The board in the picture is the first physical BiFlux PCB. It's a 7-band graphic EQ, designed to sit in the effects loop of a tube preamp. It's the board the entire BiFlux project grew out of.
The original problem
I own a Mesa Boogie TriAxis. The lead voicings are beautiful. But where the rest of the Mark series — including my Mark V — has the canonical Mesa 5-band graphic EQ with five user-adjustable sliders, the TriAxis was built in a 1U rack format with no room for those sliders. Mesa kept the same EQ circuit, but hardwired the band gains into positions that form the iconic mid-scooped V-curve — the one Metallica made famous on the Mark IIC+. The "Dynamic Voice" control on the front panel just scales how deep that V goes.
It's a clever solution for a rack-format preamp. For high-gain rhythm, it delivers a lot of the classic Mark sound. But it's not a real graphic EQ. You can't lift one band without lifting all the others. You can't reshape the curve for a clean tone, or an edge-of-breakup tone, or anything that isn't the V.
The Mark V's actual graphic EQ — at its real electrical frequencies, which work out to about 88, 372, 723, 1576, and 4823 Hz, despite the sliders being labeled 80, 240, 750, 2.2k, and 6.6k — gives me what I want for the Mark. But on the TriAxis, I'm stuck with the V. And the TriAxis controls in general step in increments that I often find myself wanting to land between — the gain, the EQ shape, all of it. I wanted finer resolution everywhere.
Anyhow, I went looking for a pedal EQ to put in the TriAxis FX loop. None of them work properly there. The pedal world runs on 9V or 18V supplies — a 9V pedal can swing maybe ±4 V peak, about +14 dBu absolute maximum, and that's before any margin for the gain stages inside. A tube preamp's FX loop will throw +14 dBu peaks at it without breaking a sweat. You hit transients above the pedal's headroom and the signal clips before it gets back into the amp's recovery stage.
So I designed my own.
What's on the board
Seven bands at 88.9, 183.8, 375, 734, 1577, 2765, and 4793 Hz. Five of those are the actual electrical frequencies of the Mesa Mark series graphic EQ — the same canonical bands the Mark V uses. The other two — 183.8 Hz and 2765 Hz — I added where I wanted finer resolution: between the bass and low-mid, and in the upper-mid where my cleaner, edge-of-breakup tones live.
The signal path is fully discrete because I didn't want to use op-amps. BC850C transistors throughout, crossfade-gyrator topology with ±12 dB per band, running on a +32 V rail. Every resistor in the signal path is thin-film, 0.1% in the matched positions and 1% elsewhere. One thick-film resistor wouldn't matter; you wouldn't hear it. But every resistor being thin-film, throughout the entire chain, is an audible improvement — quieter, less noise modulation under signal, no thermal voltage coefficient artefacts. It's the kind of decision that doesn't show up in a spec sheet but accumulates across the signal path.
For the frequency-defining capacitors — the ones that actually shape the gyrator response — I used C0G/NP0 ceramic instead of the polypropylene that's standard in audio designs. C0G is electrically cleaner than polypropylene (better dielectric absorption, zero voltage coefficient), and unlike film capacitors it has zero microphony — it doesn't pick up mechanical vibration from the chassis. In a guitar rack that lives next to tube amps and on stage near a cabinet, that matters.
Headroom enough that the summing stage swings ±7.4 V peak, about +16.6 dBu — well above what any tube preamp's FX loop will throw at it.
The band amplitudes are set by precision digital potentiometers, but those digipots act on the gyrator's inductor-simulation feedback path, not on the main signal. The audio you hear never passes through a CMOS analog switch.
That decision — keeping the audio path purely analog while letting the digital control system live alongside it — became the design center of the whole BiFlux project. The EQ is where I learned it could be done.
What's next
The schematic is locked. The 7th revision is back from fab. Next is the bring-up inside BiFlux itself: power-on, frequency response sweeps across all seven bands, distortion measurements, and listening tests in the preamp it was always going to live in.
— Sorin