# why does absolute encoder (even multiturn) have the option of an additional incremental output

Many companies make absolute encoders for motor drives some can output an optional incremental track . my question is why is that needed since the encoder have a BiSS interface that can go as high as 10Mbit/s .

for example : https://www.kuebler.com/usa/prod-sen-multiturn.html

I have read on this site somewhere : Magnetic Encoders - Tips and Tricks

Qoute :" I wouldn't recommend closing a velocity loop on absolute position feedback but rather use quadrature signals"

If this is true then my initial guess is that the cascaded Closed loop drive use Absolute output for position loop and incremental for speed loop .

But why ? It doesn't make any sense especially that absolute encoder have much higher single turn resolution (up to 17bit and even 21bit) which can be really helpful for controlling motor at very low speeds.

I have mainly used optical encoders, but this should apply to all encoder types. The main reason you want to use incremental outputs for velocity control is latency. Your very carefully designed control loop can become unstable quite easily when there is a very small delay. Even worse, you cannot model the delay as a finite linear system in an analog control loop. I once had a D.C. Motor that became unstable when a $1\ \mathrm{\mu s}$ delay was introduced.

Latency is also important when you use controllers that are not based on PID control. In a dead beat controller, for example, the transfer function needs to be controlled quite accurately.

Also keep in mind that for encoders that have higher resolution you cannot get the exact position immediately. You can get the position in the past due to the latency. For a linear encoder with resolution $100\ \mathrm{nm}$ you get 10,000,000 ticks per meter.

• I donot understand how the 1us delay would affect the PID velocity loop , assume you are working in a microcontroller with 1ms interrupt used to calculate the PID. if quadrature encoder is used , the sensor output is ready rightaway . while if you use Absolute encoder at very high speed communication you would need about 10us to get the sensor output and perform calculations . 10us is 1% of 1ms velocity loop time and it is constant at each loop , I donot see any issues if i am getting a position which is 10us old ( past ) the whole idea of feedback is using the past to correct ,please explain – ElectronS Apr 17 '17 at 8:21
• @ElectronS delay can make your loop unstable by turning negative feedback into positive. It all depends on your plant. You have also assumed that a uC is used. FPGAs are quite widespread in these applications, especially for CNCs and CMMs for the reasons I have mentioned. It is not uncommon to see ADCs and DACs in the Msps range. – user110971 Apr 17 '17 at 12:43

An industrial servo drive would usually consists of a motion controller and driver. The motion controller calculates the position trajectory and has a position loop where the actual position is relevant. It outputs the velocity setpoint, which is the input of the drive. The drive is a speed controller where the velocity feedback is relevant.

A fully integrated servo drive would have a proprietary communication protocol between motion controller and the drive (VFD like), where the encoder signals are directly connected on drive electronics. The actual position is then forwarded to the motion controller over this proprietary communication protocol.

Assuming that BiSS protocol is something new and open source, you have the ability to connect the encoder on your motion controller/PLC and connect the other quadrature or sin/cos output directly on the VFD. You do the position loop on PLC/Motion controller and the VFD does the velocity loop.

One can certainly use high-res Abs. position encoder or high-res Quadrature encoder both which yield position and direction with simple signal processing.

In some cases the Quad detector may have logic to prevent dithering on one edge when almost still where as absolute detector requires averaging.

In the end it all comes down to specs for cost, Resolution, accuracy for use in position, velocity, acceleration and interface/processing costs, latency , noise rejection.

Without specs, I don't think you can compare apples and oranges to say which is better for you.

• actually it is easier to find absolute encoder with very high resolution (17bits to 21bits) than it is to find Quadrature , due to manufacturing constraints of the optical disc ( limited space ). for example Kuebler has 5000PPR in main stream encoders and a special type with 20000PPR max . I believe in order to get more resolution you have to use SIN/COS encoders. my question was about me trying to understand the general trend used these days in industrial drives , and what is considered a best practice , rather than comparing each type . – ElectronS Apr 17 '17 at 8:15