Rotary Encoder problem with LCD Display

Question

Program reads rotary encoder correctly without bouncing in this code:

int val = 0;
byte clkPin = 3;
byte dtPin = 2;
bool lastClk;
bool lastDt;
void setup() {
  pinMode(clkPin,INPUT);
  pinMode(dtPin,INPUT);
  lastClk = digitalRead(3);
  lastDt = digitalRead(2);
  val = 0;
  attachInterrupt(digitalPinToInterrupt(clkPin), clkRise, RISING);
  Serial.begin(9600);
}

void loop() {
}

void clkRise(){
      lastClk = HIGH;
      lastDt = digitalRead(dtPin);
      if(lastDt == LOW){
        val--;
        Serial.println(val);
        }  else{val++; Serial.println(val);}
        }

but when i use lcd to show the result, i get massive bounces (i think) in this code:

#include <LiquidCrystal_I2C.h>

#include <Wire.h>

LiquidCrystal_I2C lcd(0x27,20,4);
int val = 0;
byte clkPin = 3;
byte dtPin = 2;
bool lastClk;
bool lastDt;
void setup() {
  lcd.begin();
  pinMode(clkPin,INPUT);
  pinMode(dtPin,INPUT);
  lastClk = digitalRead(3);
  lastDt = digitalRead(2);
  val = 0;
  attachInterrupt(digitalPinToInterrupt(clkPin), clkRise, RISING);
  Serial.begin(9600);
  lcd.home();
}

void loop() {
  lcd.print(val);
}

void clkRise(){
      lastClk = HIGH;
      lastDt = digitalRead(dtPin);
      if(lastDt == LOW){
        val--;
        Serial.println(val);
        }  else{val++;}
        }

it increases and decreases by random numbers between 0-5.I use KY-040 R.E. it has built-in 10k pull up resistors.

This is what i use to debounce:

A or B pin of R.E--10k R----------------------------------Arduino INPUT PIN
                         |
                         |
                         =0.1uF C
                         |
                         ---------------------GND

Answer 1

I suspect the problem is that your hardware-debounce capacitive circuit isn't working well enough to accurately debounce the contacts, and that invalid transitions are occurring on the quadrature lines as read by the program.

Note that use of Serial.println() within your clkRise() interrupt handler is problematic, because Serial.println() uses interrupts itself. As far as I know, println() is not reentrant, ie cannot be used correctly from the multiple interrupts that may start before previous println() calls finish their background processing. The bits of extra delay introduced by LCD code may merely be exposing a problem.

There is no need for either hardware or software debouncing on quadrature inputs. A correctly-implemented state machine will reject invalid conditions that arise during bounce, or will immediately follow any extra up-tick by a corresponding down-tick, or vice versa.

For some state machine code, see for example answer 16420/3917 and also see answer 18221/3917. Also see answer 19109/3917.