Why Guitar Pedals Sound Different on Batteries vs Power Supplies?

This is one of those arguments where both camps are wrong in an interesting way.

One side says batteries sound better in a fuzz, full stop, and reaches for words like "organic." The other side says a volt is a volt and it is all expectation bias. Neither is right, and the reason is that there are actually two different mechanisms in play, they have very different sizes, and almost everybody attributes the effect to the smaller one.

In true Slightly Technical fashion, let's take it apart.

The Short Version

• Yes, the difference is real, and on some pedals it is obvious.
• The dominant mechanism is rail voltage, not impedance. A fresh alkaline sits above 9 V; a used one sits well below. In a transistor fuzz, moving the rail moves every bias point in the circuit, which changes where and how it clips.
• The impedance argument everyone repeats is real but second-order — it matters only in pedals with little or no supply filtering, which mostly means vintage fuzzes and boosters.
• The genuinely important practical issue is positive ground. A PNP germanium fuzz cannot share a daisy chain with your other pedals. That is not tone, that is a short circuit.
• If you want the battery sound on demand, what you actually want is an adjustable voltage supply. That is the variable you have been hearing.

What Everyone Gets Wrong

Both sides make the same error: they assume the power supply's only job is to sit there and supply power.

In a well-designed pedal, that is very nearly true. In a poorly-decoupled pedal — which describes most of the circuits people argue about — the power rail is part of the signal path, and treating it as a passive utility is a misunderstanding of how the circuit works.

But before we get to that, the bigger effect, which is much less exotic.

Rail voltage: the mechanism nobody credits

A regulated 9 V supply produces 9.0 V. That is the entire point of a regulator: it holds that figure whatever the pedal draws, from full to empty, forever.

A battery does not.

• A fresh alkaline 9 V sits at about 9.5 V off-load, and a shade under that with a couple of milliamps drawn.
• As it discharges, the terminal voltage falls steadily. Pedal players routinely run batteries down to 7 V and below without noticing, because most pedals keep working.
• Its internal resistance rises as it depletes, so it also sags further under the same load.

So over the life of one battery, a fuzz sees its supply drift from about 9.5 V down past 7 V. And in a discrete transistor circuit, the supply voltage sets the bias points. In a Fuzz Face the second transistor's collector idles somewhere near half the rail. Drop the rail and that operating point comes down with it, the available headroom shrinks, and the clipping goes progressively more asymmetric and more gated.

That is the famous dying-battery fuzz sound. It is not mystical, it is a moving Q point, and it is the single largest reason a battery and a supply sound different.

Here is the tell that proves it: give somebody an adjustable-voltage supply, let them dial it to 7.4 V, and the "battery magic" appears with no battery anywhere in the circuit.

The Physics You Actually Need

Batteries have real impedance — more than people think

The 9 V "battery" is not a cell, it is six cells in series, and series resistances add. That construction is why the format has notably higher internal impedance than a single AA:

• Alkaline (IEC 6LR61) — roughly 1.5 to 3 Ω when fresh, rising substantially as it discharges.
• Carbon-zinc (IEC 6F22) — roughly 5 to 10 Ω even when fresh, because it is six flat cells stacked in a wrapper rather than six proper cylindrical cells.
• Lithium — the lowest of the three, and the flattest discharge curve, which is exactly why it sounds least like "a battery."

And, as Figure 1 shows, that impedance is not purely resistive. It has reactive components, it varies with frequency, and it varies with temperature and load. A battery is a slightly odd two-terminal network, not an ideal source with a resistor in front of it.

Regulated supplies essentially do not

• Switching supplies (SMPS) are compact and efficient, and put out high-frequency switching noise that has to be filtered rather than ignored.
• Linear supplies are quieter but heavier. Note the common mistake here: their residual ripple is at 100 or 120 Hz, twice the mains frequency, because full-wave rectification doubles it. Actual 50/60 Hz hum in a pedalboard is nearly always magnetic coupling or a ground loop, not supply ripple.
• Either way, a regulated output has an output impedance in the milliohm range at DC — three orders of magnitude below the battery. That difference is the whole impedance argument.

Why the power rail ends up in the signal path

Here is the part that is genuinely interesting.

Any signal current a stage draws has to come from somewhere and return somewhere. In a common-emitter stage, the collector load resistor connects to the power rail, so a portion of the AC signal current flows out of the supply, through the circuit, and back to ground.

If the supply is an ideal source, that current develops no voltage and nothing happens. If the supply has impedance, the current develops a small AC voltage on the rail itself — and because every stage in the pedal shares that rail, that voltage is now coupled into all of them.

That is a shared-impedance feedback path, and it is the real content of the impedance argument. It is not that the battery "adds warmth." It is that the battery closes a feedback loop between stages that a regulated supply leaves open.

The capacitor that switches it off

Which is why any modern pedal has a rail decoupling capacitor, typically 47 µF to 470 µF, right where the DC comes in. At audio frequencies that capacitor is a much lower impedance than the battery, so the signal current takes the easy route through it and the supply barely sees anything.

The reactance of a 100 µF capacitor at 100 Hz:

$$X_C = \frac{1}{2\pi f C} = \frac{1}{2\pi \cdot 100 \cdot 100 \times 10^{-6}} \approx 16\ \Omega$$

and at 1 kHz it is about 1.6 Ω. So even against a tired battery at 20 Ω, the capacitor is winning comfortably across most of the band — and at low frequencies, where it is weakest, is exactly where the difference survives.

This is the dividing line. Vintage designs — Fuzz Face, Tone Bender, early Big Muff, Rangemaster, LPB-1 — have minimal decoupling or none. Everything since has plenty. That is why the argument only ever seems to be about fuzz.

What This Means on the Bench

Which pedals actually care

Genuinely sensitive, mostly through the rail-voltage mechanism and secondarily through the impedance one:

• Vintage fuzzes — Fuzz Face, Tone Bender, early Big Muff. High gain, discrete transistors biased directly off the rail, little or no decoupling.
• Treble boosters and simple boosts — Rangemaster, LPB-1. Single-stage, minimal infrastructure, bias set by the rail.

Moderately affected:

• Bucket brigade delays. Their clock circuits care about supply voltage, and clock stability affects delay time and noise floor.
• Analogue modulation. LFO rate and depth can drift with rail voltage depending on topology.

Effectively immune:

• Anything with an internal regulator or a well-defined virtual ground — most Boss pedals, for example.
• Digital effects. Their DC-DC converters make the input voltage almost irrelevant within spec.
• Modern overdrives and compressors with proper rail decoupling.

The positive-ground problem — the one that actually matters

If you take one practical thing from this article, take this one, because it is the difference between "sounds slightly different" and "does not work at all."

Original-style germanium Fuzz Faces and Tone Benders use PNP transistors and are wired positive ground. The battery's positive terminal connects to the pedal's chassis and jack sleeves, not the negative one.

Now think about what happens when you daisy-chain that pedal to your negative-ground pedals from one supply. The supply's negative rail is common to every output on the chain. The fuzz needs its positive rail on ground. Your patch cables have already tied all the sleeves together. You have connected the supply's positive and negative rails through the pedalboard.

At best it does not work and hums horribly. At worst something gets hot.

Careful: A positive-ground fuzz needs either a battery or a fully isolated output on your supply. "Multiple outputs" is not the same as "isolated outputs" — a daisy chain with several plugs on it shares one rail. Check the manual, not the plug count.

This, far more than any tonal argument, is why so many fuzz players are still running batteries. They tried the supply once, it howled, and they went back.

Sag as a deliberate control

Once you understand that the variable is rail voltage, the useful move is obvious: stop trying to control it by accident.

Plenty of pedals now ship with a voltage starve or sag control, which is precisely a knob that lowers the internal rail. Plenty of supplies offer adjustable or sag-simulating outputs. Both give you the dying-battery voice on demand, repeatably, at the setting you liked, without carrying a bag of half-flat batteries to the gig.

That is the honest resolution of the whole debate. The players who insist they hear something are hearing something. They just have the wrong name for it.

Why simulating this is harder than it looks

You will see it claimed that SPICE cannot model this. That is not quite right, and the real limitation is more interesting.

LTspice will happily model a battery as a voltage source with a series R, an L and a C network in front of it — you can build an arbitrarily good equivalent circuit. The problem is not the simulator, it is the data. Manufacturers publish impedance at one frequency, at one temperature, at one state of charge, if they publish it at all. To simulate a battery honestly you need its impedance as a function of frequency and discharge state, and nobody gives you that.

So you can simulate the rail voltage mechanism perfectly well — sweep the supply from 9.5 V down to 7 V and watch the bias points move — and that is the dominant effect anyway. The impedance mechanism you can bound, but not pin down, without measuring your own cells.

Which is a fair description of a lot of this hobby: the maths is available, the data is not.

Try It Yourself

The rail-voltage mechanism is exactly what FET Playground makes visible. Set up a common-source stage, then change the supply voltage and watch the load line pivot and the operating point slide along the transfer curve. That is the same thing happening in your fuzz when the battery goes from 9.5 V to 7 V — different device, identical mechanism, and much easier to see when it is drawn for you. If you're characterizing a real JFET for a build rather than just exploring the mechanism, designing a JFET gain stage around a real part walks through measuring the specific device in your hand.

On your own bench, the experiment that settles it costs nothing:

1. Put a fresh battery in the pedal and measure the voltage while it is switched on and playing, not off-load.
2. Play it, and note what you like.
3. Set an adjustable supply to that exact voltage and plug it in.

If it sounds the same, it was the voltage. If it does not, you have found a genuine impedance effect and you have a pedal worth writing about. In my experience it is nearly always the voltage.

Careful: Measure with the pedal switched on and a signal running. A 9 V battery reading 9.4 V off-load can be sitting at 8 V under load if it is old, and off-load readings are how people convince themselves a flat battery is fine.

Sources

• Internal impedance figures: IEC 6LR61 (alkaline) roughly 1.5–3 Ω fresh; IEC 6F22 (carbon-zinc) roughly 5–10 Ω fresh, both rising as they discharge. These are manufacturer and standards figures for the 9 V format specifically — the six-cells-in-series construction is why they are so much higher than a single AA.
• The 100/120 Hz ripple figure for a linear supply follows from full-wave rectification, which doubles the mains frequency.
• Capacitive reactance figures are calculated from $X_C = 1/(2\pi f C)$ at the stated values.
• The positive-ground incompatibility is a topology fact of PNP germanium fuzz circuits, not a manufacturer quirk, and applies to any faithful reproduction of them.
• I have deliberately not put a number on the audibility of the shared-impedance mechanism. Doing that honestly requires measured impedance-versus-frequency data for the specific cell, which manufacturers do not publish.

Listen critically, measure obsessively, and don't settle for marketing hype when you can just check.