* =============================================================================
* LM3886 Non-Inverting Power Amplifier — ngspice / LTspice-compatible netlist
* =============================================================================
* Topology: Class AB, gain = 21 (≈26 dB), 8 Ω midrange driver load
*
* Official TI LM3886 SPICE model (preferred for accuracy):
*   Download from TI: https://www.ti.com/product/LM3886
*   Product folder → Design & development → Simulation models
*   Or search: "LM3886 PSpice Model" on ti.com
* Place LM3886.LIB / LM3886.MOD next to this file and uncomment the .include
* line below. Until then, a behavioral macromodel approximating the IC is used.
*
* VERIFY AGAINST DATASHEET before ordering PCBs:
*   https://www.ti.com/lit/ds/symlink/lm3886.pdf
*   Mute pin threshold, SOA, thermal pad, and pinout must match your package.
* =============================================================================

* .include LM3886.LIB          ; <-- uncomment when TI model is present

.param Rin=1k Cin=1u Rf=20k Cf=20p Ri=1k Ci=250u
.param Vsupply=25
.param Rload=8

* ----- Signal source (AC + transient sine) -----------------------------------
* Vac for .ac sweep; Vsin for .tran. Switch via XACTRAN param if needed.
Vin IN 0 DC 0 AC 1 SIN(0 0.707 1k)   ; 0.707 Vpk ≈ 0.5 Vrms → ~10.5 Vpk out

* ----- Input coupling & bias -------------------------------------------------
* Cin: AC-couples source, blocks DC offset into the non-inverting input
Cin IN N_IN  {Cin}                 ; 1 µF film — low distortion coupling
* Rin: sets HF corner with Cin (fc ≈ 1/(2π·Rin·Cin) ≈ 159 Hz with 1k… but
*      wait: Rin is series; bias Rbias sets DC path. Series Rin = 1k as specified.
Rin N_IN N_PLUS {Rin}              ; 1 kΩ series input resistance
* Input bias resistor: DC path for non-inverting input bias current → GND
Rbias N_PLUS 0 1Meg                ; 1 MΩ — keep DC offset low; VERIFY datasheet Ib

* ----- Feedback network (non-inverting gain = 1 + Rf/Ri = 21) -----------------
* Rf / Ri set closed-loop gain; Cf rolls off ultrasonic gain for stability
Rf OUT N_MINUS {Rf}                ; 20 kΩ feedback resistor
Cf OUT N_MINUS {Cf}                ; 20 pF — compensates stray C, aids stability
* Ri AC-coupled so DC gain = 1 (minimizes output DC offset from Vos)
Ri N_MINUS N_GAIN {Ri}             ; 1 kΩ gain-setting resistor
Ci N_GAIN 0 {Ci}                   ; 250 µF electrolytic — AC ground for Ri
*                               WHY Ci: at audio freqs ≈ short; at DC open → gain=1

* ----- Behavioral LM3886 macromodel (replace with TI .subckt when available) --
* Ideal VCVS with supply clamps, finite GBW roll-off, and mute switch.
* This is NOT a substitute for the official model for THD/SOA studies —
* use TI's model before trusting distortion numbers.
*
* Pin map (TO-220 style LM3886T, VERIFY datasheet for your exact package):
*   +IN, −IN, OUT, V+, V−, MUTE
*
* Open-loop gain ~100 dB, dominant pole ~10 Hz → GBW ≈ 8 MHz class behavior
Eamp N_ERR 0 N_PLUS N_MINUS 1e5
Rpole N_ERR N_AMP 1k
Cpole N_AMP 0 15.9u                ; fp ≈ 10 Hz with 1k → Aol roll-off
* Supply-limited output buffer (simple clamp via diodes + series Rout)
Ebuf N_RAW 0 N_AMP 0 1
Rout N_RAW N_PREOUT 0.05           ; output stage bulk resistance
DclipP N_PREOUT VCC DCLAMP
DclipN VEE N_PREOUT DCLAMP
.model DCLAMP D(Is=1e-12 N=0.01 Rs=0.01)

* Mute: when MUTE node is low, output is forced near 0 via series switch
* RC delay from VEE (typical LM3886 mute referenced to V−) — VERIFY datasheet
* Here: RC from +rail to mute pin through 100k + 10µF to GND ≈ 1 s delay
Rmute VCC N_MUTE 100k              ; 100 kΩ — charges Cmute after power-up
Cmute N_MUTE 0 10u                 ; 10 µF — τ ≈ R·C = 1 s turn-on delay
* Soft mute: series resistance decreases as mute voltage rises
* Approximation: voltage-controlled resistor via switch (ngspice)
Sout N_PREOUT OUT N_MUTE 0 SWMUTE
.model SWMUTE SW(Vt=0.5 Vh=0.1 Ron=0.01 Roff=10Meg)

* ----- Zobel (Boucherot) network — resistive HF load for inductive speakers ---
* WHY: speaker impedance rises with frequency; Zobel keeps amp seeing ≈Rload
Rz OUT N_ZOBEL 10                  ; 10 Ω
Cz N_ZOBEL 0 0.1u                  ; 0.1 µF film — with 10Ω ≈ 8Ω @ HF

* ----- Output inductor (Thiele network) — isolates capacitive cable loads -----
* WHY: prevents oscillation with long speaker cables / capacitive loads
Lout OUT N_SPK 0.7u                ; 0.7 µH air-core or ferrite-core inductor
RLpar OUT N_SPK 10                 ; 10 Ω parallel — damps inductor Q

* ----- Load (8 Ω midrange driver) --------------------------------------------
Rload N_SPK 0 {Rload}

* ----- Supply rails ±25 V with decoupling ------------------------------------
VCC VCC 0 {Vsupply}
VEE VEE 0 {-Vsupply}
* Ceramic HF bypass — place within mm of IC pins on real PCB
CdecP_cer VCC 0 100n
CdecN_cer VEE 0 100n
* Bulk electrolytic — reservoir for bass current peaks
CdecP_el VCC 0 1000u
CdecN_el VEE 0 1000u

* ----- Analyses --------------------------------------------------------------
* AC: frequency response 10 Hz – 100 kHz (magnitude & phase at speaker node)
.ac dec 100 10 100k
* Transient: 1 kHz sine, enough cycles to settle mute (~1 s) then measure
* For fast THD check without waiting for mute, start mute charged:
*   .ic V(N_MUTE)=5
.ic V(N_MUTE)=5
.tran 1u 10m 0 1u

* Fourier / THD estimate on speaker voltage after settle
.four 1k V(N_SPK)

* Print helpers
.print ac Mag(V(N_SPK)) Phase(V(N_SPK))
.print tran V(N_SPK) V(IN) V(N_MUTE)

.end