Helios

Helios is a stereo audio effect that adds 300B SET-style harmonic warmth to a signal. It's an asymmetric soft-saturating waveshaper — a memoryless nonlinearity — voiced specifically to produce the kind of even-dominant harmonic ladder a 300B vacuum tube in single-ended operation generates.

It's deliberately not a tube simulation. Helios doesn't model anode load lines, output-transformer saturation, or the dynamic compression of a real output stage. What it does is take the audible character — the harmonic ladder, the soft saturation, the gentle asymmetry — and deliver it through a curve that's clean, fast, and predictable.

The product surface is intentionally narrow: Drive, Mix, Auto Gain, and Bypass. The voicing is fixed; the curve shape is locked; the asymmetry is baked in. "Do one thing well" — in this case, the 300B.

Helios is meant to be heard within a few seconds of inserting it. The fastest way to hear what it does:

• Insert Helios on a stereo bus or track.
• Set Drive to around 3–6 dB. You'll hear a subtle thickening — the 2nd harmonic doing most of the work.
• Set Mix to 0 dB (full wet). Mix lower (more negative) to blend in the dry signal.
• Leave Auto Gain at its default — it compensates for the level changes Drive introduces, so the controls do what you expect.
• Push Drive higher to feel more saturation, lower to back it off. Watch the harmonics display to see the ladder build.

The default state is a transparent character pass-through with subtle warmth. The signature only really blooms with Drive turned up.

Above the knobs is an 8-bar display showing the harmonic content of the output signal. The bars are labeled 1f through 8f — multiples of the fundamental — with the amplitude on the vertical axis in dBFS.

The bars are color-coded:

• Even harmonics (2f, 4f, 6f, 8f) are drawn in amber — these are the harmonics the design centers on.
• Odd harmonics (3f, 5f, 7f) are drawn in gray — present but not dominant by design.
• 1f (the fundamental) is the tallest bar in any normal program material — always gray.

An important note: the display is probe-driven. Helios feeds a known test signal through the curve to measure the harmonic response, then renders the result. This is deliberate — it shows you the curve's character, not a live FFT of the current music. Practical implications:

• The display is decoupled from your actual input signal level. The bars show what the curve produces at a reference level, not what's currently in your audio stream.
• At high Drive settings, the display may show more activity than the music itself contains — the probe excites the curve harder than typical program material.
• Use the display to read the ladder shape — relative levels of 2f, 4f, 6f, 8f. Don't read absolute output spectrum from it.

For a true live output spectrum, run Bitscope after Helios for bit-depth and true-peak information, or use any standard spectrum analyzer plugin downstream.

End to end, what happens to a sample as it passes through Helios:

• Input — stereo, 32- or 64-bit float (host-negotiated).
• Drive gain — linear input gain set by the Drive control.
• Oversample up — minimum-phase polyphase FIR oversampling (~4x), to give the curve room to do its work without aliasing artifacts.
• Asymmetric soft-saturating waveshaper — the heart of Helios. A memoryless nonlinearity, voiced to produce the 300B-flavored even-dominant ladder.
• DC blocker — removes the static DC offset the asymmetric curve introduces.
• Oversample down — minimum-phase polyphase FIR, back to host sample rate.
• Auto-gain compensation — offsets the average loudness change from Drive.
• Mix — blend with the delay-matched dry signal.
• Output — stereo, same precision as input.

Total reported latency: ~12 samples at host sample rate (about 0.25 ms at 48 kHz). Plugin Delay Compensation handles this transparently; Helios is safe to drop into monitoring and live tracking chains.

One design choice worth flagging: the oversampler is minimum-phase, not linear-phase. A linear-phase oversampler would introduce pre-ringing that smears transient attack — an octave-up ghost preceding each kick or snare hit. Minimum-phase oversampling preserves the attack at the cost of small phase shifts at high frequencies, inaudible on real material. The transient handling matters more for warmth-on-music than phase linearity does.

Master bus / glue

Drive 1–3 dB, Mix 0 dB. Subtle thickening across the whole mix — the 2nd harmonic adds weight, the soft saturation acts as gentle compression. Try it on already-mixed material; if the harmonics display shows the 2nd in roughly the −40 to −30 dB range below the fundamental, you're sitting in 300B territory.

Bass tracks

Drive 4–8 dB, Mix 0 dB. Bass benefits enormously from 2nd-harmonic generation — the octave above the fundamental adds perceived weight on systems that don't reproduce the fundamental cleanly. A small-speaker bass trick that goes back decades.

Vocals

Sparingly. Drive 2–4 dB, Mix −3 to −6 dB. Helios adds body and analog warmth to a vocal without sounding processed. Mix lower if it starts to sound thickened.

Hi-fi playback chain

Drop Helios on the master output of a playback chain (in JRiver, MetaHost, or similar) for a single-ended-tube listening character without owning a tube amp. Drive 2–3 dB, Mix 0 dB, Auto Gain at its default. The effect is meant to be subtle here — analog character on digital playback.

Drum bus

Drive 3–6 dB, Mix −3 to 0 dB. Helios adds glue and presence to drum buses similarly to how a hardware tape machine or single-ended tube preamp would. The transient handling of minimum-phase oversampling matters here — kicks and snares stay punchy.

Helios is designed to live alongside the rest of the Pataki Audio catalog. A typical signal chain in an active-speaker setup might look like:

• Janus — linear-phase crossover splitting the bus by driver band.
• Voicing — Baxandall-style tonal balancing per band or on the sum.
• Helios — single-ended tube warmth on the master sum.
• MetaHost — routes all of the above across the multichannel speaker bus.
• Bitscope — diagnostic readout on the output of any stage you want to verify.

Helios pairs particularly well with future Chronos releases for "vintage gear" coloration: Helios handles the nonlinearity (memoryless saturation), Chronos handles the linear coloration (load an IR of any analog hardware you want to color the chain with). Together they cover the two halves of making digital sound like vintage analog.

Why does the harmonics display show more activity than my output spectrum has?

The display is probe-driven — it analyzes a known test signal through the curve, not your live audio. The probe excites the curve harder than typical music does, so the display can show more "ladder" than the music itself contains. Read it for shape (relative levels of the even harmonics), not for absolute output spectrum. For live spectrum, use Bitscope or any analyzer downstream.

I don't hear anything at Drive 0 dB.

Correct. Drive at 0 dB sits near the bottom of the saturation curve where it approximates unity gain — Helios is intentionally transparent there. Start at 3–6 dB to hear character. The plugin is designed to be heard when you turn it up, not to color everything by default.

Helios changes loudness when I turn Drive up. Isn't Auto Gain supposed to fix that?

Auto Gain compensates for the average level shift the curve introduces. Residual loudness change at high Drive is partly from saturation also compressing peaks — that's part of the character, not a bug. If the loudness change is bothersome, either keep Drive moderate or lower Mix to taste.

Why isn't there an asymmetry knob? Every other tube plugin has bias controls.

Locked by design. Helios's identity is the 300B voicing, with a specific asymmetry baked in. Exposing bias as a parameter would turn Helios into "a generic asymmetric waveshaper that happens to default to 300B" rather than "a 300B warmth plugin." Different product. If you want generic, every DAW has Saturn, Decapitator, etc. Helios does one thing on purpose.

Will there be more voicings — EL84, EL34, console-style?

Possibly in a future v1.x release if there's demand. v1.0 ships with a single 300B voicing on purpose — focused product, do one thing well first.

Why is there some odd-harmonic content in the display? I read this was an even-only design.

An earlier internal design did target strictly even-only via analytic-signal squaring through a Hilbert transform. That delivered spectrally clean output but the wrong sound — real 300B SETs have a small amount of odd-harmonic content (the 3rd at roughly −45 to −50 dB below the 2nd in the canonical 300B datasheet curves), and removing it removed the recognizable single-ended character. The shipping v1.0 architecture restores it. The honest framing: Helios is an asymmetric SET-style waveshaper, not a mathematically pure even-harmonic generator. The sound came out right when we stopped insisting on the math.

Is the latency low enough for live monitoring or tracking?

Yes. ~12 samples at host rate is about 0.25 ms at 48 kHz — well below any human perception threshold for monitoring. Helios is safe to use in live and tracking chains.

Does Helios introduce any DC offset?

The asymmetric curve produces a static DC offset by design — that asymmetry is what generates the even-harmonic content. A DC blocker on the wet path removes it before the signal hits Mix. You should not see DC at the output under any setting.

Helios began life as an even-harmonic generator using analytic-signal squaring — Hilbert-transform the input, square the analytic signal, take the real part. The math is elegant: theoretically clean even harmonics with no difference-frequency intermodulation, no DC, no odd harmonics at all. An earlier internal design (call it v2.5) was built and measured on that basis. It worked, in the sense that the spectrum was clean.

It didn't sound right. Through several listening sessions on real music, it became clear that real tube warmth — the sound you actually want when you reach for "tube warmth" — comes from the shape of the input-output transfer curve, not from the spectral purity of the harmonic content. The cleanliness was the wrong target. Real 300B SETs have measurable odd content, output-transformer interaction, dynamic compression — and removing all of that, however cleanly, removed the recognizable character.

The shipping v1.0 abandoned analytic-signal squaring entirely and rebuilt around an asymmetric soft-saturating waveshaper. The voicing was iterated by ear against canonical 300B references — the WE 300B datasheet, Stereophile measurement archives, diyAudio simulations — until the harmonic ladder lined up roughly with the 300B reference (2nd dominant, 4th at −35 to −40 dB below 2nd, 6th below −65 dB, with the small odd companion every real single-ended design carries) and the result sounded like a 300B on real material.

The acceptance criterion that mattered: sounds like the 300B, not just has the ratios. The numbers bound the voicing; the ear decided when it was right.