Why does Program Change not switch patterns on my Digitakt II (Song Mode)?
What we saw
- PC 0 was transmitted correctly but the device stayed on the song-controlled pattern.
- After disabling Song Mode via 0x54, the same PC 0 immediately produced A01.
- The device did not send a Program Change by itself when it changed pattern, so a listener cannot learn pattern changes from outgoing PCs.
Channels in our setup
In our project the auto channel was MIDI channel 16, so a pattern change was the two bytes CF nn. Tracks 1–14 listened on channels 1–14 and channel 15 carried global FX CCs. These are project/device settings, check your own MIDI config.
Timing
A queued pattern change takes effect at the next pattern-change boundary. Our tooling refused PC requests with less than 54 MIDI clock ticks (24 ppqn) of lead before that boundary; 29 and 51 ticks were rejected, later requests were accepted. This is an empirical safety margin from our setup, not a device specification.
With per-track scale, the change grid is the pattern's Change length, not the track length: in one project every change landed on a 128-tick grid (Change 32 at 3/2×).
FAQ
Which MIDI channel does Digitakt II use for Program Change?
The auto channel set in the MIDI config; in our setup channel 16 (status byte 0xCF).
Does Song Mode block Program Change?
In our tests yes: disable Song Mode first.
Observed on / Status: Observed on Digitakt II OS 1.16/1.17, October 2026. Channel layout is specific to our project.
Deutsch
Im Song-Modus überstimmt der Song-Sequencer jeden Program Change. Erst Song-Modus über 0x54 Offset 19 = 0 aus, dann PC senden. Bei uns lief PC auf dem Auto-Kanal 16; mindestens 54 Clock-Ticks Vorlauf vor der nächsten Wechselgrenze.
Related
- Where are pattern and Song Mode state in the Digitakt II project settings (0x54)?
- How can software read the live state of a Digitakt II (telemetry)?
- How can AI agents jam on a hardware sampler without wrecking the set (clipping, level locks, filter checks)?