MP Electronic Devices: Yet another algorithm for rotary encoder control

This is a part of MP Electronic Devices site.

Principles of operation

Rotary encoders are devices that typically convert the angular motion of a shaft into digital output signals. There are many web resources that explain the mechanisms of these devices and how to connecting them to a microcomputer, but this page is not intended to be a complete overview. The idea is to present another algorithm to properly read the output of mechanical rotary encoders.

[Rotary encoder signal]

In principle, common rotary encoders generate the signal by alternately closing two switches, as shown in the figure above. Depending on the connection, the neutral state can be either High (1) or Low (0). We assume here the most common connection where the neutral state is High. For clockwise movement, the action on the first switch precedes the action on the second switch, while this is reversed for counterclockwise movement. If we denote each state as a binary number with two digits representing the states of the two switches, the neutral state is represented as 11, clockwise rotation is represented by the chain of states 1101001011 and counterclockwise rotation is represented by the chain of states 1110000111.

Switch bouncing

[ADC Differential Pi]

The problem is that mechanical electrical switches are not perfect, so the phenomenon switch bouncing occurs. When a switch is toggled, the contacts must physically move from one position to another. As the components of the switch settle into their new position, they bounce mechanically, causing the underlying circuit to open and close several times. The signal obtained is shown in the figure above.

This problem can be fixed either by hardware or software. In the case of software, this involves a certain subjectively determined hold time during which the activity of the switch is ignored. However, in case of the rotary switch we can avoid the hold time altogether. Since the pair of switches passes through four successive states, we can in principle define exactly when the operation is completed.

Algorithm

[ADC Differential Pi]

The microprocessor can read out a total of four different states, and two successive readouts give a total of sixteen possible transitions, as shown in the figure above. Four transitions represent no movement, four transitions represent a quarter movement to the right, four transitions represent a quarter movement to the left, and four transitions represent theoretically impossible transitions.

Let's define a full cycle as a process that begins and ends in the neutral state. By assigning 0 for no movement, +1 for a quarter movement to the right and −1 for a quarter movement to the left, we get the aggregate of +4 for a full cycle to the right, the aggregate of −4 for a full cycle to the left and the aggregate of 0 for no movement. In particular, each individual switch bouncing, whether to the left or to the right, totals 0 and has no influence to the result. Also, the modulo of the aggregate and 4 gives us the exact state. For example, if the modulo is zero, the current state is the neutral state.

Special care must be taken with "impossible" transitions. Due to technical inadequacies, they are indeed possible. If we assign the value 0 to the impossible transitions, the aggregate of a full cycle with exactly one impossible transition is −6, −2, +2 or +6. To recognise that the full cycle contained one (or more) impossible transitions, we must assign a number larger than 10 to the impossible transitions, so that the aggregate of the full cycle will be larger than 4. However, if we assign the number 14, we not only recognise the existence of the impossible transition(s), but also the modulo of the aggregate and 4 is zero again when and only when the neutral state is reached.

Implementation

All these considerations taken together can be observed in the following Python script, intended for use on Raspberry Pi. The rotary function first updates lrmem, a binary variable with four digits, where the first two digits represent the previous state and the second two digits represent the current state, which is essentially the last transformation. The list TRANS contains the values for each transformation, as shown in the following table.

lrmemTransformationTRANSComment
0b000000000no movement
0b00010001−1movement to the left
0b00100010+1movement to the right
0b00110011+14impossible movement
0b01000100+1movement to the right
0b010101010no movement
0b01100110+14impossible movement
0b01110111−1movement to the left
0b10001000−1movement to the left
0b10011001+14impossible movement
0b101010100no movement
0b10111011+1movement to the right
0b11001100+14impossible movement
0b11011101+1movement to the right
0b11101110−1movement to the left
0b111111110no movement

The rotary function then updates lrsum, the variable containing the current aggregate. When the neutral state is reached, the function acts accordingly.

Python code for Raspberry Pi

import pigpio pi = pigpio.pi() # -1: left transition, +1: right transition, 0: no transition and 14: impossible transition TRANS = [0, -1, 1, 14, 1, 0, 14, -1, -1, 14, 0, 1, 14, 1, -1, 0] LEFT = 16 RIGHT = 20 PUSH = 21 def rotary(): global lrmem global lrsum l = pi.read(LEFT) r = pi.read(RIGHT) lrmem = (lrmem % 4)*4 + 2*l + r lrsum = lrsum + TRANS[lrmem] # encoder not in the neutral state if(lrsum % 4 != 0): return(0) # encoder in the neutral state if (lrsum == 4): lrsum=0 return(1) if (lrsum == -4): lrsum=0 return(-1) # lrsum > 0 if the impossible transition lrsum=0 return(0) pi.set_mode(LEFT, pigpio.INPUT) pi.set_mode(RIGHT, pigpio.INPUT) pi.set_mode(PUSH, pigpio.INPUT) pi.set_pull_up_down(LEFT, pigpio.PUD_UP) pi.set_pull_up_down(RIGHT, pigpio.PUD_UP) pi.set_pull_up_down(PUSH, pigpio.PUD_UP) lrmem = 3 lrsum = 0 num = 0 print(num) while(True): res = rotary() if (res!=0): num=num + res print(num) if(pi.read(PUSH)==0): break

C code for Arduino

#define LEFT 2 #define RIGHT 3 #define PUSH 4 uint8_t lrmem = 3; int lrsum = 0; int num = 0; int8_t rotary() { static int8_t TRANS[] = {0,-1,1,14,1,0,14,-1,-1,14,0,1,14,1,-1,0}; int8_t l, r; l = digitalRead(LEFT); r = digitalRead(RIGHT); lrmem = ((lrmem & 0x03) << 2) + 2*l + r; lrsum = lrsum + TRANS[lrmem]; /* encoder not in the neutral state */ if(lrsum % 4 != 0) return(0); /* encoder in the neutral state */ if (lrsum == 4) { lrsum=0; return(1); } if (lrsum == -4) { lrsum=0; return(-1); } /* lrsum > 0 if the impossible transition */ lrsum=0; return(0); } void setup() { pinMode(LEFT, INPUT); pinMode(RIGHT, INPUT); pinMode(PUSH, INPUT); pinMode(LEFT, INPUT_PULLUP); pinMode(RIGHT, INPUT_PULLUP); pinMode(PUSH, INPUT_PULLUP); Serial.begin(9600); Serial.println(num, DEC); } void loop() { int8_t res; res = rotary(); if (res!=0) { num = num + res; Serial.println(num); } if (digitalRead(PUSH) == 0) { Serial.println(num); delay(250); } }

Note that no additional electronic components are necessary, as the script uses built-in Raspberry Pi pull-up resistors.

Links to other useful resources

Rotary Encoder: H/W, S/W or No Debounce?

Rotary Encoder : How to use the Keys KY-040 Encoder on the Arduino

Please use the feedback form to tell me what else you would like to see here, suggestions, new tricks, etc.

Created by Marko Pinteric: feedback form.

Updated . Web page has been read by visitors since March 2021.