Belgrad

Xaoc Devices 14HP
fw 1976/1.4

Analog dual-core state variable filter. 10 morphable responses across two resonant peaks. Span detunes the cores up to 8 octaves, balance tilts peak emphasis, TITO switch enables self- or cross-modulation between cores.

Patch Ideas · 8

Vocal formant sweep
BB mode with two peaks tuned to formant ratios; a slow LFO scans FREQ for a vowel sweep, BALANCE CV opens consonant emphasis.
Walkthrough
  1. Patch your VCO (saw or pulse — harmonic-rich) → Belgrad INPUT.
  2. Set Belgrad MODE rotary to BB (the manual's vocal mode).
  3. Set LEVEL to ~6 (clean), RESO to 7 (just below self-osc), SPAN knob to 3.
  4. Set BALANCE knob to noon (symmetric peaks).
  5. Push the FM slider up to about 4 (modest depth).
  6. Patch a slow LFO (~0.2 Hz triangle, ±5V) → Belgrad FM input.
  7. Patch an envelope (~0.5 s decay) → Belgrad BALANCE input.
  8. Trigger the envelope on each note — peaks tilt for consonant attack.
  9. Patch Belgrad OUTPUT → Mixer In.
Signal out Belgrad OUTPUT — formant-filtered audio at modular level (≈±5V), level varies with MODE.
Listen for A vocal vowel sweep (oo → ah → ee) on each LFO cycle. The envelope on BALANCE adds a subtle consonant accent at note onset. Push SPAN higher for wider vowels.
Show diagram
Patch diagramPatch diagram with 5 modules and 4 connections. Modules: Belgrad, VCO, LFO, Envelope, Mixer. Signals: 2 audio, 2 cv.BelgradVCOLFOEnvelopeMixerMODE: BBRESO: 7SPAN: 3INPUTaudioFMcvBALANCEcvOUTPUTaudioOutaudioOutcvOutcvInaudio121. vowel scan2. consonant morphaudiocv
Dual sine VCO from self-oscillation
No INPUT. RESO past self-osc gives two sine VCOs; SPAN sets the interval, BALANCE crossfades between them, V/OCT tracks pitch.
Walkthrough
  1. Leave Belgrad INPUT unpatched (no audio source).
  2. Set MODE to BB and TITO switch to center (OFF).
  3. Set RESO to 8.5 — both cores self-oscillate as clean sines.
  4. Set FREQ to noon, SPAN knob to 4 (≈ a few octaves between the two sines).
  5. Set BALANCE to noon and the SPAN slider to 5.
  6. Patch your sequencer's pitch CV → Belgrad V/OCT (tunes both sines, ~5 oct tracking).
  7. Patch a slow LFO (~0.3 Hz, ±5V) → Belgrad BALANCE input.
  8. Patch a static offset (~+3V) → Belgrad SPAN input to nudge the interval over time.
  9. Patch Belgrad OUTPUT → VCA In, then to your mixer.
Signal out Belgrad OUTPUT — two clean sine VCOs summed, modular level (≈±5V).
Listen for A dual-sine drone tracking the sequence. BALANCE LFO crossfades emphasis from the lower sine to the upper, like a slow tremolo between two pitches. Adjust SPAN to set the interval (unison to ~8 oct).
Show diagram
Patch diagramPatch diagram with 5 modules and 4 connections. Modules: Belgrad, Sequencer, LFO, Static CV, VCA. Signals: 1 audio, 2 cv, 1 pitch.BelgradSequencerLFOStatic CVVCAMODE: BBRESO: 8.5 (self-osc)TITO: OFFV/OCT1v/octBALANCEcvSPANcvOUTPUTaudioPitch1v/octOutcvOutcvInaudio121. sine crossfade2. interval controlaudiocvpitch
XM chaotic lead
HH mode with TITO=XM at self-osc and moderate SPAN — the manual's recipe for chaotic cross-modulated timbres.
Walkthrough
  1. Leave Belgrad INPUT unpatched.
  2. Set MODE to HH and flip the TITO switch DOWN (XM cross-modulation).
  3. Set RESO to 8.5 (self-osc) and SPAN knob to 3 (moderate, per manual).
  4. Set BALANCE to noon, LEVEL to 0.
  5. Patch your keyboard CV → Belgrad V/OCT for pitch.
  6. Patch an envelope (~0.4 s) → Belgrad SPAN input — opens timbre per note.
  7. Patch a slow LFO → Belgrad BALANCE input for slow timbral drift.
  8. Patch Belgrad OUTPUT → VCA In, gated by the same envelope.
Signal out Belgrad OUTPUT — chaotic cross-modulated lead, modular level (loud, watch your VCA gain).
Listen for A warbly, chirpy lead full of FM-like sidebands. Each note opens with an evolving inharmonic burst as the envelope sweeps SPAN; BALANCE drift keeps the timbre moving. Tame SPAN if it gets too unpredictable.
Show diagram
Patch diagramPatch diagram with 5 modules and 4 connections. Modules: Belgrad, Keyboard, Envelope, LFO, VCA. Signals: 1 audio, 2 cv, 1 pitch.BelgradKeyboardEnvelopeLFOVCAMODE: HHTITO: XMRESO: 8.5SPAN: ~3 (moderate)V/OCT1v/octSPANcvBALANCEcvOUTPUTaudioCV1v/octOutcvOutcvInaudio11. timbre opensaudiocvpitch
Pingable highpass kick
HH mode just below self-osc — triggers into FM ping the highpass peak, V/OCT sets the pitch of the zap.
Walkthrough
  1. Leave Belgrad INPUT unpatched.
  2. Set MODE to HH, TITO to center (OFF).
  3. Set RESO to ~7.8 (just below self-osc — needs to ring, not sustain).
  4. Set FREQ to noon, SPAN knob to 0 (single peak for a tight ping).
  5. Push the FM slider up to ~7 (deep enough to excite the peak).
  6. Patch your clock or trigger source → Belgrad FM input — each pulse pings the peak.
  7. Patch a pitch CV (or static offset) → Belgrad V/OCT to set the ping pitch.
  8. Patch Belgrad OUTPUT → VCA In.
  9. Patch a short envelope (~0.2 s decay) → VCA CV to shape the kick body.
Signal out Belgrad OUTPUT — short percussive pinged sine at the FREQ pitch, modular level.
Listen for A zappy, electronic-kick-like thump with a high-end snap. Pitch tracks V/OCT; lower FREQ for a rounder kick, raise it for a chirpy zap. Nudge RESO closer to 8 for a longer ringing tail.
Show diagram
Patch diagramPatch diagram with 5 modules and 4 connections. Modules: Belgrad, Clock, Pitch CV, VCA, Envelope. Signals: 1 audio, 1 cv, 1 pitch, 1 trigger.BelgradClockPitch CVVCAEnvelopeMODE: HHTITO: OFFRESO: ~7.8FMtrigV/OCT1v/octOUTPUTaudioOuttrigOut1v/octInaudioCVcvOutcv11. pingaudiocvpitchtrigger
Short phaser (double notch)
NN mode with RESO low (deepest notches per manual); a slow LFO on SPAN moves the two notches like a phaser.
Walkthrough
  1. Patch a pad VCO or any harmonic-rich source → Belgrad INPUT.
  2. Set MODE to NN (double notch).
  3. Set RESO knob to 1 — manual: notches are deepest at LOW reso.
  4. Set FREQ to noon, SPAN knob to 4, BALANCE to noon.
  5. Push the SPAN slider up to ~6 (depth of LFO motion).
  6. Patch a slow LFO (~0.2 Hz triangle, ±5V) → Belgrad SPAN input — notch spacing breathes.
  7. Patch a small offset (~+1V) → Belgrad BALANCE input for asymmetric sweep (one notch deeper).
  8. Patch Belgrad OUTPUT → Reverb In, then to mixer.
Signal out Belgrad OUTPUT — the dry pad with two moving notches carved out, modular level.
Listen for A short phaser-like swoosh as the two notches glide apart and back together. BALANCE asymmetry makes one notch dominant for an off-center sweep. Stays subtle — push RESO past ~3 and the notches tighten and disappear.
Show diagram
Patch diagramPatch diagram with 5 modules and 4 connections. Modules: Belgrad, Pad VCO, Slow LFO, Offset, Reverb. Signals: 2 audio, 2 cv.BelgradPad VCOSlow LFOOffsetReverbMODE: NNRESO: 1INPUTaudioSPANcvBALANCEcvOUTPUTaudioOutaudioOutcvOutcvInaudio121. notch spacing2. asymmetryaudiocv
SM resonance grain on drums
LB mode with TITO=SM and overdriven LEVEL — drums get a textured, growling resonant flavour with stepped formants from S&H.
Walkthrough
  1. Patch your drum bus → Belgrad INPUT.
  2. Set MODE to LB, flip TITO switch UP (SM self-modulation).
  3. Set LEVEL to 7 (into overdrive per manual), RESO to 8.
  4. Set FREQ to noon, SPAN knob to 4, BALANCE to noon.
  5. Push the FM slider to ~5 and SPAN slider to ~6.
  6. Patch an envelope follower from the drum bus → Belgrad FM input — louder hits modulate cutoff.
  7. Patch a clocked S&H (clock from the drum sequencer, noise into S&H input) → Belgrad SPAN input.
  8. Patch Belgrad OUTPUT → Mixer (parallel-blend with the dry drums for control).
Signal out Belgrad OUTPUT — overdriven, resonance-textured drum bus at modular level (loud — attenuate at the mixer).
Listen for Drums gain a gritty, vocal-like growl. Each hit ducks and warps the resonance via SM; the S&H steps the formant peaks per beat for a stuttering, rhythmic timbral shift. Blend with the dry bus to keep transients.
Show diagram
Patch diagramPatch diagram with 5 modules and 4 connections. Modules: Belgrad, Drum Bus, Env Follower, S&H, Mixer. Signals: 2 audio, 2 cv.BelgradDrum BusEnv FollowerS&HMixerMODE: LBTITO: SMRESO: 8LEVEL: 7INPUTaudioFMcvSPANcvOUTPUTaudioOutaudioOutcvOutcvInaudio11. stepped formantsaudiocv
Stereo-like dual-peak spread
LH mode with wide SPAN — two opposing LFOs pan the lower and upper peaks across an external stereo mixer for pseudo-stereo from a mono source.
Walkthrough
  1. Patch a pad VCO → Belgrad INPUT.
  2. Set MODE to LH (lowpass + highpass band-reject).
  3. Set RESO to 6, SPAN knob to 8 (wide stopband, peaks far apart).
  4. Set BALANCE to noon, LEVEL to ~6.
  5. Patch a slow LFO A (~0.1 Hz triangle, ±5V) → Belgrad BALANCE input — tilts emphasis L↔R.
  6. Patch the inverted output of LFO A (LFO B) → your stereo mixer's pan CV input.
  7. Patch Belgrad OUTPUT → Mixer In (the channel whose pan is being modulated).
  8. Pan modulation + BALANCE tilt creates the illusion of two peaks moving across the stereo field.
Signal out Belgrad OUTPUT (mono) → modulated mixer pan = pseudo-stereo image at line/modular level.
Listen for A wide, drifting filtered pad that feels stereo despite the mono source. The lower peak appears to swing left while the upper swings right (and vice versa) as BALANCE tilts and the mixer pan sweeps in opposition.
Show diagram
Patch diagramPatch diagram with 6 modules and 4 connections. Modules: Belgrad, Pad VCO, LFO A, LFO B, Pan, Mixer. Signals: 2 audio, 2 cv.BelgradPad VCOLFO ALFO BPanMixerMODE: LHRESO: 6SPAN: 8INPUTaudioBALANCEcvOUTPUTaudioOutaudioOutcvinvcvCVcvInaudio121. peak tilt L2. peak tilt Raudiocv
Formant arp via BALANCE CV
BB mode held at one root pitch; clocked S&H steps BALANCE for a two-formant arpeggio from a static drone.
Walkthrough
  1. Patch a drone VCO (saw or square) → Belgrad INPUT.
  2. Set MODE to BB, RESO to 7, SPAN knob to 3 (vocal sweet spot).
  3. Set FREQ to a musical root note, BALANCE to noon, LEVEL to ~6.
  4. Patch a static root-pitch CV → Belgrad V/OCT (holds the root steady).
  5. Patch your clock → S&H trigger input.
  6. Patch noise → S&H signal input.
  7. Patch S&H output → Belgrad BALANCE input — each clock step jumps to a new peak emphasis.
  8. Patch Belgrad OUTPUT → Mixer In.
Signal out Belgrad OUTPUT — drone with stepped two-formant emphasis, modular level.
Listen for A static drone that arpeggiates between vowel-like timbres. Each clock pulse the random S&H value picks a new BALANCE position, jumping between lower-peak and upper-peak emphasis — like an arpeggiator but on formant emphasis instead of pitch.
Show diagram
Patch diagramPatch diagram with 7 modules and 6 connections. Signals: 2 audio, 2 cv, 1 pitch, 1 clock.BelgradDrone VCORoot CVClockS&HNoiseMixerMODE: BBRESO: 7SPAN: 3INPUTaudioV/OCT1v/octBALANCEcvOUTPUTaudioOutaudioOut1v/octOutclkTrigclkIncvOutcvOutcvInaudio11. formant stepaudiocvpitchclock

Behaviors

Mode dial morphs LP → HP MODE rotary

10 positions ordered so rotation CCW→CW gradually moves from lowpass (LL) to highpass (HH) via bandpass and notch hybrids. LL and HH turn into classic 2-pole vs 4-pole responses at SPAN extremes.

Dual-peak span detuning SPAN knob / CV

The two cores are symmetrically offset from FREQ. SPAN=0 collapses to a single peak; SPAN=max spreads up to ~8 octaves. In BB mode this produces two-formant vocal/bowed-string responses.

Asymmetric peak emphasis BALANCE knob / CV

Tilts resonance strength between the lower and upper peak. Particularly useful in NL and HN modes to expose the notch, and in NN/LH modes to morph between single and dual rejection.

Self-oscillation as sine VCO RESO ≥ ~8

Both cores produce a clean sine at resonance. V/OCT tracks about 5 octaves when tuned. SPAN sets the interval between the two sine VCOs; BALANCE crossfades between them.

Nonlinear feedback saturation RESO pushed past self-oscillation

Past the self-osc point the internal feedback paths saturate, warping the resonant sines into shaped/distorted waves. SPAN and BALANCE sculpt the resulting timbre.

TITO self-modulation (SM) TITO up · high RESO

Each core audio-rate-modulates itself. Yields a radical, grainy, textured resonance character. Most audible with RESO in self-oscillation range.

TITO cross-modulation (XM) TITO down · high RESO

The two cores modulate each other at audio rate, producing warbly/chirpy FM-like sidebands. BH and HH modes with XM and moderate SPAN give chaotic, unpredictable timbres.

Notch prominence at low RESO MODE = NN / NL / HN · low RESO

Counter-intuitively, notch modes sound deepest with RESO near zero — higher RESO narrows the notch until it is inaudible. Asymmetric BALANCE helps reveal the stopband in NL and HN.

Vocal formant sweet spot MODE = LL / LB / BB · RESO 6–7 · SPAN 2–4

The manual's recommended vocal-like setting. Two peaks land on typical formant ratios; sweeping FREQ scans vowel shapes. BALANCE adds consonant emphasis.

Mode-dependent output level changing MODE

Different topologies have radically different gain. Narrow bandpass (HL) is loud, narrow notch (NN) is quiet. Expect to rebalance the mixer after switching modes.

Input overdrive via LEVEL LEVEL past ~6

The input stage progressively saturates before entering the cores. Useful for adding grit without relying on RESO or TITO.

Controls

Global LEVEL Sets input amplitude into the filter core. Pushes into soft overdrive past unity.
0–10 · clean below ~6 · progressive input saturation above
Global FREQ Large center-frequency knob for both cores. Scaled 0–10 on the dial.
~4Hz – 28kHz · sums with V/OCT and attenuated FM
Global SPAN Symmetrically detunes the two filter cores around FREQ. Non-linear taper. Zero = cores aligned, max ≈ 8 oct split.
0–10 · non-linear · 0 = single peak · 10 ≈ 8 oct spread
Global RESO Overall resonance / Q for both cores. Self-oscillates around 8 depending on mode.
0–10 · self-osc threshold ≈ 8 · sine at resonance, saturates when pushed
Global BALANCE Tilts resonance emphasis between the two peaks. CCW = lower peak dominant · noon = symmetric · CW = upper peak dominant.
bipolar taper · CCW/CW extremes mute the opposite peak
Global FM attenuator (slider) Illuminated slider attenuating the FM input before it reaches FREQ.
0–10 · unipolar attenuator
Global SPAN CV attenuator (slider) Illuminated slider attenuating the SPAN CV input.
0–10 · unipolar attenuator
Global TITO 3-way switch routing audio-rate cross-coupling between the cores. Up = SM (self-modulation) · center = off · down = XM (cross-modulation).
SM: textured radical resonance · OFF: clean · XM: warbly chirp / FM-like
Global MODE 10-position rotary selecting the two-core topology. Progresses lowpass → highpass across the dial.
LL · LB · NL · NN · LH · BB · HL · HN · BH · HH

I/O

IN · 6

  • INPUT AUDIO
    Main audio input into the filter, scaled by LEVEL.
  • V/OCT 1V/oct CV
    1V/octave pitch input for FREQ. Used when self-oscillating as a sine VCO; tracks ~5 octaves when tuned.
  • FM audio rate capable CV
    Frequency modulation input into FREQ, scaled by the FM slider.
  • RESO 0 to +8V CV
    CV input for resonance amount. Sums with the RESO knob.
  • SPAN -5V to +5V CV
    CV input for the core detuning amount, scaled by the SPAN slider.
  • BALANCE -5V to +5V CV
    CV input for peak emphasis tilt between lower and upper resonant peaks.

OUT · 1

  • OUTPUT
    Main filter output. Level varies significantly with MODE — expect loud bandpass, quiet notch.