The mechanical properties of the transformation mechanism will influence encoder selection and gear reduction and mechanical play need to be taken into account. 

Environmental conditions such as temperature, vibration and electromagnetic interference can also have an influence on encoder selection. Optical encoders, for example, need to be protected against dust. Magnetic encoders may be sensitive to external magnetic fields – including those of the motor – and may require shielding.

The characteristic parameter of an incremental encoder is the number of rectangular pulses per motor revolution. Typically, there are two channels delivering the same pulse number. The two signals have a relative phase shift of one-quarter of a pulse length. This arrangement allows the detection of the direction of motor rotation and gives four distinctive states per pulse, or quadcounts. They represent the real resolution which is four times higher than the number of pulses on one channel. An encoder with 1000 cpt (counts or pulses per turn) gives 4,000 states per turn or a nominal resolution of 360°/4000 = 0.09°.

The encoder resolution can vary and there are many factors that influence the achievable encoder resolution – the underlying physical principle (optical, magnetic, inductive), the primary signal type (analogue or digital), the signal treatment (such as interpolation), and the mechanical layout to name a few. 

How accurate are encoders?
Resolution – the number of states – gives the nominal accuracy, the position is known within an error of one state. However, encoder pulse lengths may vary due to mechanical tolerances (for example, shaft runout, length of magnetic poles and others). The pulses in one range of motor rotation may be shorter than the pulses of other ranges. As a result the measured position deviates from the real position in a periodic way over one motor revolution. 

The maximum deviation (peak to peak) is called Integrated Non-Linearity (INL). INL is important in applications that require absolute position accuracy. Repeatability – i.e. always reaching the same position for a given set value - is not affected by INL. Repeatability is rather a question of signal jitter that typically amounts to less than one state.

Absolute positions
Incremental encoders just give position changes. For absolute positioning, a reference or home position must first be established by moving the mechanism to an external reference. 

Some encoders feature a third channel with one pulse per turn. The edges of this index channel give absolute position references within one turn. The limited accuracy of external references can be improved by an additional move to one of the index channel edges. However, the index channel is not a prerequisite for positioning. In fact, machine builders try to avoid using the Index for referencing because it requires new calibration if a motor-encoder unit has to be replaced.

Some controllers use the index channel to crosscheck the encoder signal and supervise the encoder counts per turn. Line drivers are recommended for transmission over long lines and for a better signal quality. For positioning, a line driver avoids missing encoder pulses.

Line drivers generate inverted signals for each channel. Each signal pair is transmitted together and the difference is evaluated, filtering out any electromagnetic interference during signal transmission. As a side effect, the signal quality is improved, the signal edges are more clearly defined and the driver function enables the transmission of the signal over longer distances.

Positioning systems
For positioning systems the required positioning resolution of the application will dictate the encoder resolution to be selected. A well-tuned system can maintain the position within one encoder state. Hence, the encoder resolution in quadcounts should, at least, correspond to the maximum permissible positioning error. Depending on the response time of the system, a higher encoder resolution should be chosen in order for the controller to detect deviations faster and counteract quicker. 

Signal jitter, particularly if large compared to the nominal state width of the encoder, reduces accuracy in terms of achievable repeatability. In this respect, direct sensing optical encoders have advantages over interpolated magnetic encoders. Direct sensing larger optical encoders also have advantages concerning the absolute accuracy. Their Integrated Non-Linearity (INL) is very small. 

Improving the accuracy of the reference position by an additional  move to the edge of the index channel signal.
Improving the accuracy of the reference position by an additional move to the edge of the index channel signal.

Mechanical transformation
A very high accuracy in positioning is difficult to achieve with mechanical transformation and the associated play so high resolution encoders only make sense on direct drive applications. High precision positioning often requires a high number of states and a high absolute accuracy. Optical encoders have advantages here – both due to a high resolution and a low INL. Drive systems with mechanical transformations, such as gearheads or lead and ball screws, do not require a high encoder resolution. The resolution of the encoder mounted on the motor will be multiplied by the gear reduction. Similarly, on a screw with 5mm pitch a moderate encoder resolution of 512 quadcounts will result in a theoretical position accuracy of the nut of about 10 microns. 

Absolute encoders
Incremental encoders measure only changes in position and require a homing procedure for absolute position reading. This is typically performed at low speeds, taking time that is not available in some applications. In multi-axis systems homing could cause collisions and damage. In such cases, absolute encoders can be used as an alternative to incremental models. 

In industrial applications, absolute encoders with a serial interface are often used transmitting the actual position as a bit-stream. A total of only six lines is sufficient for the supply voltage, data transmission and synchronisation of the transmission timing. 

For single-turn absolute encoders, one axis revolution is coded in N steps. The coding repeats when rotating more than 360°. Typical resolutions are 12-bit and more per revolution. In multi-turn absolute encoders, the numbers of revolutions are additionally coded and stored in the same bit stream. Multi-turn encoders are required when the number of measurement steps of a single-turn encoder is not sufficient, for instance for longer paths. 

Speed control
The highest encoder resolutions are required for very precise speed control. The encoder resolution increases with the square of the demanded speed accuracy. In addition, a fast speed control loop is needed and a high mass inertia has a beneficial effect on speed stability.
Speed is evaluated in the controller by counting the number of state changes within a given time interval. The actual speed of the motor will assume the set value, and will maintain it because of the mechanical inertia. 

Speed control at high speeds: The electronic components of the encoder limit the maximum pulse frequency that can be handled. In some cases, this restriction stems from mechanical considerations such as unbalance and mounting tolerances. 

The frequency constraints at the encoder input on the controller side should also be considered. If very high speeds are required, a correspondingly low encoder resolution should be chosen. A relative speed variation of a few percent at high speeds of several thousand rpm corresponds to several 10 rpm absolute accuracy and is quite easy to achieve. 

Speed control at low speeds: While the state counting type of speed evaluation results in a good speed control at high speeds, it becomes difficult at very low speed. 

To reduce the absolute speed variation requires higher encoder resolution and a faster controller. Just imagine an encoder with 5000 cpt in the situation described above; you get ten times more feedback. However, at low speeds the control loop should be able to react faster keeping the absolute speed deviation small. Both requirements increase the demands on the encoder. The encoder resolution increases with the square of the absolute speed stability: Half the permitted speed variation require a four times higher encoder resolution. 

At very low speeds, some controllers can offer an alternative speed evaluation solution measuring the time that elapses between two states. The speed feedback values will be more homogeneous, allowing a stiffer and more dynamic control. 

Urs Kafader is head of training at maxon motor.

Contact Details and Archive...

Print this page | E-mail this page