DIMI-A audio input

DIMI-A has two input channels. DIMI-A inputs and outputs are hooked up differently as you would expect, so inputs go to XLR male jack while outputs use XLR female jack. Only one jack is used for both channels, and pinout is: pin 1 = input 1, pin 2 = input 2 and pin 3 is ground.

From input jack signal travels to a standard non-inverting opamp buffer circuit. There is one trimmer for each channel for adjusting input gain. DIMI-A is the first Kurenniemi instrument using an integrated opamp chip, µa709.

µa709 was introduced in 1965 by Fairchild Semiconductor and was designed by American electronics engineer Bob Widlar. It works similarly as opamps such as TL071 we still use today. It requires few external components to stabilize purposes, but you can drop in and replace it with a newer single opamp chip with eight legs.

Misusing µa709 and especially its stabilizing signal paths is the basic principle of Michel Waisvizs’ Cracklebox or Kraakdoos. On DIMI this chip is used as intended.

From the input buffer the signal travels to the comparator section. Comparator section has two parallel comparators. Both comparators will transform the signal to either digital HIGH or LOW.

First section is based on SN72710N (similar as LM710) and will output HIGH or LOW depending if the signal voltage happens to be negative or positive.

Second section is based on SN72711N (similar as LM711) and will output LOW when signal is close to zero volts, between -0.7V and +0.7V to be exact (diode voltage drop).

Digital signals from comparators continue then to the motherboard, where they are combined with a simple network of SN7404N inverter and SN7451N AND-OR-Invert (AOI) gate.

The whole process is necessary, as the signals travelling in DIMI-A motherboard are all digital, so at each moment in time, they can either be HIGH (+5V) or LOW (0V).

In DIMI-A, Kurenniemi uses a weird approach, where he adds a third possible signal level. In addition to levels HIGH and LOW, he adds a position that I will now call MIDDLE. The trick is to use ultrasonic square wave to represent a state where the signal is between these two levels. Kurenniemi uses ~3.1Mhz signal with 50% duty rate.

The idea is that after a low pass filtering at the end of the signal chain you would end up with a signal having +2.5V offset. Then potentially you would have MIDDLE level in addition to HIGH and LOW. The math kind of works in this scenario, but so far I’m not sure if after the output RC filter stage (every audio equipment has in some form or another) there is any difference from the more straightforward approach.

What all this means, is that now the AOI gate needs to transform the signals it is receiving into these three modes. End result is that when the input signal is negative, AOI output is LOW. When it is positive it is HIGH. And when it is close to zero, we will switch to an ultrasonic carrier.

I’m thinking in some scenarios the gating of the signal for instance could work better with this approach compared to using levels HIGH or LOW as mute. But switching carrier wave on every time waveform cycles from negative to positive is potentially noisy. Also as the ultrasonic carrier wave is not in sync with input signal, sonically rich and messy things will happen to the signal when it travels through DIMI-A frequency dividers and output filter sections.


  1. https://en.wikipedia.org/wiki/Bob_Widlar
  2. https://en.wikipedia.org/wiki/Kraakdoos