Sphinx 101

Congratulations!

You just joined a very small group of engineers and producers who own one of the most comprehensively modeled analog master bus processors out there. Over 2,000 automated DSP measurements across eleven independent test suites verify every analog mechanism on every release. Sphinx 101 models the individual electronic components inside the hardware, wires them together in the real circuit topology, and lets the sound emerge from how they interact. Welcome our TrueRail technology.

What's in the Box

Sphinx 101 audio plugin (macOS AU, Windows VST3), two visual skins (Amok, Nevy), default factory presets, this manual, signal test results, and your serial key.

Installation and Authorization

macOS: Double-click the installer .pkg.

The plugin installs to /Library/Audio/Plug-Ins/Components/Sphinx 101.component (AU). The manual, signal test results, and an uninstaller .command are placed at /Applications/Analog Realism/Sphinx 101/ — visible in Finder, parallel to the Windows layout below. Your per-machine activation state (calibration.dat) lives in ~/Library/Application Support/Sphinx 101/ and is preserved across upgrades.

Windows: Run the installer .exe.

The plugin installs to C:\Program Files\Common Files\VST3\Sphinx 101.vst3 (VST3). The manual and signal test results are copied to C:\Program Files\Analog Realism\Sphinx 101\. Your per-machine activation state (calibration.dat) lives in %APPDATA%\Sphinx 101\ and is preserved across upgrades.

Open your DAW, insert Sphinx on any stereo bus.

Authorization (one-time): Click the key icon, paste your serial, click Authorize. A challenge code appears - copy it, visit the combined activation/deactivation web page shown, paste your serial and challenge on the Activate side, click Activate. Copy the response code back into Sphinx, click Activate. Done - zero internet needed from the plugin itself. Your serial works on 2 machines. Re-activating after an OS reinstall is seamless - the system recognizes your machine.

Deauthorization: Click the key icon, click Deauthorize. Copy the removal code, visit the same combined web page, switch to the Deactivate side, enter your email and removal code, copy the response back, and click Confirm removal from the plugin itself. Your seat is freed. Need the removal code again? Just repeat - the system re-issues it.

Demo Mode: Full functionality with a brief 2-second silence approximately every 1.5 minutes.

Requirements:

• macOS: macOS 11 (Big Sur) or later, any AU-compatible host. Two separate single-architecture builds ship: an Apple Silicon (arm64) installer for M-series Macs, and an Intel (x86_64) installer that runs natively on Intel Macs and on Apple Silicon via Rosetta 2. Pick the one that matches the Mac you're installing on — Rosetta 2 is the Apple Silicon translator for x86_64 binaries and does not run on Intel Macs.
• Windows: Windows 10 or later, x64, any VST3-compatible host
• RAM: 4 GB minimum, 8 GB recommended for large sessions
• Disk: approximately 50 MB
• Activation: one-time internet connection for web activation

Five-Minute Quick Start

Insert Sphinx on your master bus, hit play. Try SLL (punch), Nevy (warmth), Amok (tube depth) main circuits. Push the Input TX Drive amount hard - that's iron-core saturation you're hearing, modeled with Jiles-Atherton magnetic-hysteresis equations and tuned to the harmonic signatures of real transformer builds. Switch transformer models and listen to how for example the bottom end changes character. Wiggle Width module to open the stereo image. Push the compressor threshold and feel the musical, natural glue. Try different EQ circuits to hear the true silky nature of real hardware EQs.

The Big Idea - Why Sphinx Sounds Different

Many analog modeling plugins use "black box" modeling: measure what hardware does, build a filter that copies it. Sphinx models what the hardware IS - every transistor, tube, transformer core, capacitor, circuit and inductor simulated as individual components wired in the real circuit topology.

The difference is profound. In black-box models, the EQ doesn't know the compressor exists. In Sphinx, everything connects through a shared power rail. When the compressor works hard, it draws more current, causing voltage sag that shifts the bias of every other stage. The EQ's inductors saturate differently. The transformer's operating point shifts. The entire circuit breathes as one organism. This is TrueRail - twelve interacting mechanisms producing complexity that no hand-tuned algorithm can replicate.

TrueRail - The Twelve Mechanisms

All twelve are always active, all verified by automated measurement:

1. Summing amplifier bandwidth - frequency-dependent gain rolloff adding subtle warmth no EQ can replicate
2. Component tolerance - every component randomized to real manufacturing specs, creating natural L/R stereo depth impossible with mathematically perfect circuits
3. Thermal drift - three slow oscillations modulating the circuit over time, so the sound is never quite static
4. Power supply sag - shared current draw creates the "glue" that makes analog bus compressors feel different from digital
5. Cross-channel crosstalk - frequency-dependent L/R coupling creating wide-but-cohesive imaging
6. Transformer hysteresis - Jiles-Atherton magnetic model with memory: the transformer remembers its recent history, producing program-dependent saturation no static waveshaper can match
7. Harmonic chain accumulation - each stage adds harmonics that interact through the chain
8. Class-A crossover nonlinearity - the texture people describe as "warmth" or "presence"
9. Crosstalk frequency shaping - coupling stronger at certain frequencies, consistent with how real PCB-trace coupling behaves
10. Compressor program dependence - behavior changes based on recent signal history
11. Transformer memory - a loud bass note changes how the transformer handles the next transient
12. Inter-module phase interaction - subtle phase shifts creating characteristic analog "depth"

Signal Flow and Routing

Audio passes through: Input Gain → Drive Stage → Input Transformer → Filters → Sub Bass Compressor → HF Compressor → HF SoftClip → Main Compressor → EQ → Width/Space → Output Transformer → Output Gain.

Input ─► [Input Gain] ─► [Drive Stage*] ─► [Input TX†] ─► [Filter] ─►
         [Sub Comp] ─► [HF Comp] ─► [HF SoftClip] ─► [Compressor] ─►
         [EQ] ─► [Width / Space] ─► [Output TX†] ─► [AutoGain] ─► [Output Gain] ─► Output

  *  Drive Stage = main circuit (SLL / Nevy / Amok) — always active
                   when the Input module is enabled.
  †  TX = Transformer (optional, independently bypassable per side).

  ┌──────────── TrueRail™ shared infrastructure (always active) ────────────┐
  │  Power Supply Rail  │  Crosstalk Matrix  │  Thermal Drift                │
  │  Component Tolerance │ Instance Seed     │  Summing-Node Bandwidth       │
  └─────────────────────────────────────────────────────────────────────────┘

Band-limited modules (Sub Comp, HF Comp, HF SoftClip) use sidechain-only processing - they detect energy in their frequency range and apply gain changes to the full-band signal but only in their own frequency areas, avoiding the phase artifacts of multiband splitting. Your audio is never chopped up and reassembled.

The Three Souls - SLL, Nevy, Amok

SLL - solid-state precision inspired by legendary British mixing consoles. Push-pull-style BJT topology modeled with the Ebers-Moll equations. The BJT's exponential transfer curve produces balanced even/odd harmonics (H2 ≈ H3). The cleanest of the three personalities at the -6 dBFS operating sweet spot, and the fastest transient response. Choose SLL when you want "finished" — polished and professional.

Nevy - single-ended Class-A discrete transistor with asymmetric off-centre bias (quadratic H2 enrichment) plus a gentle low-shelf around 200 Hz modeling classic iron-core output-transformer coupling. The asymmetry lifts H2 clearly above H3 — well above SLL's balanced harmonics while staying clearly below the tube path's depth. A true warm-transistor middle ground — not a tube clone. The compressor has a distinctive "hang" in its release — a slow initial decay that accelerates, creating the breathing quality real-life users and engineers love. Choose Nevy when you want "expensive."

Amok - tube-flavored signal path modeled with Norman Koren's plate-current equations on a 5670 dual triode (a mastering-grade vari-mu valve, not a 12AX7), full Miller capacitance, single-ended Class-A. The rest of the personality is a mastering-console scaffold: ±17.5 V op-amp rails, a VCA compressor with dual-detector (RMS + peak, max-combined) topology and soft knee, transformer-coupled summing bus. Complex even-harmonic distortion, the slowest rise time of the three (Miller capacitance effect), the deepest saturation. H2 dominates H3 by a wide margin — dramatically richer 2nd-harmonic content than solid-state. Choose Amok when you want "alive."

Each circuit changes the entire chain - not just the drive stage, but the compressor character, EQ behavior, and transformer interaction.

TrueRail Tiers - A, B, C

The TrueRail tier scales the per-instance manufacturing-tolerance spread. Same instance seed across all tiers - switching tiers does NOT re-roll the random draw, it widens or narrows the same underlying variation pattern.

Tier A - Factory Spec. Tightest tolerances (±0.5-1% typical, scale factor 0.10). Closest match between L and R channels, smallest deviation from nominal component values. The "studio reference" sound - what a freshly-calibrated production unit would measure.

Tier B - Production Standard. Moderate tolerances (±1-2%, scale factor 0.15). Adds natural character without demanding attention. The sweet spot for most work.

Tier C - Vintage Wide (default). Widest tolerances (±2-4%, scale factor 0.20) - hand-wired vintage character. Most L/R asymmetry, most deviation from nominal values, the most "alive" sound. The welcome preset ships on Tier C because the per-instance L/R variation it produces is the most audibly distinctive of the three tiers.

The buttons run left-to-right A → B → C, ascending from least to most variation - matching the "left=least, right=most" convention. Every plugin instance generates its own unique set of component tolerances - just like no two pieces of real hardware are identical.

Module Reference

Knob Interactions

Every value knob shares the same interaction set: drag vertically to adjust its value, hold Shift while dragging for fine adjustment, double-click a knob to type a value directly, scroll the mouse wheel for fine or fast adjustment depending on scroll speed, and Alt+Click (Option+Click on Mac) to restore the value the knob held before your last edit. The Alt+Click undo is single-step per knob — a second Alt+Click does nothing until you make another change. The undo state persists when you save and reload your DAW session.

Input Gain

Adjusts the input level into Sphinx with a range from -12 to +12 dB. Use Input Gain to hit the -6 dBFS sweet spot when necessary.

Hold Shift while dragging (or while scrolling the mouse wheel) to simultaneously counter-rotate the Output Gain by the same amount in the opposite direction — drive the plugin harder without changing the output level. Useful for auditioning saturation and color without volume bias.

Transformers

Four distinct transformer models, available on both input and output positions. Each represents a different core material with its own magnetic character:

M1166 - silicon steel core. Warm, smooth saturation. Low-frequency resonant peak in the deep sub band adds subtle low-end weight. The classic "console sound." Best for: warmth and body.

C9049 - nickel alloy core. Authoritative, controlled saturation. Higher-frequency resonant peak (above the M1166's) gives a tighter, more defined low end. Best for: precision and punch.

K1166 - grain-oriented silicon steel. Tight, fast saturation with minimal overshoot. Best for: transient-heavy material that needs transformer color without smearing.

L1544 - amorphous strip core. The most transparent. Produces an odd-harmonic-dominant character (H3 greater than H2), consistent with the symmetric saturation behavior real amorphous strip cores exhibit. Essentially flat — no LF resonance peak, no audible saturation. Best for: transparency with a hint of iron.

The input transformer uses full Jiles-Atherton magnetic hysteresis — it remembers its recent magnetization history, producing saturation that depends on what came before. The output transformer uses memoryless Langevin saturation — same per-core harmonic shape, no magnetic-state memory carried between samples — which keeps the chain stable when upstream EQ or filter changes shift the spectral content (the equilibrium of a fully-stateful Jiles-Atherton model would otherwise tilt under filter HPF phase rotation at the ~19 Hz LF resonance and produce sub-bass overshoot; the memoryless output is the engineering choice that delivers a stable, predictable chain). The Output TX still carries the per-core LF resonance peak and HF rolloff that give each core its frequency-domain personality — those come from the transformer's linear shell, not from hysteresis. The compressor and EQ tests bound any sample-to-sample filter-state carryover (DC blocker, LF resonance biquad) to a negligible level, so a loud passage cannot leak residual energy into the next musical phrase.

Filters

High-pass and low-pass filters with circuit-modeled rolloff. These model the actual component behavior of passive filter networks. Positioned early in the signal chain - before the compressors and EQ - so they clean up the frequency extremes before dynamics processing acts on them. Like the EQ, these aren't just digital biquads - each circuit topology models the actual component behavior of real filter networks, including the way source and load impedances interact with the filter elements.

Pultey (Passive LC): - the gentlest slope. Modeled after passive inductor-capacitor filter networks with their naturally smooth rolloff and subtle phase character. The HPF has a gentle resonant bump near the cutoff that adds a slight "lift" just above the filter point - cleaning up the sub without thinning the low end. Best for: mastering, where transparency matters more than precision.

Nevy (Discrete Class-A): - steeper rolloff with the warm, slightly rounded character of discrete transistor filter circuits. The real component tolerances create a filter that isn't mathematically perfect - and that imperfection is what makes it sound musical rather than clinical. Best for: general mixing, cleaning up recordings without sterile artifacts.

Sllamok (Active Op-Amp): - hybrid of SLL/Amok circuits. Clean Butterworth-derived response with predictable, measured behavior. Minimal phase disturbance in the passband. Best for: surgical cleanup, post-production, situations where the filter must be invisible.

Maney (Tube Buffer): - passive filter network followed by a tube buffer stage. The tube adds subtle 2nd-harmonic warmth to whatever passes through the filter, so even the act of filtering becomes a tonal enhancement. Best for: creative filtering, when you want the HPF or LPF to add character rather than just subtract frequencies.

Sharp mode: Activates a steeper rolloff slope with a tighter Q at the cutoff frequency. In normal mode the filters roll off gently with broad, musical curves. In Sharp mode they cut more aggressively — useful for removing problematic low-end rumble or taming harsh high-frequency content without affecting the nearby frequencies. The Sharp HPF includes a small resonant lift at the cutoff frequency that prevents the bass from sounding "scooped" even with aggressive filtering.

Dynamic Modules

Each compressor module in Sphinx 101 includes their own selection of four circuit topologies, each modeled from real hardware behavior:

SLL (VCA) - fast, punchy, predictable. The hardest knee of the four. Single exponential release. The "glue" compressor. Best for: pop, rock, electronic — anything that needs rhythmic punch.

Nevy (Transformer-coupled) - warm, musical, forgiving. Medium knee width. Dual time constant release with characteristic "hang." Best for: jazz, soul, acoustic — organic material that needs cohesion without aggression.

Amok (Tube) - thick, deep, saturated. Soft knee. Tube-dependent harmonics increase with gain reduction depth — the harder it works, the more harmonics it adds. Best for: dense mixes, hip-hop, cinematic — when you want the compressor to be heard.

Maney (Vari-Mu) - gentle, program-dependent, self-adjusting. The softest knee of the four. Hyperbolic release (slower than exponential). The tube gain element shifts operating point with signal level, making it naturally program-dependent. Best for: mastering, classical, delicate material.

Sub Bass Compressor

Controls the low-end foundation. Operates via sidechain - detects energy below the crossover frequency and applies gain changes only to its own frequency area of the full signal, so your low end gets tighter without multi-band phase artifacts.

HF Compressor

Tames high-frequency energy - sibilance, harsh cymbals, brittle high-mids. Sidechain detects energy above the crossover frequency and applies gain changes only to its own frequency area of the full signal. Your highs get smoother without the artifacts of traditional de-essing.

HF SoftClip

A saturation-based high-frequency limiter - think of it as the analog equivalent of a tape machine's natural HF compression. Two circuits:

Tube - warm, rounded saturation. Even harmonics. Smooths harsh peaks into musical warmth. Sidechain detects energy above the crossover frequency and applies gain changes only to its own frequency area of the full signal.

DPX - tighter, more aggressive clipping with a hardware-inspired British character. Adds edge and presence while preventing harshness.

Timing is fixed - this is a clipper circuit, not a compressor. The response is inherent to the saturation characteristic, producing instant, transparent peak control.

Compressor

EQ

Sphinx 101's three-band EQ is not just another digital filter. Each band contains a real inductor model with saturation characteristics - at moderate settings it's a clean, precise EQ, but push the gains harder and the inductors begin to saturate, adding harmonic complexity that's physically generated, not artificially added. Four circuit topologies offer fundamentally different approaches to equalization:

Sllamok (Active Op-Amp): - hybrid of SLL/Amok circuits, clean, precise, zero coloration. Constant-Q design: the bandwidth stays the same regardless of how much you boost or cut. No inductor saturation — this is a purely mathematical EQ implemented with modeled op-amp circuits. The cleanest of the four topologies, with negligible THD at any setting. The surgical option. Best for: corrective work, precise notching, clinical frequency adjustments where you want zero tonal footprint from the EQ itself.

Nevy (Discrete Class-A with Inductors): - warm and musical with signal-dependent character. Proportional-Q: small boosts are broad and gentle, large boosts narrow and focused — the way vintage British console EQs naturally behave. Real inductor saturation adds warmth as you push the gains harder. Swinging inductance shifts the center frequency slightly with signal level — the subtle modulation effect that's the recognizable signature of vintage inductor EQs. Best for: musical tone shaping, adding warmth to vocals, giving guitars presence — anything that benefits from the EQ having its own character.

Pultey (Passive LC + Tube Buffer): - the strongest inductor saturation of all four circuits, modeled after the revered American passive equalizer. Overwhelmingly rich 2nd-harmonic warmth. The famous passive-EQ "boost and cut at the same frequency" trick works here: simultaneously boosting and cutting the same band creates a scooped-midrange-with-boosted-sub curve that's impossible with conventional EQ. The inductors saturate at the right point and in the right way — by design, acting as a cross between an EQ and a low-frequency limiter. Best for: mastering sweetening, adding weight and air simultaneously, the classic tube passive EQ sound that has defined hit records for decades.

Maney (Passive Parallel Topology): - modeled after a celebrated American tube passive mastering equalizer with a unique parallel band architecture. Unlike conventional series EQs where each band stacks on top of the previous one, the parallel topology sums bands independently — boosting multiple bands at the same frequency caps gracefully instead of stacking linearly. This "soft addition" means you can make aggressive-looking settings without the output becoming ugly. Strong inductor saturation with musical character. Best for: mastering, broad tonal sculpting, situations where you want to use multiple overlapping bands aggressively without the sound falling apart.

Three-band EQ with real inductor saturation modeling. At moderate settings, it's a clean, musical EQ. Push the gains harder and the inductors begin to saturate, adding harmonic complexity that static EQ plugins can't produce - because in Sphinx, the EQ components are actually saturating, not just filtering.

Four circuit topologies match the main circuit selection, each with subtly different Q behavior and saturation character.

Width Module

Three controls for psychoacoustic spatial enhancement, inspired by the dimensional processing found in classic hardware spatial processors.

Width (range: -100% to +100%) — tube-stabilized stereo expander. Positive values widen the stereo image by processing the side signal through a 5751 triode stage that adds subtle 2nd-harmonic warmth to off-center content, psycho-acoustically stabilizing the center image as the sides expand. Negative values narrow toward mono while retaining the tube's warm character — like pulling down the stereo bus width on a real console where iron and tubes are still in the signal path. At 0% the tube is bypassed: fully transparent. The sweet spot for most masters is +10% to +40%. The tube participates in the TrueRail shared power rail — when the compressor pumps hard, Width's spatial enhancement breathes with the dynamics. 12 o'clock = 0% (no change).

Space (range: -100% to +100%) — frequency-targeted spatial enhancement. Opens or focuses the stereo image by gently shaping the side signal's presence range (a broad bell centered around 3 kHz, the 2–5 kHz band where human spatial perception is most acute). Positive values boost that band on the sides — off-center elements like panned guitars, background vocals, and room ambience feel wider and more open. Negative values cut it, pulling the image toward a more focused, compact center. The shaped side then feeds the Width tube, gaining analog harmonic character. It is a steady-state filter — no modulation, no artifacts. At 0% fully transparent. Effective at subtle settings (+5% to +30%). 12 o'clock = 0% (no change).

Mono Sum (range: 20 to 200 Hz) — frequencies below the set value are smoothly summed to mono. Essential for vinyl cutting, club systems, and tightening the sub-bass foundation. The filter is transparent and artifact-free. Default: 75 Hz.

Output Gain

Use Output Gain for final level matching back into DAW. Ranges from -12 to +12 dB.

Hold Shift while dragging (or while scrolling the mouse wheel) to simultaneously counter-rotate the Input Gain by the same amount in the opposite direction.

AutoGain

Disabled by default — toggle on from the Output section when you want level-matched A/B comparisons of processing changes. Unlike conventional auto-gain systems that measure input vs. output levels (which fights compression and misjudges silence), Sphinx 101's AutoGain operates per-module — each processing stage independently estimates and compensates its own gain contribution, preserving the natural interaction between modules.

It uses parameter-based prediction for frequency-shaping modules and a slow (~6-second) running-average of actual gain reduction for dynamics modules. Crucially, AutoGain restores the average level only — the per-sample envelope is left untouched, so rhythmic pumping, transient character, crest factor, and dynamic shape are all fully preserved.

Turn off when targeting a specific output loudness for delivery. Does not compensate the Output Gain knob — that remains your final manual level control. Manual makeup gain knobs in each module add on top of AutoGain's correction.

The Parallel Mix Secret

The master Parallel Mix control is an effect-intensity scaler — it does NOT digitally mix wet and dry signals (which would cause comb filtering through all the parallel paths). Instead, it scales each module's processing intensity: at 0%, all modules operate at their default/bypass values. At 100%, all modules operate at their user-set values. At 50%, every parameter sits halfway between its default and your setting. This is why Sphinx's parallel mix sounds clean at every position. The per-module Parallel knobs work in the same spirit — think of them as "Strength" knobs, dialling each module's effect up or down. See each module's reference table for exactly what its Parallel knob scales.

Oversampling and Antialiasing

Oversampling processes the heaviest nonlinear stages at a higher internal rate to reduce aliasing. The DriveStage (main BJT/tube saturator) wraps its nonlinearity in a linear-phase FIR equiripple half-band filter (80 dB stopband target, 64 taps); both transformers (Input TX and Output TX) wrap their hysteresis cores in a polyphase IIR half-band. All three run at the user-selected oversampling rate (2× / 4× / 8× / 16×). The lighter nonlinearities — the EQ inductor saturation, the Compressor topology color polynomials, and the Width tube — run inline at base sample rate. Under typical mastering programme content their alias contribution sits below the per-topology component noise floor; under aggressive single-frequency test probes the measured alias levels are documented per element in the shipped HF Alias Regression test report.

All four oversampling levels are fully functional and user-selectable in real time. Higher rates reduce alias contribution from the OS-eligible stages at the cost of CPU. The default 2× is sufficient for most work. Use 4× or higher for final masters, especially when pushing the saturation stages hard.

The full-chain analog noise floor sits in roughly the −70 to −74 dB region from the complete per-topology component noise model — VCA / diode-bridge / triode shot noise, BJT shot inside DriveStage, tube shot inside Width when engaged, NE5534 op-amp noise in the summing node, per-topology PSRR coupling — plus ThermalDrift and TrueRail rail-sag sub-audio parameter modulation rendered as broadband smearing around each harmonic spike. This analog character floor is largely independent of the oversampling factor because the OS factor only changes the aliasing contribution from the oversampled stages, not the analog noise model. Real hardware behaves the same way: the console's own noise floor is the limiting measurement, not the digital conversion.

Metering and Visual Feedback

Calibrated to real hardware ballistics - 300 ms rise time, 1.25% overshoot (VU mode); 1 ms attack, 2.8 s decay (PPM mode). LED indicators use smooth opacity breathing, never binary on/off. VU meters calibrated to -6 dBFS = 0 VU (the recommended operating level). Gain reduction meter for real-time compression depth. Spatial LED phase indicator. Knob value display on hover and adjust.

Skins

Switch instantly between skins by clicking the skin selector "handle area" in the plugin's GUI faceplate.

Amok (default) - light-styled metal with industrial accents.

Nevy - Alternative aesthetic, same controls, dark-styled layout.

Gain Staging - Getting the Sweet Spot

Sphinx operates at -6 dBFS peak. This is where the nonlinear components produce their most musical harmonics.

Too hot (peaking at 0 dBFS): harsh highs, pumping, lost definition - reduce Input Gain 3-6 dB.

Too cold (below -12 dBFS): thin and lifeless - increase Input Gain.

The TX Drive knobs push transformer saturation independently: 100% anchors the harmonic balance to the model's nominal operating point (matched to published electrical measurements of the modeled core class); above 150% extrapolates the modeled physics into obvious warmth; below 100% backs off into cleaner-than-hardware operation.

Under the Hood - Components

Sphinx 101 models analog circuits at the individual component level. Each electronic part is computed from published component physics — Ebers-Moll BJTs, Koren tubes, Jiles-Atherton transformer cores — augmented with light per-model harmonic shaping so the audible character matches measured behavior of the modeled units. None of it is a generic saturation curve or static waveshaper. The componenttest suite verifies every model directly against its expected behavior; the sections below name the model and list what is measured.

Vacuum Tube - Koren Model

The Amok circuit's triode stage uses Norman Koren's plate current equations, applied to a 5670 dual triode (a mastering-grade vari-mu valve — the low-mu type used by classic tube mastering hardware, distinct from the higher-mu 12AX7 found in guitar/preamp circuits). The model computes plate current as a function of grid voltage, plate voltage, and tube-specific parameters (mu, kp, kvb, Ex). This produces the asymmetric transfer curve that gives tubes their characteristic even-harmonic warmth - H2 rises naturally with signal level, exactly as in real hardware. Miller capacitance is modeled for accurate HF rolloff, and a cathode-bypass capacitor sets the operating bias.

Verified: plate current accuracy, amplification factor (mu), H2 enrichment vs level, THD-rises-with-level curve, output level still alive at hot drive, per-instance seed variation in tube parameters.

Bipolar Junction Transistor - Ebers-Moll Model

The SLL and Nevy circuits use BJT gain stages based on the Ebers-Moll equations. The model computes collector current from base-emitter voltage using the transistor's forward current gain (hFE), Early voltage, and thermal voltage. The SLL topology uses push-pull configuration (symmetric, odd-harmonic dominant). The Nevy topology adds a DC bias offset for Class-A operation, producing the asymmetric transfer curve that enriches even harmonics. Two real part numbers are modeled — BC184 (NPN, low-noise audio) and BC214 (PNP complement) — with the Ebers-Moll equations populated from their datasheets.

Verified: BC184 THD at -6 dBFS, BC214 THD at -6 dBFS, NPN vs PNP transfer-curve differential (they should not be identical), monotonic THD vs level, output level alive under drive, per-instance seed variation.

VCA - Exponential Control Topology

The SLL compressor uses a voltage-controlled amplifier. The model computes gain as an exponential function of the control voltage, providing dB-linear gain control. THD at unity gain and gain accuracy at attenuated settings are modeled from published specifications for low-noise mastering-grade VCAs.

Verified: unity-gain output level lands at the calibrated reference, attenuation tracks the dB-linear law, monotonic THD ordering across grade tiers, low-level output stays above the per-topology noise floor.

Operational Amplifier - NE5534

Active EQ stages use an op-amp model based on the NE5534 — the industry standard for analog audio. The model implements the part's published datasheet behavior: finite gain-bandwidth product (10 MHz), open-loop gain (100 dB), output slewing into the supply rails, and the resulting THD from bandwidth-limited negative feedback at high frequencies. The summing-node op-amp also contributes its own datasheet-derived input noise to the chain.

Verified: unity-gain transfer at 1 kHz tracks the input voltage, finite-GBW HF rolloff visible above audio band, output clamps at the supply rails (slew/clip ceiling), per-instance seed variation in component values around the op-amp.

Passive Components

Resistors, capacitors, and inductors are modeled with per-instance tolerance variation. Each plugin instance receives a unique seed that determines its component values within the TrueRail tier's tolerance range:

• Tier A (tightest): ±0.5-1% — precision matched components
• Tier B: ±1-2% — standard professional grade
• Tier C (default, vintage wide): ±2-4% — aged, drifted components

Metal-film resistors show tighter tolerance than carbon-composition; film capacitors hold value more closely than electrolytics. Inductors exhibit swinging behavior — inductance decreases with signal level via the Jiles-Atherton current-dependent saturation curve, which affects EQ response under high-level transients.

Verified: metal-film 1% resistor standard deviation falls inside the tolerance band, carbon-composition spread exceeds metal-film, film capacitor 1% spread inside tolerance, electrolytic spread exceeds film, inductor saturates monotonically with current.

Transformer Cores - Jiles-Atherton Hysteresis

Four transformer core materials are modeled using the Jiles-Atherton magnetic hysteresis equations (see also section 9 — Transformers). The JA model computes magnetization from field intensity using five physical parameters: saturation magnetization (Ms), anhysteretic shape (a), domain coupling (alpha), pinning/loss coefficient (k), and reversibility (c). Each core has its own JA parameter set plus a per-core harmonic-shaping term tuned to match published electrical measurements of the modeled units — M1166 silicon steel ends up H2-rich (warm even-harmonic character); L1544 amorphous ends up H3-dominant with H2 effectively absent (the symmetric-loop signature real amorphous strip produces).

Verified: M1166 H2/H3 ratio asymmetric (>1.0), M1166 more H2-dominant than L1544, measurable M-vs-L spread in the H2/H3 ratio, per-instance JA-parameter tolerance variation non-zero.

Metering Ballistics

VU, PPM, and LED meters use standard-compliant ballistics:

• VU: ~300 ms rise time, 1-1.5% overshoot, calibrated 0 VU = -6 dBFS.
• PPM: <20 ms attack, ~2.8 s decay (IEC Type-I shape).
• LED: smooth opacity breathing (never binary on/off), per-LED attack and decay so the bar feels natural rather than digital.

Verified (VU): needle rests at zero with no input, 0 VU lands at the -6 dBFS reference tone within the IEC tolerance band, needle rises over the IEC-spec rise time. (PPM): needle at zero on silence, attack catches a transient within the IEC-spec attack time, sustained tone holds high. (LED): opacity responds to level, attack is smooth (not stepped), sustained signal settles high.

Robustness Validation

A 30-probe runtime audit (RV01-RV30) drives the full plugin through pathological inputs — zero-length buffers, single-sample buffers, transport jumps, repeated prepareToPlay calls, full-scale and hot signals, every knob at min/max, rapid circuit-switch crests, skin reloads, preset restores, two simultaneous instances — and asserts that every output sample is finite, bounded, and click-free. The same audit verifies AutoGain doesn't kill level, flat-EQ doesn't color, parallel mix doesn't comb, mono-sum stays smooth, the sidechain HPF does not leak into the audio path, and the entire plugin produces bit-identical output run-to-run (deterministic seeding).

Verified: 30 robustness probes pass on every build (see the Signal Verification chapter for the full list).

Shared Infrastructure (Always Active)

Beyond the per-component models, six "chassis" subsystems are always on - they are how the components interact, not what they do individually:

• TrueRail Power Supply (RailState): shared virtual rail. Every module pulls current from it in proportion to its current workload, AND every module reads the resulting rail-voltage deviation to adjust its own operating point — transformer saturation headroom, output PSRR coupling, inter-channel crosstalk, and per-module bias all respond. Rail sag is personality-dependent: SLL is the tightest (most headroom, least cross-module interaction); Amok is the loosest (most "breathing"); Nevy sits in between, producing the characteristic "glue."
• CrosstalkMatrix: frequency-dependent L/R coupling through virtual ground impedance.
• ThermalDrift: three independent slow LFOs modulating component values over time.
• SummingNode: finite-GBW summing amplifier with frequency-dependent gain rolloff.
• InstanceSeed: deterministic per-instance random draw of component tolerances, scaled by TrueRail tier.
• Per-Module Bypass: every module uses a smooth 20 ms crossfade for click-free enable/disable. When fully bypassed (enableGain = 0), output equals dry input - true passthrough - and the module stops drawing current from the shared rail.

Technical Specifications

Formats: Audio Unit (AU) for macOS, VST3 for Windows.

Platform: macOS 11 (Big Sur) or later — separate Apple Silicon (arm64) and Intel (x86_64) installers; the Intel build also runs on Apple Silicon via Rosetta 2. Windows 10 or later (x64).

Sample rates: 44.1-192 kHz (verified identical character).

Buffer sizes: 64-2048 (verified bit-identical output).

Internal precision: 64-bit.

Oversampling: 2× / 4× / 8× / 16× user-selectable, applied to the DriveStage and both transformers. FIR equiripple half-band on the DriveStage (80 dB stopband target, 64 taps) and IIR polyphase half-band on the transformers. The lighter nonlinearities — EQ inductor saturation, Compressor topology color polynomials, and the Width tube — run inline at base sample rate. Per-element measured alias levels documented in the shipped HF Alias Regression report.

Noise floor: approximately -110 dBFS RMS (silent input) — the per-topology analog noise model floor.

Operating level: -6 dBFS.

Gain structure: unity loudness at default settings — output RMS tracks input within about ±0.5 dB at the -6 dBFS sweet spot. (Peak crest expands a few dB through the chain from cascaded analog phase rotation and oversampling inter-sample reconstruction, exactly as a real console strip does — this is loudness-transparent, not a gain error, and is verified on real program material.)

Bypass transparency: the TrueRail chassis still imparts a subtle baseline coloration — bypass is NOT bit-perfect by design (it's part of "always-on analog" character).

Polarity: non-inverting at all frequencies.

Latency: sample-accurate DAW compensation.

Parameters: 72, all automation-compatible.

TrueRail mechanisms: 12 simultaneous.

Authorization: one-time web visit, serial + C/R, 2 seats per serial.

Demo mode: 2 sec silence in approximately 1.5 minute intervals.

Signal Verification

Every copy is built from the same codebase that passes over 2,000 automated signal measurements across eleven categories on every release: full signal chain integrity (includes envelope-preservation and phase-integrity tests), compressor dynamic behavior, EQ accuracy, transformer verification (four iron cores: M1166 / C9049 / K1166 / L1544), module isolation, parameter sweep verification, oversampling and anti-aliasing, component-model verification with edge-case robustness tests, configuration audit, noise-floor regression gate (verifies every component model's noise source is live across personality × compressor × buffer-size combinations), music-material integrity, and HF alias regression (per-element alias floor at 15 kHz / −6 dBFS probe). The shipped test reports list the exact per-suite measurement counts for the version installed.

Beyond the synthetic test signals (sine, multitone, noise), every suite also runs a real-music level section: a calibrated commercial master is pushed through the actual welcome preset and metered the way a DAW meter does. This is the authoritative loudness check — mono and synthetic tones can mislead on level (a mono tone collapses the stereo side and hides the chain's true behavior), so the design's unity-loudness invariant is confirmed against real, bass-heavy, stereo program material alongside the synthetic measurements.

Signal Analysis & Measurement Guide

Sphinx is built to be measured. If you open it in an analyzer such as Plugin Doctor and inspect its harmonic distortion, Hammerstein kernels, or phase response, this section explains exactly what you might see and why — with the detail of a hardware technical sheet. None of the behaviors below are flaws; each is the physical consequence of modeling real console circuitry at the component level.

The three console personalities model three classes of legendary large-format console hardware. We use original names — SLL, Nevy, Amok — out of respect for the trademark holders, but the topologies and the measured harmonic signatures below identify the sonic lineage to any engineer who knows the genre:

• SLL — the transparent, solid-state VCA-bus British console class (BJT Class-A drive).
• Nevy — the transformer-coupled, discrete Class-A British console class.
• Amok — the valve (vacuum-tube) console class (single-ended Class-A triode).

All values below are taken at the −6 dBFS operating sweet spot with the default preset, the same conditions an analyzer uses by default.

19.1 Total Harmonic Distortion (THD)

The three personalities produce distinctly different, deliberate THD — this is musical harmonic content from the modeled circuitry, not digital distortion:

For exact current-release values, see the shipped Signal Chain Verification and Component Verification reports. The relative ordering (SLL < Nevy < Amok by orders of magnitude) is the architectural fact; the absolute percentages drift slightly between releases as the analog models are refined.

The harmonic balance is also visible in the Hammerstein-kernel G2/G3 levels, which rise with the "warmth" of the topology: SLL G2 ≈ G3 (balanced — the BJT's exponential transfer curve), Nevy G2 above G3 by several dB (transformer-coupled H2 enrichment), and Amok G2 dominates G3 by a wide margin (the 5670 triode's asymmetric transfer curve). These signatures match the harmonic profiles of the real console classes they model.

The measurement floor. Below the harmonic spikes sits a smooth analog noise floor in roughly the −70 to −74 dB region. It is produced by six physically-modeled device noise sources working together:

1. The SLL and Amok compressor circuits' VCA models include a continuous signal-modulated quiescent noise term, analogous to the noise floor of a real studio-grade low-noise VCA — the floor moves slightly with signal level, just like the real device.
2. The Nevy compressor models the shot noise of its balanced discrete diode-bridge; the Maney compressor models the plate shot noise of its vacuum-triode gain cell. Both are physically derived from In = √(2qIΔf) at the modeled quiescent current. These fill in the silent floor on the personalities that don't use a VCA, so the noise model is topology-complete — no compressor topology produces "digital silence."
3. The DriveStage's input transistor or triode adds its own shot noise — BJT shot on SLL/Nevy, triode plate shot on Amok — physically correct and inside the oversampled saturation loop.
4. The Width module's inline vacuum-triode saturator adds plate shot noise when engaged (Width ≠ 0 %); silent and bypass-transparent at Width 0 %.
5. The SummingNode's NE5534 op-amp model contributes its own input-stage noise.
6. Per-topology PSRR (Power Supply Rejection Ratio) couples residual rail-voltage deviations into the audio output amplitude. The SLL personality has the tightest rejection (sub-floor contribution); Nevy is moderate; Amok is the loosest (matches the looser regulation of tube-grade B+ supplies, and audibly part of the "tube console" character).

Plus, the ThermalDrift and TrueRail rail-sag models modulate circuit parameters at sub-audio rates, which an FFT analyzer with finite frequency resolution renders as a slight broadband smearing around each harmonic spike.

The transformer cores' low-frequency resonance peaks (e.g. the M1166 resonance around 19 Hz on the SLL personality) apply equally to signal and to the upstream noise floor; this is the same behavior real iron-core transformers exhibit and is part of the per-core sound. If you are evaluating transformer character at high Output TX Drive and high monitoring gain and the bass region floor sounds elevated, that is the LF resonance amplifying the upstream device noise floor through the modeled resonance — try the L1544 core (flat, no peak) for a quieter floor in that band.

What to look for. Clean harmonic spikes at exact integer multiples of the test frequency, rising above a smooth floor.

19.2 Hammerstein Kernels

A Hammerstein decomposition separates the nonlinearity into frequency-dependent polynomial kernels — G1 is the linear (fundamental) response, G2 the 2nd-order term, G3 the 3rd-order, and so on. Each personality has a distinct fingerprint:

• SLL: G2 and G3 sit close together (within a few dB) through the midband. The balanced even/odd character reflects the BJT's exponential transfer curve — "punchy but clean."
• Nevy: the even-order kernels (G2, G4) sit clearly above the odd-order kernels (G3, G5), with G2 the most prominent. This even-harmonic dominance is the signature of transformer-coupled iron — the "warmth."
• Amok: G2 dominates by a wide margin over G3, with the higher orders falling away progressively. This near-pure 2nd-harmonic character is the hallmark of a single-triode valve stage working on the knee of its transfer curve.

Low-frequency behavior. Below ~50–200 Hz the higher-order kernels roll off, and on the Amok personality they show a characteristic oscillation. This is the interaction of the coupling capacitors with the valve stage's plate-decoupling and cathode-bypass networks at sub-audio frequencies — the same behavior measurable in real tube equipment with iron coupling, and absent from the solid-state SLL and transformer-coupled Nevy paths.

High-frequency behavior. Above ~5–10 kHz the kernels converge into the noise floor. This is physically correct: real transformers and amplifiers are bandwidth-limited, which naturally suppresses high-frequency harmonic generation.

19.3 Phase Response

Sphinx's phase response is dominated by its transformer models. The DriveStage (main circuit) contributes effectively zero phase rotation in the audio band — its oversampling uses a linear-phase FIR half-band filter that adds only pure delay, fully compensated by the DAW's plugin delay compensation. Each transformer (Input TX and Output TX) contributes approximately one full rotation ("wrap") from its reactive elements: the coupling capacitor, winding inductance and resistance, and shunt capacitance that form the transformer's linear shell model. With both transformers enabled, the full chain shows roughly two wraps — monotonic downward rotation with no direction reversals.

The band-split compressors (Sub Comp, HF Comp, HF Softclip) are deliberately sidechain-only — their crossover filters act on a detection copy, never the main audio, so they add no phase rotation to the output. The EQ, Compressor color saturation, and Width modules operate inline at base sample rate without oversampling, contributing no additional phase. The measured phase is almost entirely the transformers.

At the highest frequencies (above ~18 kHz), a small amount of additional phase rotation appears from the IIR halfband anti-aliasing filters on the Input and Output Transformer oversamplers. This is inaudible and confined to the extreme top of the spectrum.

Because the phase rotation comes primarily from the transformers rather than the DriveStage, the three main circuits (SLL, Nevy, Amok) produce very similar phase responses. Subtle personality-specific differences appear at low frequencies from each circuit's bias point and gain structure interacting slightly differently with the transformer's reactive load.

Phase rotation and peak levels. Even with modest phase rotation, the frequency-dependent phase shifts from the transformer models reshape the waveform of complex program material. On music this can raise instantaneous peak levels by a few dB while RMS (loudness) stays at unity — exactly what happens when you run audio through a real analog console with iron transformers. It is not a gain error: a DAW peak meter will read higher, but a loudness (LUFS) meter will read the same as the input. AutoGain holds RMS unity; the peak rise is the physical consequence of frequency-dependent phase rotation on a complex waveform.

19.4 Oversampling and Anti-Aliasing

The heaviest nonlinear stages — the DriveStage (main BJT/tube saturator) and both transformers (Input TX and Output TX) — run at the user-selected oversampling rate (2× / 4× / 8× / 16×). The DriveStage uses an FIR equiripple half-band (80 dB stopband target, 64 taps) for linear-phase HF stop-band control; the transformers use IIR polyphase half-band filters. The lighter nonlinearities — the EQ inductor saturation, the Compressor topology color polynomials, and the Width tube — operate inline at base sample rate.

The shipped HF Alias Regression test report measures each nonlinear element's alias level in isolation at a 15 kHz / −6 dBFS probe (48 kHz session) and pins it as a per-item regression baseline with +3 dB headroom — so a future code change cannot worsen any element's alias level by more than 3 dB without failing the suite. The current-release values are documented in the shipped report; the picture they paint is consistent across releases:

• The DriveStage and the inductor-saturating EQ topologies (Pultey / Maney / Nevy) sit comfortably below audibility at typical operating levels — the OS pass on the DriveStage is doing its job, and the inline EQ topologies' nonlinearities are gentle enough that fold-back products land far below the music.
• The active op-amp EQ (Sllamok) measures dramatically lower than the inductor topologies because it has no inductor saturation — only a clean op-amp model, so its alias contribution sits far below audibility.
• The Width saturator at maximum engagement and the Maney compressor under gain reduction sit higher up the alias scale — still well below audibility under typical programme material but worth knowing about for the user who runs single-tone tests at extreme settings.
• The full-chain worst case (every nonlinearity engaged simultaneously at aggressive settings, single-tone probe) is the highest measurement in the report — by design, it documents the upper-bound aliasing behavior, not what mastering programme content actually produces.

Under typical mastering programme content the music itself fills the spectrum and most harmonic spurs are masked into the broadband analog noise floor (roughly the −70 to −74 dB region that the per-topology component noise model establishes — VCA / diode-bridge / triode shot noise, BJT shot inside DriveStage, tube shot inside Width when engaged, NE5534 op-amp noise in the summing node, per-topology PSRR coupling, plus ThermalDrift and TrueRail rail-sag sub-audio parameter modulation). Higher oversampling factors reduce the alias contribution from the OS-eligible stages further; the analog floor itself is largely independent of the OS factor because it's the modeled analog character, not the digital conversion.

The oversampling filters are phase-transparent in the audio band. The FIR half-band on the DriveStage adds only pure delay (linear phase), fully compensated by the DAW's plugin delay compensation. The IIR half-band on the transformers adds a small amount of HF phase rotation above the audio band — inaudible and at the extreme top of the spectrum.

FAQ

Does the plugin have a guide inside the DAW? Yes, click on the book icon on the lower-right corner of Sphinx GUI to open a popup guide.

Transfer license? - Deauthorize old machine, activate new.

Can't access old machine? - Email support.

Lost response code? - Repeat the activation or deactivation steps - it's re-issued.

Internet required? - One-time web activation only, from any computer. Plugin does not call home, ever, because everyone deserves privacy.

Server down? - Existing installs unaffected.

Mac + Windows? - Both ship today (AU for macOS, VST3 for Windows). Each install counts as one seat.

96/192 kHz? - Verified identical character.

CPU? - Around the usual industry top-dogs.

Multiple instances? - Yes, each gets unique component tolerances. Multiple instances never null due to their analog, unpredictable nature.

Troubleshooting

Plugin not found: - rescan plugins.

Clicks on DAW transport location changes: - increase buffer.

High CPU: - try 2x oversampling.

Sounds congested: reduce input 1-3 dB.

Challenge/Response rejected: - copy exactly, it's case-sensitive. Or email support with your serial and email.

Max seats reached: - deauthorize old machine or email support.

Air-gapped studio: - you can use any internet-connected computer with the web activation page to get the codes.

Credits and Legal

Sphinx 101 developed by Analog Realism. Component models based on published scientific equations: Koren (1996), Jiles-Atherton (1986), Ebers-Moll (1954). Made in Finland. (C) 2026 Analog Realism. All rights reserved.