This repository contains details of my derivative of Hagiwo's Euclidean Rhythms synthesiser module.

In his notes about the module, Hagiwo comments that there is considerable latency due to the OLED display and also mentions instability in the program possibly caused by memory issues. He partially addresses the latency issue by only updating the OLED on clock inputs, but this makes it difficult to update the settings in the absence of a clock input. During analysis of the code I found several things which would compound these issues, e.g.

• use of a relatively slow I2C OLED display. Swapped for an alternative using much faster hardware SPI
• (using a graphics library which keeps an in-memory copy of the display. Changed to the less memory-intensive U8glib library). Subsequent testing has shown that code using the Adafruit_SSD1306 library runs roughly twice as fast as either u8glib or u8g2. I'd therefore recommend using the EuclideanRhythms_AF sketch. It's larger than the u8glib based code but has shown no signs of instability despite its greater functionality
• the original code uses multiple loops over the channels, each processing a single aspect of the code. Restructured to carry out as many of these operations as possible in a single loop
• use of the Arduino digitalWrite() function (intrinsically very slow) to output the triggers. Replaced by direct port operations to update all trigger outputs much faster and simultaneously
• the rotary encoder originally used both interrupt pins of the Arduino with simple polling of the clock input, potentially leading to short clock pulses being missed. The Encoder library can also be used in a single-interrupt mode allowing the clock input to use the other interrupt, with the interrupt service routine directly generating the trigger outputs

These changes significantly improve the performance and allow the display to be updated on each clock and when the rotary encoder is used. The display update is still relatively slow even using hardware SPI resulting in long & variable length trigger outputs. Also the OLED is always on even if the module is not in use. I therefore added a screensaver which blanks the OLED after 30 seconds plus LEDs to monitor the trigger outputs. The OLED is enabled again when the rotary encoder is used. This allows quite high clock speeds once the screensaver has kicked in.

Some extra functionality has been added to the code:

• ability to 'Put' the current settings into the Arduino EEPROM and 'Get' them from EEPROM in a subsequent session. This version of the code starts up with predefined default settings, but it would be easy to modify so that the stored settings are retrieved automatically at start up
• a probability control by potentiometer or input CV to vary the likelyhood of changes in Random mode. As in Hagiwo's original code the allowed ranges of randomness are predefined in the code with the probability controls influencing how likely the changes are
• option for a gate input to re-zero all the output channels to the starting position. I didn't implement this hardware due to a lack of space on my Eurorack panel but the code for it is present

There are some more details of the changes in the Arduino code comments. The hardware is similar to the original but uses an Arduino Pro Mini rather than a Nano, has buffering on the outputs for protection and to give higher voltage triggers, and has additional circuitry for the probability controls & manual step button.

Please note that I built this module using stripboard rather than a PCB. The KiCad files and Gerbers have been used to build an SPI version of Hagiwo's 6-channel trigger sequencer so the PCB layout has now been verified apart from the probability CV input section.

08-10-2025 - updated KiCad files (amended silk-screen labelling, minor header-pin changes); added Arduino code using the faster Adafruit_SSD1306 library - RECOMMENDED