/* Phaser.cpp - Phasing and Approximate digital model of an analog JFET phaser. Analog modeling implemented by Ryan Billing aka Transmogrifox. ZynAddSubFX - a software synthesizer Phaser.cpp - Phaser effect Copyright (C) 2002-2005 Nasca Octavian Paul Copyright (C) 2009-2010 Ryan Billing Copyright (C) 2010-2010 Mark McCurry Author: Nasca Octavian Paul Ryan Billing Mark McCurry DSP analog modeling theory & practice largely influenced by various CCRMA publications, particularly works by Julius O. Smith. This program is free software; you can redistribute it and/or modify it under the terms of version 2 of the GNU General Public License as published by the Free Software Foundation. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License (version 2 or later) for more details. You should have received a copy of the GNU General Public License (version 2) along with this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ #include #include #include "Phaser.h" using namespace std; #define PHASER_LFO_SHAPE 2 #define ONE_ 0.99999f // To prevent LFO ever reaching 1.0f for filter stability purposes #define ZERO_ 0.00001f // Same idea as above. Phaser::Phaser(const int &insertion_, float *efxoutl_, float *efxoutr_, unsigned int srate, int bufsize) :Effect(insertion_, efxoutl_, efxoutr_, NULL, 0, srate, bufsize), lfo(srate, bufsize), old(NULL), xn1(NULL), yn1(NULL), diff(0.0f), oldgain(0.0f), fb(0.0f) { analog_setup(); setpreset(Ppreset); cleanup(); } void Phaser::analog_setup() { //model mismatch between JFET devices offset[0] = -0.2509303f; offset[1] = 0.9408924f; offset[2] = 0.998f; offset[3] = -0.3486182f; offset[4] = -0.2762545f; offset[5] = -0.5215785f; offset[6] = 0.2509303f; offset[7] = -0.9408924f; offset[8] = -0.998f; offset[9] = 0.3486182f; offset[10] = 0.2762545f; offset[11] = 0.5215785f; barber = 0; //Deactivate barber pole phasing by default mis = 1.0f; Rmin = 625.0f; // 2N5457 typical on resistance at Vgs = 0 Rmax = 22000.0f; // Resistor parallel to FET Rmx = Rmin / Rmax; Rconst = 1.0f + Rmx; // Handle parallel resistor relationship C = 0.00000005f; // 50 nF CFs = 2.0f * samplerate_f * C; invperiod = 1.0f / buffersize_f; } Phaser::~Phaser() { if(old.l) delete[] old.l; if(xn1.l) delete[] xn1.l; if(yn1.l) delete[] yn1.l; if(old.r) delete[] old.r; if(xn1.r) delete[] xn1.r; if(yn1.r) delete[] yn1.r; } /* * Effect output */ void Phaser::out(const Stereo &input) { if(Panalog) AnalogPhase(input); else normalPhase(input); } void Phaser::AnalogPhase(const Stereo &input) { Stereo gain(0.0f), lfoVal(0.0f), mod(0.0f), g(0.0f), b(0.0f), hpf( 0.0f); lfo.effectlfoout(&lfoVal.l, &lfoVal.r); mod.l = lfoVal.l * width + (depth - 0.5f); mod.r = lfoVal.r * width + (depth - 0.5f); mod.l = limit(mod.l, ZERO_, ONE_); mod.r = limit(mod.r, ZERO_, ONE_); if(Phyper) { //Triangle wave squared is approximately sin on bottom, tri on top //Result is exponential sweep more akin to filter in synth with //exponential generator circuitry. mod.l *= mod.l; mod.r *= mod.r; } //g.l,g.r is Vp - Vgs. Typical FET drain-source resistance follows constant/[1-sqrt(Vp - Vgs)] mod.l = sqrtf(1.0f - mod.l); mod.r = sqrtf(1.0f - mod.r); diff.r = (mod.r - oldgain.r) * invperiod; diff.l = (mod.l - oldgain.l) * invperiod; g = oldgain; oldgain = mod; for(int i = 0; i < buffersize; ++i) { g.l += diff.l; // Linear interpolation between LFO samples g.r += diff.r; Stereo xn(input.l[i] * pangainL, input.r[i] * pangainR); if(barber) { g.l = fmodf((g.l + 0.25f), ONE_); g.r = fmodf((g.r + 0.25f), ONE_); } xn.l = applyPhase(xn.l, g.l, fb.l, hpf.l, yn1.l, xn1.l); xn.r = applyPhase(xn.r, g.r, fb.r, hpf.r, yn1.r, xn1.r); fb.l = xn.l * feedback; fb.r = xn.r * feedback; efxoutl[i] = xn.l; efxoutr[i] = xn.r; } if(Poutsub) { invSignal(efxoutl, buffersize); invSignal(efxoutr, buffersize); } } float Phaser::applyPhase(float x, float g, float fb, float &hpf, float *yn1, float *xn1) { for(int j = 0; j < Pstages; ++j) { //Phasing routine mis = 1.0f + offsetpct * offset[j]; //This is symmetrical. //FET is not, so this deviates slightly, however sym dist. is //better sounding than a real FET. float d = (1.0f + 2.0f * (0.25f + g) * hpf * hpf * distortion) * mis; Rconst = 1.0f + mis * Rmx; // This is 1/R. R is being modulated to control filter fc. float b = (Rconst - g) / (d * Rmin); float gain = (CFs - b) / (CFs + b); yn1[j] = gain * (x + yn1[j]) - xn1[j]; //high pass filter: //Distortion depends on the high-pass part of the AP stage. hpf = yn1[j] + (1.0f - gain) * xn1[j]; xn1[j] = x; x = yn1[j]; if(j == 1) x += fb; //Insert feedback after first phase stage } return x; } void Phaser::normalPhase(const Stereo &input) { Stereo gain(0.0f), lfoVal(0.0f); lfo.effectlfoout(&lfoVal.l, &lfoVal.r); gain.l = (expf(lfoVal.l * PHASER_LFO_SHAPE) - 1) / (expf(PHASER_LFO_SHAPE) - 1.0f); gain.r = (expf(lfoVal.r * PHASER_LFO_SHAPE) - 1) / (expf(PHASER_LFO_SHAPE) - 1.0f); gain.l = 1.0f - phase * (1.0f - depth) - (1.0f - phase) * gain.l * depth; gain.r = 1.0f - phase * (1.0f - depth) - (1.0f - phase) * gain.r * depth; gain.l = limit(gain.l, ZERO_, ONE_); gain.r = limit(gain.r, ZERO_, ONE_); for(int i = 0; i < buffersize; ++i) { float x = (float)i / buffersize_f; float x1 = 1.0f - x; //TODO think about making panning an external feature Stereo xn(input.l[i] * pangainL + fb.l, input.r[i] * pangainR + fb.r); Stereo g(gain.l * x + oldgain.l * x1, gain.r * x + oldgain.r * x1); xn.l = applyPhase(xn.l, g.l, old.l); xn.r = applyPhase(xn.r, g.r, old.r); //Left/Right crossing crossover(xn.l, xn.r, lrcross); fb.l = xn.l * feedback; fb.r = xn.r * feedback; efxoutl[i] = xn.l; efxoutr[i] = xn.r; } oldgain = gain; if(Poutsub) { invSignal(efxoutl, buffersize); invSignal(efxoutr, buffersize); } } float Phaser::applyPhase(float x, float g, float *old) { for(int j = 0; j < Pstages * 2; ++j) { //Phasing routine float tmp = old[j]; old[j] = g * tmp + x; x = tmp - g * old[j]; } return x; } /* * Cleanup the effect */ void Phaser::cleanup() { fb = oldgain = Stereo(0.0f); for(int i = 0; i < Pstages * 2; ++i) { old.l[i] = 0.0f; old.r[i] = 0.0f; } for(int i = 0; i < Pstages; ++i) { xn1.l[i] = 0.0f; yn1.l[i] = 0.0f; xn1.r[i] = 0.0f; yn1.r[i] = 0.0f; } } /* * Parameter control */ void Phaser::setwidth(unsigned char Pwidth) { this->Pwidth = Pwidth; width = ((float)Pwidth / 127.0f); } void Phaser::setfb(unsigned char Pfb) { this->Pfb = Pfb; feedback = (float) (Pfb - 64) / 64.2f; } void Phaser::setvolume(unsigned char Pvolume) { this->Pvolume = Pvolume; outvolume = Pvolume / 127.0f; if(insertion == 0) volume = 1.0f; else volume = outvolume; } void Phaser::setdistortion(unsigned char Pdistortion) { this->Pdistortion = Pdistortion; distortion = (float)Pdistortion / 127.0f; } void Phaser::setoffset(unsigned char Poffset) { this->Poffset = Poffset; offsetpct = (float)Poffset / 127.0f; } void Phaser::setstages(unsigned char Pstages) { if(old.l) delete[] old.l; if(xn1.l) delete[] xn1.l; if(yn1.l) delete[] yn1.l; if(old.r) delete[] old.r; if(xn1.r) delete[] xn1.r; if(yn1.r) delete[] yn1.r; this->Pstages = min(MAX_PHASER_STAGES, (int)Pstages); old = Stereo(new float[Pstages * 2], new float[Pstages * 2]); xn1 = Stereo(new float[Pstages], new float[Pstages]); yn1 = Stereo(new float[Pstages], new float[Pstages]); cleanup(); } void Phaser::setphase(unsigned char Pphase) { this->Pphase = Pphase; phase = (Pphase / 127.0f); } void Phaser::setdepth(unsigned char Pdepth) { this->Pdepth = Pdepth; depth = (float)(Pdepth) / 127.0f; } void Phaser::setpreset(unsigned char npreset) { const int PRESET_SIZE = 15; const int NUM_PRESETS = 12; unsigned char presets[NUM_PRESETS][PRESET_SIZE] = { //Phaser //0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 {64, 64, 36, 0, 0, 64, 110, 64, 1, 0, 0, 20, 0, 0, 0 }, {64, 64, 35, 0, 0, 88, 40, 64, 3, 0, 0, 20, 0, 0, 0 }, {64, 64, 31, 0, 0, 66, 68, 107, 2, 0, 0, 20, 0, 0, 0 }, {39, 64, 22, 0, 0, 66, 67, 10, 5, 0, 1, 20, 0, 0, 0 }, {64, 64, 20, 0, 1, 110, 67, 78, 10, 0, 0, 20, 0, 0, 0 }, {64, 64, 53, 100, 0, 58, 37, 78, 3, 0, 0, 20, 0, 0, 0 }, //APhaser //0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 {64, 64, 14, 0, 1, 64, 64, 40, 4, 10, 0, 110,1, 20, 1 }, {64, 64, 14, 5, 1, 64, 70, 40, 6, 10, 0, 110,1, 20, 1 }, {64, 64, 9, 0, 0, 64, 60, 40, 8, 10, 0, 40, 0, 20, 1 }, {64, 64, 14, 10, 0, 64, 45, 80, 7, 10, 1, 110,1, 20, 1 }, {25, 64, 127, 10, 0, 64, 25, 16, 8, 100, 0, 25, 0, 20, 1 }, {64, 64, 1, 10, 1, 64, 70, 40, 12, 10, 0, 110,1, 20, 1 } }; if(npreset >= NUM_PRESETS) npreset = NUM_PRESETS - 1; for(int n = 0; n < PRESET_SIZE; ++n) changepar(n, presets[npreset][n]); Ppreset = npreset; } void Phaser::changepar(int npar, unsigned char value) { switch(npar) { case 0: setvolume(value); break; case 1: setpanning(value); break; case 2: lfo.Pfreq = value; lfo.updateparams(); break; case 3: lfo.Prandomness = value; lfo.updateparams(); break; case 4: lfo.PLFOtype = value; lfo.updateparams(); barber = (2 == value); break; case 5: lfo.Pstereo = value; lfo.updateparams(); break; case 6: setdepth(value); break; case 7: setfb(value); break; case 8: setstages(value); break; case 9: setlrcross(value); setoffset(value); break; case 10: Poutsub = min((int)value, 1); break; case 11: setphase(value); setwidth(value); break; case 12: Phyper = min((int)value, 1); break; case 13: setdistortion(value); break; case 14: Panalog = value; break; } } unsigned char Phaser::getpar(int npar) const { switch(npar) { case 0: return Pvolume; case 1: return Ppanning; case 2: return lfo.Pfreq; case 3: return lfo.Prandomness; case 4: return lfo.PLFOtype; case 5: return lfo.Pstereo; case 6: return Pdepth; case 7: return Pfb; case 8: return Pstages; case 9: return Plrcross; return Poffset; //same case 10: return Poutsub; case 11: return Pphase; return Pwidth; //same case 12: return Phyper; case 13: return Pdistortion; case 14: return Panalog; default: return 0; } }