EQ in a layer: an independent low shelf first, then a fixed bank #122

New issue

Open

opened 2026-10-05 23:35:05 +00:00 by bcox · 0 comments

bcox commented

2026-10-05 23:35:05 +00:00

Owner

bcox asked on 2026-10-05 for a ticket for each engine gap that the third set of instruments in his moregaps.md left without one. Engine claims are from reading the tree at f6aa8eba on main. Instrument claims are from that day's research, read mostly as archive.org OCR of the manuals by research agents and not re-checked; the sources are listed in that document. Nothing here is built, measured or listened to.

Summary

A layer's only EQ is the strip's tilt: a coupled pair of shelves, 2^tilt on the high and 2^-tilt on the low, so full travel is exactly 6 dB, with a pivot, and not a modulation destination (dsp/layer_strip.h:363-381). The FM pair has three formant bands (dsp/fm_pair_layer.h:352-372). The console strip's EQ is four bands (fabric/inserts.h:82-91) and project-level, not a patch's. #14 (closed) asked for a filter or an EQ at the end of the FM pair's chain.

moregaps.md's recommendation since its first set: a low shelf in the layer is the engine change with the widest payoff. 80 of the Alpha Juno's 128 factory tones sit on its bass-boost setting; 70 of the Juno-106's 128.

What the instruments do

A low shelf or boost: Alpha Juno and Juno-106 (the highpass control's bass-boost position), Poly-800 II (bass and treble), Trident strings (two bands), Delta strings (bass and treble), Pro 2 (Girth and Air, a low and a high shelf).
A fixed bank: K5's DFT, an 11-band graphic equalizer labeled C-1 to C9 over both sources; Teisco S-110F, eight fixed bands from 250 Hz to 2.8 kHz with up to 25 dB of boost; Polymoog's resonators, three bands each with frequency, emphasis and gain, and one lowpass, band-pass or highpass mode for all three.

A fact that shapes it

A static linear EQ after the sum equals the same EQ on every voice. So a channel-scope instance covers every case above (the Polymoog's resonators and the K5's DFT are shared in the hardware too), and per-voice cost is never needed for an EQ that nothing modulates per note.

The work, as a sketch

Two sizes, in order:

A decoupled low shelf on the strip. The smallest change: a second gain beside the tilt, on the low shelf alone, in the same SvfState width. It covers both Junos, the Poly-800 II, the Trident and the Delta, and most of what the file set tallied.
An EQ layer kind, channel scope by default: up to some number of bands, each a two-pole SVF whose taps give peak, low-shelf and high-shelf responses (dsp/svf.h computes all three taps in one recursion), with frequency and gain as lanes. A fixed bank is peaks at fixed frequencies; the Polymoog's resonators are three bands with a shared mode. Same slot machinery as the delay, reverb and phaser layers.

Open

Whether 1 is enough for a long time, and 2 waits for a patch that needs the K5's eleven bands.
Bands as destinations: the K5's DFT has no modulation; the Polymoog's resonators are swept by hand, which is a macro.

bcox asked on 2026-10-05 for a ticket for each engine gap that the third set of instruments in his `moregaps.md` left without one. Engine claims are from reading the tree at `f6aa8eba` on `main`. Instrument claims are from that day's research, read mostly as archive.org OCR of the manuals by research agents and not re-checked; the sources are listed in that document. **Nothing here is built, measured or listened to.** ## Summary A layer's only EQ is the strip's tilt: a coupled pair of shelves, 2^tilt on the high and 2^-tilt on the low, so full travel is exactly 6 dB, with a pivot, and not a modulation destination (`dsp/layer_strip.h:363-381`). The FM pair has three formant bands (`dsp/fm_pair_layer.h:352-372`). The console strip's EQ is four bands (`fabric/inserts.h:82-91`) and project-level, not a patch's. #14 (closed) asked for a filter or an EQ at the end of the FM pair's chain. `moregaps.md`'s recommendation since its first set: a low shelf in the layer is the engine change with the widest payoff. 80 of the Alpha Juno's 128 factory tones sit on its bass-boost setting; 70 of the Juno-106's 128. ## What the instruments do - A low shelf or boost: Alpha Juno and Juno-106 (the highpass control's bass-boost position), Poly-800 II (bass and treble), Trident strings (two bands), Delta strings (bass and treble), Pro 2 (Girth and Air, a low and a high shelf). - A fixed bank: K5's DFT, an 11-band graphic equalizer labeled C-1 to C9 over both sources; Teisco S-110F, eight fixed bands from 250 Hz to 2.8 kHz with up to 25 dB of boost; Polymoog's resonators, three bands each with frequency, emphasis and gain, and one lowpass, band-pass or highpass mode for all three. ## A fact that shapes it A static linear EQ after the sum equals the same EQ on every voice. So a channel-scope instance covers every case above (the Polymoog's resonators and the K5's DFT are shared in the hardware too), and per-voice cost is never needed for an EQ that nothing modulates per note. ## The work, as a sketch Two sizes, in order: 1. **A decoupled low shelf on the strip.** The smallest change: a second gain beside the tilt, on the low shelf alone, in the same `SvfState` width. It covers both Junos, the Poly-800 II, the Trident and the Delta, and most of what the file set tallied. 2. **An EQ layer kind**, channel scope by default: up to some number of bands, each a two-pole SVF whose taps give peak, low-shelf and high-shelf responses (`dsp/svf.h` computes all three taps in one recursion), with frequency and gain as lanes. A fixed bank is peaks at fixed frequencies; the Polymoog's resonators are three bands with a shared mode. Same slot machinery as the delay, reverb and phaser layers. ## Open - Whether 1 is enough for a long time, and 2 waits for a patch that needs the K5's eleven bands. - Bands as destinations: the K5's DFT has no modulation; the Polymoog's resonators are swept by hand, which is a macro.