Manuals, downloads, and a real person at the other end.

Sphinx 101 is a master bus processor DAW audio plugin that models an entire analog mastering unit at the component, circuit and even rail level. Instead of sampling what hardware sounds like, we model the individual electronic components - every transistor, vacuum tube, transformer core, and capacitor - using published component physics (Ebers-Moll, Koren, Jiles-Atherton) and wire them together in real circuit topologies. The sound emerges from how those models interact, not from a static black-box filter. The result is multiple simultaneous analog modeling mechanisms (TrueRail) operating on your audio at all times, producing the depth, warmth and dimension that makes real hardware sound alive.

Most plugins model the input/output relationship of hardware - they measure what goes in and what comes out, then build a filter that copies it. Sphinx models what's INSIDE the hardware. Every component shares a virtual power supply, so when the compressor works hard, the voltage sags, and the EQ and transformers respond - exactly like real circuits. This interconnection produces emergent behaviour that no static algorithm can replicate. We verify this every release with 2,000+ automated measurements across thirteen independent test suites - full reports ship with the plugin and are public on this site.

Sphinx 101 ships in seven installers covering Mac and Windows. The two universal Mac installers (AU and VST3) each contain both Apple Silicon and Intel binaries in a single fat file - these are the recommended choice for most Mac users because a single download runs natively on any Mac. Single-architecture Mac builds are also available for Apple Silicon and Intel separately, useful when you need to force a specific-architecture install or troubleshoot. Windows uses a VST3 x64 installer. Minimum macOS 11 (Big Sur). Windows 10 or later. Approximately 30-50 MB disk space, and minimum 4 GB RAM recommended.

Any DAW that supports AU (Mac) or VST3 (Mac + Windows): Logic Pro, Ableton Live, Cubase, Studio One, Reaper, FL Studio, Bitwig, Digital Performer, Pro Tools, and others.

No. Sphinx runs on any Mac or Windows PC meeting the system requirements. No iLok dongle, no external hardware, no special audio interface needed. A good pair of headphones or monitors is recommended so you can hear the subtleties of the analog modelling.

The plugin itself never goes online - neither before, during, nor after activation. Only your browser does, on our activation page. You paste your serial in the plugin, copy the challenge code, paste it on the page from any computer with a browser, get a response code, paste that back into the plugin. The whole exchange takes about 60 seconds. No dongles, no cloud verification, no background services. We collect only your email, name, serial and machine ID - nothing else.

Two. Your serial activates on up to 2 machines simultaneously. If you need to move to a new computer, deauthorize the old one first - it takes 60 seconds and frees the seat immediately.

If you still have access to the old one: open Sphinx, click the key icon, click Deauthorize, and follow the short steps. Your seat is freed instantly and you can activate on the new machine. If you can't access the old machine (it died, was stolen, etc.), email support with your serial number and we'll reset your activation.

Yes. Sphinx runs in full demo mode without a serial key - every feature, every circuit, every transformer model, no restrictions. The only difference is a brief 2-second silence about every 90 seconds. This lets you evaluate the actual sound quality on your own material before committing.

Nothing happens to already-authorized installations. Your license is stored locally and works offline permanently. The server is only needed during the one-time activation moment. If it's temporarily down, just try again later - your serial doesn't expire.

Insert it on your master bus (or any stereo bus), hit play, and listen. The default preset is carefully tuned as a starting point. Then try switching between the three circuits (SLL, Nevy, Amok) to hear how each one changes the character of your mix without changing the volume. From there, experiment with the transformer drive and the compressor threshold.

Sphinx is calibrated for -6 dBFS peak. This is the sweet spot where the analog modelling produces its most musical harmonics. If your mix is peaking at 0 dBFS, pull the Input Gain down by about 6 dB. If it's very quiet (below -12 dBFS), bring it up. The VU meters are calibrated so that 0 VU = -6 dBFS.

Three complete circuit personalities. SLL is solid-state precision - tight, punchy, transparent, inspired by classic British console bus amplifiers. Nevy is transformer-coupled warmth - rich, full, dimensional, inspired by legendary British recording consoles. Amok is all-tube - deep, saturated, breathing, modeled with real vacuum tube equations. Each one changes not just the drive character but the entire chain: compressor behaviour, EQ response, and transformer interaction all adapt.

They select how much manufacturing variation is in the circuit's individual components themselves. They offer the experience of having three separate production units of the same hardware, with slight factory-build component variations. They offer a range of sonic options from subtle, nearly transparent - perfect for multi-instance use or when you want just a hint of analog warmth, to balanced amount of component variation - great for most work, up to maximum variation - closest to real hardware and ideal for final mastering where you want every nuance of real-life like circuits.

Two global controls that shape the power-supply and output-load context the modeled circuits sit inside. Mains selects the mains frequency of the virtual power supply: 50 Hz (default, European standard) produces a 100 Hz ripple fundamental after full-wave rectification; 60 Hz (North American / Japanese standard) produces a 120 Hz ripple fundamental. The ripple sits well below the noise floor either way, but the choice matches the mains you would find in your part of the world. OHM selects the output load (the termination) the transformers drive: 600 ohm (default, studio-console standard) keeps the low-end response tight; 150 ohm loads the transformer more heavily and slightly lifts the low-frequency shoulder; OPEN removes the load entirely for a more resonant, high-impedance-fed character. Neither control changes the fundamental sound of the plugin; both let you match the electrical environment the modeled circuits would have lived in.

Because every real piece of hardware does too. No two resistors are exactly 10 kΩ. No two capacitors are exactly 100 nF. Sphinx randomizes every component value within real manufacturing tolerances, seeded per plugin instance. Your left and right channels process through slightly different circuits, and your Sphinx sounds microscopically different from anyone else's - just like two units of the same hardware model are never perfectly identical.

Yes. Sphinx models physical component warm-up. When you first load the plugin, the virtual components start slightly below their settled operating temperature. The initial warm-up settles within the first second - fast enough that you will not notice it in normal use. Beyond the initial warm-up, the thermal model continues to respond to your programme material over time: loud passages gradually shift component parameters (transformer permeability, transistor bias voltage, tube transconductance) toward their heat-saturated values, and quiet passages let them cool back. This ongoing thermal breathing is physically motivated and contributes to the organic, session-specific character that separates component-level modeling from static processing.

Four physically different core materials, each with distinct character. M1166 (grain-oriented silicon steel): warm, rich, earliest saturation onset - the classic "console transformer" sound. C9049 (nickel alloy): authoritative, controlled, strongest bass compression - punchy and defined. K1166 (silicon steel variant): tight, fast, minimal overshoot - great for transient-heavy material. L1544 (amorphous core): ultra-transparent, odd-harmonic character, essentially zero saturation - when you want iron-path bandwidth without colouration.

It controls how hard you push the transformer's magnetic core. At 100 % (12 o'clock), the transformer operates at its designed sweet spot, adding subtle harmonic character. Push past 150 % and you'll hear the core saturate - bass gets rounder, harmonics get richer, transients get smoother. This isn't a digital gain boost followed by a waveshaper. It's the actual nonlinear saturation curve of the modeled core material responding to increased magnetic flux.

Four distinct topologies, each modeled from real hardware behaviour. The SLL circuit grabs fast with a hard knee - the classic VCA bus comp feel. The Nevy circuit has a characteristic "hang" in its release where it holds briefly before letting go - the musical breathing quality. The Amok circuit adds harmonics that increase with gain reduction - it gets warmer as it works harder. The Maney (Vari-Mu) circuit is fully program-dependent - it adapts to your material over time because the tube's operating point shifts with the signal history. In our measurements, beat 8 produces measurably different compression from beat 1.

No. Each EQ band has a real inductor model with saturation characteristics. At gentle settings, it's a clean, precise EQ. Push the gains harder and the inductors saturate, adding harmonics that interact with the frequency shaping - warmth that's physically generated, not artificially added. The four circuit topologies each have different Q behaviour and saturation character which real hardware users will recognize.

Only one knob in Sphinx is a true dry/wet blend, and it's called Parallel. The three frequency-selective modules and the chain-wide master knob use a different mix architecture called Strength, because true dry/wet on their signal path would comb-filter.

Parallel (Compressor only). The main Compressor's Parallel knob is a true dry/wet blend, the classic "NY-style" parallel compression: at 0% only the unprocessed dry signal passes through, at 100% only the fully compressed signal is heard, at 50% half the dry mixes with the compressed output so fast transients survive intact on the dry path while the wet path shapes the sustained body. This works cleanly because the main Compressor's wet path is a per-sample gain multiply with no crossover filter, so there is nothing to phase-mismatch when the dry copy is mixed back in.

Strength (Sub Comp, HF Comp, HF SoftClip). The three frequency-selective modules use sidechain-driven shelf filters, and their wet path is phase-rotated relative to a phase-flat dry copy - true dry/wet on those would comb-filter at the crossover corner. Instead, the Strength knob scales the depth of each module's band-targeted gain change in place: at 0% the shelf is at 0 dB (no effect), at 100% the shelf carries the full detected gain reduction. The audio is never band-split, so no phase mismatch can occur.

Output Strength (master). The chain-wide Output Strength knob in the Output section is a global effect-intensity scaler. At 0% every module operates at its default/bypass settings; at 100% every module operates at your chosen settings; at 50% every parameter sits halfway between its default and your setting. Rather than mixing wet and dry chains (which would comb-filter at every stateful stage), the master Strength scales each module's active-processing parameters in place, keeping the chain phase-coherent at every setting. A small loudness compensation is applied at the chain output so sweeping the master Strength stays perceptually level-matched.

For some work, yes. Sphinx 101's AutoGain uses static parameter-based compensation combined with a slow (~6-second) gain-reduction average to hold your loudness at unity. It preserves all dynamic shape - compression pumping, transient character, crest factor - and only corrects the static offset. It does not chase the signal sample-by-sample, so it never fights your compressor. It does not compensate the Output Gain knob, which remains your final manual level control. Turn AutoGain off when you're targeting a specific output loudness for delivery.

Sphinx runs the entire signal chain at the user-selected oversampling rate, so CPU scales with that selection: 2× for tracking and mixing, 4× or higher for final masters. The many simultaneous TrueRail mechanisms are interacting circuit models, which is more work than equivalent filter-based plugins. Your DAW's plugin CPU meter shows the actual figure for your machine and buffer size.

2× is the default and suitable for most work. 4× for final masters and critical listening. 8× and 16× for offline rendering. All four are selectable in real-time. The entire signal chain runs at the user-selected oversampling rate: a single minimum-phase polyphase IIR halfband oversampler upsamples at the input, every module - DriveStage, transformers, EQ, Compressor, Width - processes at the elevated rate, and a matching oversampler downsamples at the output. Minimum-phase means the impulse response is causal - no pre-ringing, the way every analog console behaves. The latency from the input and output filters is reported sample-accurately to the DAW for automatic delay compensation. Higher rates progressively lower the alias floor; per-configuration alias levels are documented in the shipped HF Alias Regression test report.

Yes, a small amount from the oversampling filters. The latency is reported sample-accurately to your DAW for automatic delay compensation. You won't hear any timing offset in your mix.

Yes. All 79 automation-compatible parameters (plus a non-automatable GUI Size preference) can be automated. Parameter changes are smoothly interpolated with no clicks or zipper noise.

Verified identical sonic character at 44.1, 48, 88.2, 96, 176.4, and 192 kHz. We run specific tests at every sample rate to confirm consistent behaviour across the full range.

The Signal Analysis grid shows Plugin Doctor measurements of Sphinx 101's three console circuits (SLL, Nevy, Amok) across three measurement types.

Top row (Hammerstein) shows the polynomial nonlinear-kernel decomposition of each circuit. Each colored line represents a kernel order - G(1) is the linear (fundamental) response, G(2) the 2nd-order nonlinear mechanism, G(3) the 3rd-order, and so on. SLL shows an odd-harmonic-dominant fingerprint, characteristic of push-pull BJT circuit symmetry that cancels even-order nonlinearity in the kernel decomposition - the signature of precision solid-state design. Nevy shows even-order enrichment, characteristic of Class-A single-ended topology whose asymmetric operation lifts the even-order kernels closer to the odd-order levels, producing the warm even-harmonic character of transformer-coupled discrete amplification. Amok shows an even-harmonic-dominant fingerprint, characteristic of the triode's asymmetric transfer curve (a low-mu mastering-grade dual triode of the kind used in classic tube mastering hardware) which produces prominent even-order kernel content, the deep, rich character of tube amplification. Note: Hammerstein kernel levels show the strength of each nonlinear mechanism, not the audible harmonic amplitudes directly - in the spectrum the harmonic ordering can differ because each harmonic line combines contributions from multiple kernels via Chebyshev expansion. Both the input and output transformers' Jiles-Atherton hysteresis adds memory that produces particularly strong higher-order kernels, so the audible harmonic spectrum reflects each topology's circuit physics rather than the raw kernel amplitudes. For current-release per-personality kernel levels, see the shipped Signal Chain Verification report. The architectural character described here is stable across releases; individual harmonic amplitudes may shift as the analog models are refined.

Middle row (THD) is Total Harmonic Distortion at 500 Hz. The pink spectrum shows all harmonic energy generated by each circuit. SLL is the cleanest, Nevy the warm middle ground, Amok the deepest - each sitting in the THD range expected for its modeled console class, exactly the order of magnitude real hardware in each class produces. Exact values vary slightly between sessions due to TrueRail's thermal drift modeling - the circuit's operating point shifts over time, exactly like real hardware. For current-release per-circuit measurements, see the shipped Signal Chain Verification report.

Bottom row (Phase) shows how each circuit shifts the phase of the signal across frequency. The distinctive curves below 200 Hz reveal each circuit's analog character - the low-frequency phase rotation that gives analog consoles their dimensional quality. Measured with the K1166 input transformer at 100 % drive.

All measurements were taken at -6 dBFS (our calibrated operating level), 48 kHz, with 16× oversampling and TrueRail mode B active.

Yes - this is one of the more physically faithful behaviours in Sphinx 101. The undulation is TrueRail's thermal drift modulating the virtual tube's operating point, exactly as happens in real vacuum tube hardware. A tube's plate resistance changes with temperature, shifting the harmonic balance over time. Since the left and right channels have independent thermal drift, they oscillate at different rates - creating the wave pattern.

SLL and Nevy use transistor topologies with stronger negative feedback that stabilises the operating point. If you measure a real tube preamp with Plugin Doctor for 60 seconds, you'll see the same behaviour. The tube is breathing.

No - these are the deliberate THD values of the three console personalities at the -6 dBFS operating sweet spot, and they match the console classes they model: SLL is the cleanest (transparent VCA-bus British console class), Nevy is the warm middle ground (transformer-coupled British console class), and Amok is the deepest (valve console class). Each sits in the range expected for its console class - exactly the order of magnitude real hardware in each class produces. This is musical harmonic content from the modeled circuitry, not digital distortion. It is exactly what gives each personality its distinctive character. (See the Owner's Manual, "Signal Analysis & Measurement Guide," for the full breakdown, and the shipped Signal Chain Verification report for current-release values.)

The floor is primarily analog modelling, not aliasing. Several mechanisms produce it: (1) the SLL and Amok compressor circuits' VCA models include a continuous quiescent noise term consistent with a real studio-grade low-noise VCA, (2) the Nevy compressor models the shot noise of its balanced discrete diode-bridge and the Maney compressor models the plate shot noise of its vacuum-triode gain cell, (3) the DriveStage's input transistor or triode adds its own shot noise inside the oversampled saturation loop, (4) the Width module's inline vacuum-triode saturator adds plate shot noise when engaged, (5) per-topology PSRR couples residual rail-voltage deviations into the audio output amplitude, (6) the ThermalDrift and TrueRail rail-sag models modulate circuit parameters at sub-audio rates, which an FFT analyser with finite frequency resolution renders as broadband smearing around each harmonic spike, (7) each device noise source carries a per-topology 1/f spectral shape - silicon paths sit close to flat, discrete BJT and diode-bridge paths carry a moderate low-frequency emphasis, and tube plate stages have the warmest low-frequency character, matching how real component noise spectra differ between silicon, discrete, and tube construction, and (8) a per-topology mains-frequency power supply ripple runs continuously on top of the rail sag (100 Hz at the default 50 Hz mains setting, 120 Hz when the Mains selector is switched to 60 Hz), modeling the residual mains-frequency content present in real power supplies (tightest on SLL, strongest on Nevy) at levels well below the analog noise floor. The entire signal chain - including the EQ inductor saturation, Compressor topology color, and Width tube saturator - runs at the user-selected oversampling rate through a single minimum-phase polyphase IIR halfband oversampler pair. The per-element and per-configuration alias levels are documented in the shipped HF Alias Regression test report.

That's the designed PSU ripple - a per-topology power supply oscillator built into TrueRail that models the residual mains-frequency content from full-wave rectification in real hardware power supplies. At the default 50 Hz mains setting the fundamental sits at 100 Hz; if you switch the Mains selector to 60 Hz the fundamental moves to 120 Hz, matching North American / Japanese mains. The amplitude varies by topology: SLL shows it smallest (matching modern tight regulation), Nevy shows it largest (matching an unregulated vintage supply), and Amok sits in between. The once-per-second pulsation is an analyzer artifact - the ripple free-runs at its fixed frequency and drifts in and out of phase with the analyzer's FFT frames, creating the slow beat pattern you see. The ripple sits well below the analog noise floor and is inaudible in a musical context - it contributes to the organic baseline rather than to audible hum. A real console on an analyzer rig shows the same behaviour from its actual power supply.

Yes. With the main circuit only (no transformers), the phase is very close to zero across the entire audio band. When transformers are enabled, each one adds its reactive phase signature - the curve you see is the physical consequence of iron-core transformer models with coupling capacitors and winding inductances. The shape and amount of rotation matches what you'd measure on real analog consoles with transformer-coupled bus paths.

Each transformer contributes approximately one full rotation from its reactive elements - the coupling capacitor, winding inductance and resistance, and shunt capacitance. With both the Input and Output Transformer enabled, the full chain shows roughly two wraps. The DriveStage contributes effectively zero phase rotation in the audio band. The chain's oversampling uses a minimum-phase polyphase IIR filter, contributing a small, monotonic high-frequency phase rotation in the same analog direction the transformers already rotate - well inside the cumulative analog-chain phase budget. The band-split compressors (Sub Comp, HF Comp, HF Softclip) are deliberately sidechain-only - their crossover filters never touch the main audio, adding no phase rotation.

No. These come from the valve stage's plate-decoupling and cathode-bypass capacitor interaction at sub-audio frequencies. The behaviour is measurable in real tube equipment and is unique to the valve topology - it does not appear on the solid-state SLL or the transformer-coupled Nevy. Below 5 Hz it is completely inaudible.

This is correct analog behaviour, not a gain error. Cumulative phase rotation from the modeled circuit shifts the waveform shape of complex material, which raises instantaneous peaks while RMS (perceived loudness) stays at unity - exactly what happens when you run audio through a real analog console. AutoGain holds the loudness; the peak rise is the physical consequence of frequency-dependent phase rotation. Your LUFS meter will confirm the loudness is unchanged.

Many emulations model a console's frequency response and harmonic distortion but use minimum-phase EQ curves rather than physically reactive component models. Sphinx's phase rotation comes from actual transformer models with coupling capacitors, winding inductance, and core hysteresis - producing the same kind of frequency-dependent phase shift you'd measure on real iron-path hardware. The result is approximately two full rotations in the full chain, consistent with what a real transformer-coupled analog console produces.

Rescan your plugin folder in your DAW's preferences. In Logic Pro, it usually appears automatically. If not, try restarting your DAW or your computer.

Make sure you copy the challenge code exactly from the plugin - use the Copy button, don't type it manually. The code is case-sensitive and any modification will be detected and rejected. If it still fails, click Cancel in the plugin, click Authorize again to generate a fresh challenge, and try once more.

Try reducing oversampling to 2×. Increase your DAW's buffer size. Bypass any modules you're not actively using. The entire chain runs at the selected oversampling rate, so dropping the rate is the largest single lever.

You're probably driving the input too hot. Sphinx is designed for -6 dBFS peak, not 0 dBFS. Pull the Input Gain down by 2-4 dB and listen to how the sound opens up. The nonlinear components sound their best when they're not being overdriven beyond their designed operating range - just like real hardware.

No. One-time purchase, yours forever. Free updates within the same major version. No recurring charges.

We offer a 14-day refund policy from the date of purchase. If Sphinx 101 doesn't meet your expectations, contact support with your order ID within 14 days and we will process your refund - no questions asked. The full demo mode lets you evaluate every feature on your own material before buying, so we encourage you to spend real time with the demo first.

No. The activation server stores only your serial number, email, name, and a hashed machine identifier for seat tracking. The plugin itself contains no analytics or telemetry and never connects to the internet. The website uses Google Analytics and Meta Pixel for anonymous visitor analytics - both are consent-gated and only activate after you accept cookies. Your serial, machine hash, and payment details are never sent to Google or Meta. We share data only with Paddle (payments), Google (page-view analytics), and Meta (page-view analytics). No data sold, no ads.

Email support. We're a small team and we actually read every message. Include your serial number if it's an authorization issue.