AmpSim follow-up: measure Brad's Behringer GT2 clone before the class is built

[bdbarnett]

The effects program's follow-up additions (roadmap, after Phase 7) include AmpSim, modelled on the Tech 21 SansAmp GT2 (dossier: workspace docs/effects-internal/dossiers/AmpSim.md, a Station A discovery seed of 2026-09-08 graded on the owner's manual and a third-party component trace, since Tech 21 never published a schematic).

Brad owns a Behringer clone of the GT2 (2026-09-17) — the GDI21 V-Tone Guitar Driver DI, to be confirmed on the box — currently in storage. It is the first physical unit any drive dossier has had. Before the class is built: fetch it, and capture through the usbif/audiodev rig (or a USB interface) a measured transfer curve and harmonic spectra at each amp/mod/mode switch position and a few drive settings, at a stated input level. Those captures are gold reference material for the dossier, graded above the trace, and the SPICE run of the trace gets something real to be checked against — the accuracy program's inversion of the usual order (docs/effects-vision.md §4).

Not to be started before Phase 7; filed so the bench fact does not die with a session.

[bdbarnett]

How to connect it (Arthur, 2026-09-17; corrected the same evening). Through the PC and a USB audio interface with a line-level input, not through our boards — the measurement wants the most trusted chain, and the boards are what we are calibrating, not the instrument. Nothing on the bench is recorded as having one. My first draft named a Behringer codec "already on the bench"; that was a guess from usbif#1's "Burr-Brown CODEC (monitors)", which is the playback-only DAC feeding the monitors, and the C-Media mic is mic-level. So this needs an interface with line in and out (any PCM290x-class two-channel box, 48 kHz/16-bit is enough for the bandwidth argument below); until one is on the bench, nothing here can be run. The chain below assumes that box.

Chain: DAW or a Python capture script → interface line out → GDI21 1/4" input (1 MΩ, nominal −10 dBV / 250 mV, so consumer line level is the right ballpark; trim in the DAW) → GDI21 1/4" output (the amp-level one, not the mic-level XLR DI) → interface line in → record. Pedal on a 9 V battery, not the wall adapter, so the only ground is USB's and nothing hums; log the battery voltage per capture.

Calibrate before anything: (1) loopback, interface out straight into interface in, a 1 kHz sine at −12 dBFS and a log sweep — that is the chain's own gain and response, divided out of every later capture; (2) volts: play the same 1 kHz tone and read the interface's output with the DMM in AC volts, so every level in the record is a voltage at the pedal's jack and not a dBFS number nobody can reproduce.

Then per switch position (amp / mod / mode, photographed): a low-level log sweep for the frequency response, and 1 kHz and 100 Hz sines at stepped input voltages (about 10 mV to 1 V RMS) for the drive law and its harmonic spectra, read over whole cycles. Keep tones at or below 1 kHz so twenty harmonics sit under the interface's 24 kHz anti-alias filter; the pedal's higher harmonics are removed before sampling, not folded. Two limits to write into the dossier: the pedal sees a near-zero source impedance instead of a pickup (immaterial for an op-amp input like this one, unlike a Fuzz Face), and 48 kHz captures cannot see anything above 24 kHz.

Captures go under audiocomponents/.reference-captures/gdi21/ (gitignored, like the other reference packs) with a JSON manifest per file: switch positions, input volts, battery volts, the loopback calibration file it was corrected with.

[bdbarnett]

The interface (2026-09-17). Brad may have a Line 6 TonePort UX1 in storage. It fits: 1/4" line inputs and 1/4" analog outputs at 44.1/48 kHz, loopback-calibratable, and the line inputs have no gain knob in the path. It is not USB-class-compliant, so it needs Line 6's driver — their Windows 11 page lists the POD Studio/TonePort drivers as working (https://kb.line6.com/windows-11-compatibility), with community reports that Windows 11 refuses to keep Memory Integrity on beside that driver. No POD Farm or licence is needed for raw recording; the loopback capture is the check that nothing in Line 6's software touches the path. Recording is on Windows (REAPER over the Line 6 ASIO driver), not in WSL.

[bdbarnett]

Parked (Brad, 2026-09-23). It reopens when the effects program reaches AmpSim, after Phase 7. Phase 5 is itself parked. The bench notes above, the Behringer GT2 clone and the Line 6 TonePort UX1 as interface, are the plan at that point.