ZoomMultistompZDL

Stereo Routing

Last updated: 2026-05-19

This file answers the recurring community question: how does a ZDL declare itself stereo so the pedal switches to stereo mode? The static answer, from the stock corpus and the ZDL header layout, is: it doesn’t. There is no stereo flag in any ZDL.

1. The header carries no stereo bit

The 76-byte ZdlInfo header was diffed across mono/stereo pairs:

pair	only field that differs
CHORUS.ZDL vs STCHO.ZDL	sort_index (browse order)
DELAY.ZDL vs STDELAY.ZDL	sort_index (browse order)

real_type, knob_type, bass_flags, sort_fx_type, unknown1, unknown2, fx_version, and version_string are identical between mono and stereo versions of the same effect family. Whatever distinguishes stereo is not in the header.

2. The ELF carries no stereo flag either

Section layout for the same pairs is essentially identical: same sections, same VAs, same alignments. The only structural deltas are size — extra DSP code and extra const data — and the exported symbol names.

Examples:

• CHORUS.ZDL exports Fx_MOD_Chorus* symbols; STCHO.ZDL exports Fx_MOD_StereoCho* with the same parameter set (depth, mix, outLv, rate, tone). The .audio section is the same size in both (0x4a0 bytes). The stereo version is a separately compiled DSP that reads/writes both channel buffers; the host UI sees the same controls.
• DELAY.ZDL exports Fx_DLY_Delay* with time/fb/mix/...; STDELAY.ZDL exports Fx_DLY_StereoDly* with timeL/timeR/timeLR/fbL/fbR/mixL/mixR. Different parameter count, different .audio size. Stock “stereo Delay” is structurally a different effect, not a flag on the mono one.

3. Pedal hardware decides routing, not the ZDL

Same-named effect ZDLs are byte-for-byte identical across pedals with different channel counts:

MS-70CDR (stereo)	MS-50G (mono)	identical?
CHORUS.ZDL	MS-50G_CHORUS.ZDL	yes (12878 B)
DELAY.ZDL	MS-50G_DELAY.ZDL	yes (16557 B)
AIR.ZDL	MS-50G_AIR.ZDL	yes (13948 B)
AUTOPAN.ZDL	MS-50G_AUTOPAN.ZDL	yes (13834 B)

The same compiled effect is shipped to both pedals. The pedal hardware / firmware is what determines whether the right-channel buffer carries real audio or is silent. Nothing in the ZDL adapts.

4. What this means for custom ZDLs

• Don’t look for a stereo flag. There isn’t one in the descriptor, the header, the SonicStomp entry, or the effectTypeImageInfo.
• If a custom effect wants to be stereo, write the audio function so it reads from and writes to both the L and R buffers. The LLLLLLLL RRRRRRRR 8-sample block layout (see build/ABI.md §5.2) is the same on mono and stereo pedals.
• On a mono pedal the right-channel pointer may carry a silent or uninitialized buffer; the audio function still gets handed a buffer pointer and must not crash. Stock stereo effects ship to the MS-50G unchanged and are known to load there, so the firmware tolerates this.
• If a custom effect wants different parameters per channel (as in STDELAY’s timeL/timeR/fbL/fbR), that’s just a parameter-count decision; it doesn’t change the ABI.

5. Open questions deliberately not closed by this doc

• Whether the pedal exposes any dynamic stereo signal (e.g. ctx[13] / ctx[14]) to custom effects: still open, tracked in docs/STATE-ABI-PROGRESS.md. Stock modulation effects use those slots; their meaning for custom DSP is unresolved.
• Whether the firmware silently passes ctx[5]_R (right wet buffer) on mono pedals or actively masks it: would need a hardware probe writing a known constant into the R wet buffer and watching for crashes on a mono pedal.
• How preset routing (the global stereo/dual chain mode) affects which pedals are “in” the L vs R path: that is a pedal-level feature, not a ZDL-level one, and is out of scope here.

6. Where this finding lives in the toolchain

Nothing to change. The build/linker.py path already has no stereo bit because stock builds don’t have one either. A “make this effect stereo” control would belong in the per-plugin manifest.json only as a hint to the author about which buffers to touch in their C code, not as a flag the host reads.

This site is open source. Improve this page.