This page is read from MIDI Instrument Definitions, a Python library used in the author's other projects, in which every number is traced to a document its maker published. Its source is on GitHub.

Novation Circuit MIDI implementation

The MIDI implementation of the Novation Circuit: what it answers to, and the 299 controls it offers, on firmware 1.8.

At a glance

Firmware

This page describes firmware 1.8.

MIDI channel

Any one of 1 to 15.

Parts

Synth, 2 of them: each on a MIDI channel of its own, taking notes, controls, and program changes, with 6 voices each.
Drums: on a MIDI channel of its own, taking notes and controls.
Session: on MIDI channel 16, taking controls and program changes.

Voices

Each part's own, under Parts.

Note range

Not recorded.

Velocity

Not recorded.

Aftertouch

Not recorded.

Pitch bend

Not recorded.

Program change

Received and sent, with 64 presets.

MIDI clock

Sent and received.

Start and stop

Sent and received.

Control change

95 controls, in the tables below.

NRPN

Supported. 204 of the controls below have one.

System exclusive

Supported.

Controls

In the order the definition gives them, by part and group. Under each control is the name a program calls it by, once pymidiinstrumentdefs.load("novation/circuit") has loaded them.

Synth

Voice

Oscillator

Mixer

Filter

Envelope

LFO

Effects and EQ

Mod matrix

Drums

Drum control

Session

Reverb

Delay

Master filter

Sidechain

Mixer

Additional controls

Where these facts come from

Source documents

• Programmer's Reference Guide, 1.3, dated 17 February 2022, its MIDI implementation, read on 4 October 2026 (the file read).
• Programmer's Reference Guide, 1.3, dated 17 February 2022, its MIDI implementation, read on 4 October 2026 (the file read).
• Circuit User Guide, 1.6, its user manual, read on 4 October 2026 (the file read).
• Circuit 1.8 User Guide, 1.8, dated 26 February 2019, its release notes, read on 4 October 2026 (the file read).

How they were read

Programmer's Reference Guide 1.3 (20pp), which carries all three control tables: Synths 1 & 2 at pp. 3-7, Drum Control at p. 10, Session Control at p. 11, with the value tables they point at at pp. 8-9 and the patch and session program changes at p. 9. The User Guide (72pp, edition 1.6) for the voices, the channels and the MIDI settings; the v1.7 addendum; and the 1.8 User Guide for the one thing that changed the channels.

THE GUIDE PRINTS SEVENTY-FOUR OF ITS 307 SYNTH ROWS MORE THAN ONCE. Pages 4, 5 and 6 repeat whole blocks - the tail of the LFO section from Lfo 1 common sync to lfo 2 fade mode, and the entire Effects and EQ section from distortion level to chorus delay. Every repeat is identical in every cell, which is asserted by the script that reads them, so this costs nothing but the counting: 307 rows, 233 distinct synth controls.

A repeat does not carry its own section, which is the trap in reducing them. A section name is printed once, on its first row, so a block repeated further down a later page inherits whatever section was open there: Lfo 2 waveform is under LFO where it is first printed and under Mod Matrix where it is repeated. And the section cannot simply be ignored either, because in the Session table the same name under two sections is two different controls - synth 1 send level is CC 88 under Reverb and CC 111 under Delay. A repeat is therefore a row whose every cell but the section matches an earlier one, which tells the two apart.

AND THE REPRINT DID NOT ADD CONTENT, IT REPLACED IT. Thirty-nine rows the instrument plainly has are missing from this guide, and what sits where they should be is the duplicated blocks:

• Macro knobs 1 and 2 are absent altogether - thirty-four rows. Knobs 4 to 8 each have seventeen: a position, and for each of four slots A to D a destination, a start position, an end position and a depth.
• Macro knob 3 begins four rows in, at macro knob 3 depth A. Its position, destination A, start position A and end position A are not printed.
• mod matrix 11 destination is missing, where mod matrix 1 to 10 each have one.

The numbering makes the gap unmistakable: the macro positions run CC 83, 84, 85, 86, 87 for knobs 4 to 8, and mod matrix 11 depth is NRPN 2:8 where 10's destination is 2:4. The missing numbers are predictable and are not recorded, because predicting a controller number is not reading one. What a player cannot reach from this document is two of the eight macro knobs, most of a third, and one modulation destination.

TWO NRPN ADDRESSES CARRY TEN PARAMETERS BETWEEN THEM, told apart by the value sent rather than by the address. NRPN 0:122 is eight switches - lfo 1 one shot at 12-13, lfo 1 key sync at 14-15, Lfo 1 common sync at 16-17 and so on to lfo 2 delay trigger at 28-29 - and NRPN 0:123 is two, Lfo 1 fade mode at 0-3 and lfo 2 fade mode at 4-7. Each is a control of its own below with its own range, which is what a sender needs: the address says nothing and the value says everything.

THE GUIDE IS OUT OF DATE ABOUT ITS OWN CHANNELS, AND IT IS THE NEWER DOCUMENT. It heads the synth table "Send messages on MIDI Channel 1 for Synth 1 and Channel 2 for Synth" (guide p. 3) and the drum table "Messages on MIDI Channel 10" (guide p. 10), with no suggestion that either can move. Firmware 1.8 changed that: "MIDI channels for each track can now be changed from the setup menu. Synth 1, Synth 2 and Drum 1/2/3/4 can now be set to MIDI channels 1-15. Channel 16 is reserved for the session." (update_1_8 p. 7). The 1.8 guide is dated 2019 and this edition of the reference guide was made in 2022, so the newer document describes the older instrument. The channels below follow 1.8.

The guide's own sentence about those channels is cut short: it reads "Channel 2 for Synth" (guide p. 3) where it means Synth 2.

NOVATION SERVES THIS GUIDE TWICE AT ONE NAME, and the second copy is not listed. The download page offers Circuit Programmer's Reference Guide 1.3_2.pdf; _1 is live at the same path and nothing points at it. They are identical in text on all twenty pages and differ only in how their images are compressed - 2.4 MB against 371 KB - and they were exported eighty-three minutes apart on the same afternoon. So trying the numbers found a second file and it says exactly the same thing, which is worth recording as the outcome it is.

AND THIS GUIDE'S CONTENTS PAGE IS RIGHT, for all twenty of its entries. Its successor's is a page out for eighteen of twenty, which cost real time when that one was read. Same maker, same template, four years apart, and this is the one that is correct.

ONE NOTE IS SET IN THE NEXT ROW'S CELL. The Filter routing note wraps over four printed lines and the fourth - the single word filter - is typeset inside the drive row's Notes cell. Read row by row, routing ends mid-sentence at "bypasses the" (guide p. 3) and drive acquires a one-word note that means nothing. They are put back together, by name rather than by position, so that a re-issue which fixes the typesetting simply stops matching.

AND ONE SESSION ROW LOOKS LIKE A TYPO THAT IS NOT CORRECTED HERE. The Sidechain section's synth 2 depth is NRPN 1:69 where its five siblings run 2:65 to 2:68 and synth 1's equivalent is 2:59. Both readings found 1:69 in the file. It is almost certainly meant to be 2:69, and the published number is what is recorded, because correcting a maker's number is not reading it.

NOT RECORDED:

• the range of notes either synth answers to, which no document states;
• what the drums do with a note outside the four in the Drum Notes Table;
• and whether the Circuit answers to aftertouch or pitch bend, neither of which is mentioned in any of these documents.