Setting up your rig

One loop, two cables:

interface OUT (first output pair) ──► pedal IN
pedal OUT ──► interface IN (the input you pick)

That’s the whole rig. PedalScope plays test signals out of the interface, through the pedal, and reads what comes back. Your guitar is not part of the loop — never plug it into this path, and never route the test loop to anything you’d hear at stage volume.

Which jacks

Interfaces differ at the panel and agree at the pedal, so the rule is stated at the pedal and your panel is read against it:

Calibration and measurement traverse the same electrical path, and that path is single-ended — one signal conductor and one ground — where it meets the pedal. A pedal jack is tip and sleeve, nothing else. The calibration is a picture of the loop it measured, subtracted from every measurement made afterwards, so the loopback you calibrate with must be the path the pedal will drive: same output pair, same input, same cables, same adapters. A different path in the calibration quietly poisons the compensation built from it. Plain TS (“mono” guitar) cables everywhere is that rule’s ordinary form — on a ¼″ output and a ¼″ input a TS cable is the whole path and there is nothing else to think about. When a rig can’t follow the cable rule literally, reason from the invariant.

One interface, doing both jobs. PedalScope needs an audio interface that both plays and records — one device, because calibration and measurement must traverse the same electrical path and the loop’s two ends must share a clock. A Mac on its own does not qualify: the built-in speakers and the built-in microphone are two devices, and combining them in a macOS aggregate device is not a supported measurement path — an aggregate that spans two clock domains applies drift correction, and drift correction is resampling, which the measurement path forbids in as many words. (Whether the built-in devices share a clock on Apple silicon is measurable and has not been measured; the route is unsupported, not proven impossible.) The headphone jack’s TRRS microphone contact is a microphone input: it supplies a bias voltage and expects millivolts, so a pedal’s output overloads it and the bias puts DC on the pedal’s output jack. The site’s FAQ carries the short form: which interfaces work.

Four things to establish about your interface, once:

• Where the signal leaves. PedalScope drives the interface’s first output pair — outputs 1 and 2, whatever the panel calls them (Main, Line Out 1/2, Monitor L/R, OUT 1/2) — with the same signal on both channels. There is no output to choose, and the Source panel says so; plug the pedal into either channel of that pair. On a ¼″ TRS output a plain TS cable does the single-ending for you: the plug’s sleeve shorts the jack’s ring to ground inside the interface. On an XLR-only output no TS cable fits, so the same thing is done by an adapter — XLR female to ¼″, pin 2 to tip, pin 1 to sleeve, and pin 3 either to sleeve or left floating; the interface’s own documentation says which its output stage wants, and the standard references disagree because the hardware does. Where that adapter sits is the one thing this page does not settle:

  • At the interface — a short XLRF→TS pigtail on the output, then a plain TS cable into the rig. The cold leg is terminated at the source, which is exactly what a TS plug does inside a ¼″ TRS output jack: the reference bench’s own topology, every cable after it plain TS, and the pin-3 decision a one-time property of one small part. A pigtail rather than a rigid barrel adapter, which is a long lever on a desktop unit’s jack.

  • At the far end — an XLRF→TRS cable into the pedal’s or the Breakout Box’s TS jack. Electrically almost the same, with one difference: the cold leg’s current now runs the length of the cable and returns along the sleeve, the conductor that is also the signal’s ground reference, so whatever voltage it develops across the sleeve’s resistance sits in series with the signal — a small, linear perturbation the calibration absorbs while the output stage stays in its linear range.

  Which is better on a given interface is confirmed, not argued: a bracketing null run through the Breakout Box, with the actual cable in place. If the floor sits where it sits on a known-good rig, the arrangement is fine. With a Breakout Box in the rig, interface → IN is the only link that changes for an XLR-only interface; everything downstream of IN is ¼″ and unchanged.

• The 6 dB, and why it isn’t a problem. Nearly every interface’s line outputs are balanced — a hot and a cold leg, on TRS and XLR alike — and the level in the manual is the differential figure. Single-ending one, by any of the routes above, uses one leg of the two, so the level that reaches the pedal is about 6 dB under the manual’s number. Measured for this guide on two interfaces from two manufacturers at three reference output settings (+9, +19 and +24 dBu nominal), the shortfall was the same 6.0–6.1 dB every time — through a plain TS cable on a ¼″ TRS output just as through an XLR adapter. It is a balanced-output tax, not an XLR tax: normal, expected of the ordinary balanced output stage (a few cross-coupled designs compensate and lose nothing), and the reference bench’s own figures already include it. So don’t expect an XLR rig to land 6 dB under a TRS one; expect every rig to land about 6 dB under its manual, and let the calibration absorb it — a static offset in a linear loop. What records where you actually landed is the volts factor: a property of one interface, which is why the app empties it when you pick a different device.

• Where it returns, and what the app calls that jack. The input you return the pedal into is what the app calls “Input 1”, “Input 2”, … counted from 1 in the order the driver reports the device’s input channels — which is not always the order printed on the panel. On a two-input box the XLR is usually 1 and the ¼″ jack 2; on a larger interface the ¼″ instrument jacks may come after the mic inputs, an XLR may be an output rather than an input, and the count itself can change with the sample rate (an ADAT-equipped interface has fewer channels at 96 kHz — the Input picker shows what the device has right now). Don’t predict the number from the panel: pick one, run the calibration with the loopback cable, and let the cable check say — it fails on the wrong jack, since nothing arrives, and passes on the right one.

• What sets that input’s sensitivity. Pedal outputs are low-impedance, so the line setting is fine and the instrument (Hi-Z) setting is fine too; what matters is that whatever sets the sensitivity — an INST/LINE switch, a PAD, a Boost / −10 dBV / +4 dBu selector, a gain knob, a control in the interface’s own mixer software — is chosen once, calibrated with, and left alone. A calibration is a picture of the loop with those controls where they were.

Two worked examples, one from each shape of interface:

• A two-input bus-powered box (the reference bench is a Focusrite Scarlett Solo). Outputs: two ¼″ TRS line outputs on the back — the first pair by definition — and a plain TS cable into either one. Inputs: the XLR mic jack is Input 1 and the ¼″ instrument/line jack is Input 2, so the pedal returns into Input 2, with the INST switch set once (either way) and the gain knob staged for the pedal. A Direct Monitor button on the front routes the inputs into the outputs — off, for the reason given under Listening while you measure.

• A desktop interface with XLR line outputs and a routing matrix (an RME Babyface Pro). Outputs: the XLR line outputs are outputs 1–2, so the pedal is reached through the XLR-to-¼″ adapter described above — into the Breakout Box’s IN where there is one. Inputs: the two XLR mic inputs are 1–2 and the ¼″ jacks are 3–4, permanently instrument inputs with a Boost / −10 dBV / +4 dBu sensitivity selector in place of an INST switch — so the pedal returns into Input 3 or Input 4, confirmed by the cable check. There is no direct-monitor button: monitoring is whatever the routing grid sends to each output pair, which is the point of the next section.

Levels

Interface line outputs can drive a pedal far harder than any guitar ever would. PedalScope expresses drive levels in dBFS — decibels below the loudest signal your interface can produce — and draws a shaded “guitar range” band on level axes for orientation. Two habits matter:

• Stage the input gain for the pedal first. With the pedal in the loop at the loudest drive you plan to use, set the interface input gain so peaks land in the −12…−6 dBFS window — using the converter’s resolution, with headroom to spare. Calibrate at that same gain (the calibration page walks the full order of operations), then leave every knob where it is; changing input gain or output volume afterwards silently invalidates the calibration. If a knob moves, recalibrate — it’s two minutes.

• Start at the defaults. The default sweep level (−26 dBFS) at minimum interface input gain lands near hot-pickup territory on typical interfaces. Push toward −12 dBFS to hear a pedal’s ceiling, drop toward −50 dBFS to find where it cleans up.

Listening while you measure

You’ll want to hear what’s happening, and headphones on the interface are the way to do it. What you hear there:

• The outgoing sweep, always. The headphone jack mirrors the playback being sent to the line outputs. Listening is a parallel tap of that signal, so the headphone volume knob is yours to move freely, mid-run included. (The main monitor/output knob is a different story: it sets the level actually driving the pedal, so it’s part of the calibrated chain — calibrate with it where you’ll measure, then leave it, like every other knob.)

• The pedal’s return, if you route it to your ears. A small interface has a direct monitor button or an input/playback blend knob; a matrix-routed interface has no such control — its software mixer is a routing grid, and any input sent to an output pair is direct monitoring of that pair. Either way the interface mixes the input into what you hear, in analog or in its own DSP, before the computer ever sees it. On interfaces whose monitor mix feeds only the headphone bus, that’s entirely outside the measured loop. But on several common interfaces — the smaller bus-powered boxes included — the same mix also reaches the line outputs, i.e. the very pair driving the pedal, and a return blended into the loop’s output is in the measured path. So the safe setting is nothing routed into the pair that drives the pedal — the direct monitor button off, or that pair’s input sends at −∞ in the routing grid — and listen on the headphone bus, which is outside the loop. A routing grid gives you something a button cannot: you can see that no input reaches the pedal’s pair, which is direct evidence rather than an inference. If you monitor through the measurement pair anyway, the loop carries it: treat the routing as part of the rig, set it before you calibrate, and leave it there. The calibration’s cable check measures the loop with whatever routing is active, so it describes the path you actually measured through. Whether it would single out a monitor mix as the cause is untested — so set the routing the safe way rather than relying on the check to catch it.

And start low: a full-range sweep is unpleasant listening at any volume, and a cranked pedal’s return is more so. Set the headphone level low before the first run and bring it up to comfortable.

The output level guard

The measurement path is a loop, but the interface’s output is not private to it. On most interfaces — the smaller Scarletts included — the pair driving the pedal is the same pair feeding the monitors and the headphone jack, so a sweep meant for a pedal also plays into the room, at whatever the monitor knob happens to be set to. Put headphones on before you notice that, and a run at the top of the level range arrives all at once.

PedalScope asks first, in two situations:

• Before the first test signal on a given interface. One screen, once per output: the loop is described and you’re reminded to turn the monitors down or take the headphones off. It doesn’t come back.

• Before starting output at or above −10 dBFS. The dialog states the level — in volts at the pedal input when your calibration measured the volts factor, in dBFS otherwise — and how far above the standard −26 dBFS measurement level it sits. Confirm once and the rest of that bench session at that level or quieter proceeds without asking; something louder asks again, and so does a different output.

Ordinary work never sees the level dialog. The standard sweep level sits 16 dB under the threshold, and the compression, gain-map and Live-mode ranges all stop at −12 dBFS — the whole default bench workflow is below it. What crosses the line is deliberate: a sweep pushed to −6 dBFS, or the test tone set loud.

The guard covers every path that puts a signal on the interface — measurement runs, family runs, the calibration sweep, the test tone, and Live mode. It is always resolved before a run starts, never between pressing the button and the first sample.

Test signals already fade in. Nearly every stimulus PedalScope emits rises from silence rather than snapping to level: the sweeps carry a 50 ms raised-cosine fade behind half a second of silence, the transfer-curve tone 20 ms, the two-tone 50 ms, the test tone 5 ms, and the compression ladder simply starts at its quietest step (−70 dBFS by default) and climbs. Those fades are part of the measurement — the sweep’s fade-in is literally what defines the bottom of the band it excites — so they are not adjustable, and there is no “safety ramp” setting to look for.

The exception is Live mode in Tone acquisition, whose stimulus is a seamless looping sine block: a fade inside a looping block would repeat as tremolo every block and corrupt the readings it exists to produce, so that one starts at level. It reaches full amplitude in about a millisecond, and the guard asks before it starts. (Live mode’s Sweep acquisition is faded like every other sweep.)

Source impedance (why your fuzz measures “wrong”)

A guitar pickup is a weak, high-impedance source; an interface output is a strong, low-impedance one. Most pedals don’t care. But fuzz faces, treble boosters, and other circuits whose input loads the pickup behave differently when fed from an interface — typically brighter and harder.

Practical fix: put a series resistor (start around 10 kΩ) — or a re-amp box — between interface out and pedal in, and keep it there for every measurement of that pedal. The absolute numbers still won’t be guitar-exact, but comparisons between settings and between pedals stay fair, which is what the charts are for. Declare it, so the record knows: set the pedal’s SOURCE control to the fixture and enter the resistance in the calibration’s source-impedance field. (A declared 10 kΩ series resistor is also the standard fuzz fixture of the PSC-1 condition — see Comparing measurements.) The build — one resistor, or the Breakout Box’s OUT 10k jack — and where along the cable it has to sit, are on Bench fixtures.

Hum and ground loops

The loop connects two devices to the same computer, so hum is usually mild, but if a pedal chains through a power supply shared with other gear you can pick up 50/60 Hz garbage that pollutes the low end of a measurement.

• Power the pedal from a battery or an isolated supply if you see unexplained low-frequency content.

• Keep the loop’s cables short and away from power bricks.

• Never lift a safety ground to chase hum.

Oscillation

High-gain pedals at some settings can self-oscillate — often ultrasonically or just below Nyquist, where you can’t hear it but every measurement can. PedalScope checks the stimulus-silent pre-roll of each sweep for strong narrowband tones (mains-hum families excluded) and warns — both during the run and as a flag on the saved record. When it fires: back off the gain, try a different power supply, and re-measure. Numbers taken while a pedal squeals to itself describe the squeal, not the pedal.

The warning shows its evidence. A verdict alone is hard to evaluate — a plausible squeal and detector confusion over gated fuzz hiss can look identical from a one-line flag. So the warning states what the detector actually judged: the tone’s frequency, its level in dBFS, how far it stood above the surrounding spectral floor, and where in the capture the analyzed stimulus-silent window sat. A Hear the flagged segment button plays the captured window itself (offline-extracted, normalized for listening with the applied gain labeled — ten seconds of listening settles what no threshold can). The snippet is kept in a regenerable cache; if the system has purged it on an old record, the numbers remain. Two honest caveats: a genuinely ultrasonic tone stays inaudible however much it is amplified — there the numbers are the evidence — and the detector only listens during the pre-roll, so it reports energy present with the stimulus silent, which no stimulus can explain.

The zero-cost hardware check. Your interface’s monitor path to the headphone bus — a direct-monitor button, or a routing-grid send — is outside the measured chain: headphones on it during the sweep let you listen to what the pedal is actually sending back, non-invasively — an audible squeal on decay tails is exactly what this catches. See “Listening while you measure” above for how that routing interacts with the loop on small interfaces (set it before you calibrate, and leave it).

The microphone permission

The first time you run a calibration or measurement — or play a sound through the interface, see Permissions — macOS asks whether PedalScope may access the microphone. That wording is Apple’s, not ours: macOS files all audio input — including your interface’s line and instrument jacks — under the Microphone permission. PedalScope never listens outside a run you started; allow it and the prompt never returns.

If you clicked Don’t Allow (or IT policy blocked it), every run will fail with a clear message and an “Open Microphone Settings…” link — flip PedalScope on under System Settings → Privacy & Security → Microphone and run again.

Before you measure

Run calibration once per interface + sample rate + knob setting. It takes the pedal out of the loop (straight cable out → in), measures the interface’s own latency, tone, and distortion floor, and subtracts them from everything you measure afterwards.