Olig’s code reads the ports directly without using Arduino’s digitalRead. To get Oleg’s version working on the ESP32 I made these changes. However, you can stop here and use Oleg’s code if you are using a Nano or Uno, or use the version. Also, the interrupt version uses PROGMEM that also appears to be incompatible with the ESP32, or at least as it is done in the example. It uses port read to read the value of the encoder pins, which I expect is a very efficient thing to do, but didn’t work with the ESP32. Oleg’s code worked well too, but I couldn’t get it to work on the ESP32 without a couple of changes. I found some others had similar code that I believe is based on the code by Oleg. Next I came across code by Oleg Mazurov’s pages Reading rotary encoder on Arduino and Rotary encoder interrupt service routine for AVR micros It’s still a great read and I am grateful for Best Microcontroller Projects for providing the resource. However, I want to use code that I can include in projects that I can place online to share with others. Firstly, it uses polling and I wanted to use interrupts and secondly it has a copyright notice and I respect that. There were only two things missing that I wanted. I tried the Code For Improved Table Decode and was really surprised just how well it worked without any hardware filtering. I recommend reading it if you want to learn more about it.

I came across this page Rotary Encoder : How to use the Keys KY-040 Encoder on the Arduino. That is to use a table to compare the previous state with the new state and using a lookup table to ignore those changes that are not valid for a legitimate change. In some forums I read there was sometimes a comment saying there was a more reliable way. After all a reliable encoder will not necessarily stay reliable forever. I thought it was time to revisit debouncing. I don’t know if that is a possibility, but I now believe that the encoder I was using was extremely noisy. I assumed that this was because the ESP32 was so running much faster than the Nanos I’ve been using and triggering on even quicker bounces. Turning a single indent resulted in a large number of false increments. The first time I started using one with an encoder the bouncing was terrible.

I persevered with what I knew, at least until my methods didn’t work. In forums I read people recommending to use a lookup table, but not me. I thought I had it figured out by either using a couple of 100nF capacitors or by using code ignore quick changes but this was not always reliable and sometimes missed legitimate changes when rotating at speed. The issue of bounce with them is significant and for years I’ve been looking a reliable method of dealing with it. I love those simple cheap rotary encoders as used in the KY-040 modules as a method of getting user input with Arduino and ESP32 projects.

Rotary encoder arduino