How to Measure Guitar Amplifier Output Power?

Now, this is a topic that I tackle on a daily basis. Since most of us are not technically proficient, and yet a certain part of the population has forgotten some elementary physics, I find that the vast majority of people choose to go with the (mis)information that has been circulating for decades.

So let's do the whole thing properly. What the number on the back of your amp means, how it is actually arrived at, and — the part nobody ever gets to — how that number relates to standing in a room and deciding which amp is louder. Along the way we get to settle two arguments I have had more times than I can count:

Why are tube amps louder than transistor amps? Or are they?

Why are Marshalls so loud?

The Short Version

• A "100 W" rating means 100 W at 1 kHz, into a resistive load, at almost no distortion. That is one operating point, and it is not one you have ever played at.
• Cranked into clipping, the same amplifier delivers considerably more. The waveform turns into a square wave, and a square wave's RMS value is its peak value — so the power roughly doubles. I measured 170 W on my 1976 Superlead.
• Doubling the power buys you 3 dB. That is barely a change. To sound twice as loud you need about 10 dB, which is ten times the power.
• Which means the watts are almost never the reason one amp is louder than another. Speaker sensitivity and cabinet size are, and a 4×12 is worth about as much as quadrupling your amplifier.
• Tube amps are not inherently louder. They are inherently usable while clipping, and we have all quietly agreed that clipping is normal.

What Everyone Gets Wrong

The mistake is not believing the number. The number is fine. The mistake is believing the number describes the amplifier, when it actually describes a single test condition that has almost nothing to do with playing guitar.

Here is what a spec sheet is really telling you when it says 100 W at 1 kHz, 0.01 % THD, 8 Ω:

Somebody fed the amplifier a 1 kHz sine wave. They connected an 8 Ω resistor instead of a speaker. They turned it up until the output just started to distort — 0.01 % of what came out being harmonic content that did not go in — and stopped there. At that exact point, the power delivered to the resistor was 100 W.

Every part of that is a laboratory condition:

• 1 kHz is one frequency. The amp will do something different at 100 Hz and something different again at 5 kHz.
• A resistor is not a speaker. A speaker's impedance swings from maybe 6 Ω to over 40 Ω across the audio band, and what the amp delivers changes with it.
• 0.01 % THD is cleaner than a guitar amp has ever been asked to be. In guitar-player terms, this is 100 W of clean mids.

So the industry number is a relative figure. It is genuinely useful for comparing two amplifiers of similar type against each other, and close to useless for predicting what either one does in a room.

The Physics You Actually Need

Not much, honestly. Three ideas.

Power is a rate

Power is the rate at which energy is transferred — joules per second. One watt is one joule per second.

$$P = \frac{W}{t}$$

In an electrical circuit, that works out to voltage times current:

$$P = V \cdot I$$

If we know the voltage across the speaker and the current through it, we multiply and we have the power. Easy enough? Not really — because that formula is written for DC, a steady voltage that does not move. Our signal moves constantly. So we need one more idea.

Why the test signal is a sine wave

You will read that a sine wave is used for testing because it is "closest to the real signal." It is not. A plucked string is a rich harmonic stack with a percussive attack and a decaying envelope; a sine wave is none of those things.

The real reason is much better than that. A sine wave is the only signal a perfectly linear amplifier passes without changing its shape. Feed a sine in, and anything that comes out at a different frequency was manufactured by the amplifier. That makes distortion trivially measurable — you look at the output spectrum, and everything that is not the fundamental is the amp's fault. No other test signal gives you that for free. It is also mathematically simple, which mattered a great deal more when this convention was set than it does now.

RMS, and why it is lower than the peak

A sine wave spends most of its time away from its peak. A DC voltage sitting at the same value as that peak would deliver far more energy, because it never lets up.

So to use our DC power formula with an AC signal, we need the AC value that does the same work as a given DC value. That is the RMS — root mean square. For a sine wave:

$$V_{\text{RMS}} = \frac{V_{\text{Peak}}}{\sqrt{2}} \approx 0.707 \cdot V_{\text{Peak}}$$

This is worth internalising, because the entire trick at the end of this article depends on it: RMS is not a fixed fraction of the peak. It depends on the shape of the wave. For a sine it is 0.707 of the peak. For other shapes it is something else entirely.

And by the way — when we quote AC values, we almost always quote RMS. The 230 V in your wall socket is an RMS figure. Put a scope on it and the waveform actually swings to about ±325 V. In the States, 120 V RMS peaks at about ±170 V.

Rule of thumb: Unless somebody explicitly says "peak," an AC voltage is RMS. Amplifier power ratings are RMS-based. So is your mains supply.

One more piece: THD

Total harmonic distortion. Strictly, it is the ratio of the RMS value of all the harmonics the amplifier added to the RMS value of the fundamental it was given, expressed as a percentage. The maths for extracting it is genuinely involved and we do not need it here.

What matters is the job it does in a spec: it is the stopping condition. The power figure is meaningless without it, because you can get almost any power number you like out of an amplifier if you never say how much distortion you accepted on the way. A hi-fi amp is measured at 0.01 %. Guitar amps are sometimes rated at 5 % THD, which is a substantially louder — and dirtier — operating point. Fender rates the Blues Junior at 15 W minimum into 8 Ω at 5 % THD, and it is printed right there on the factory service diagram.

Running the Numbers

What 100 W into 8 Ω actually looks like

We know that $P = V \cdot I$, and Ohm's law says $V = I \cdot R$. Combine them and you can work in voltage alone:

$$P = \frac{V^2}{R}$$

Rearranged for the voltage we need:

$$V = \sqrt{P \cdot R} = \sqrt{100 \cdot 8} \approx 28.3\ \text{V}$$

And the current:

$$I = \frac{V}{R} = \frac{28.3}{8} \approx 3.54\ \text{A}$$

So a 100 W amplifier into 8 Ω is producing about 28.3 V RMS and delivering about 3.5 A. That is the industry measurement, and that is what the sticker on the back is telling you.

Keep hold of that 28.3 V. It is an RMS figure, for a sine wave.

What happens when you crank it

Let's take a Marshall Plexi-style amplifier, my favourite example. Four EL34s, rated 100 W, and somehow so much more powerful-sounding than plenty of other 100 W amps — especially their solid-state counterparts.

Part of the answer is architecture. Older designs like the Plexi don't have a volume control at the end of the preamp; that arrived later, with the 2204/2203 master-volume circuits. The volume pot on a Plexi behaves more like a gain control, and everything downstream of it is wide open. Anybody who has played a good Superlead knows the amp is already extremely loud at 2 on the dial.

Early Marshalls are 100 W amps, and they deliver 100 W of "clean" — with the volume pot at about 2.

Go past 2 and the volume keeps climbing, along with our excitement and our neighbour's rage. But something specific is happening to the waveform. It starts to compress, then to clip, and as it clips it approaches a square wave.

Here is the interesting part. The peak amplitude cannot grow. It is pinned by the supply rail and the transformer; the tops of the wave hit a ceiling and flatten out. So if the peak voltage is fixed and the speaker impedance is fixed, why does the amp keep getting louder?

Because RMS depends on shape, and the shape changed.

A square wave never leaves its peak. It sits at $+V_{\text{Peak}}$ for half a cycle and $-V_{\text{Peak}}$ for the other half. Its RMS value is therefore equal to its peak value:

$$V_{\text{RMS}} = V_{\text{Peak}}$$

Our sine was 28.3 V RMS, so its peak was:

$$V_{\text{Peak}} = 28.3 \cdot \sqrt{2} \approx 40\ \text{V}$$

Clip that sine into a square wave and the RMS climbs from 28.3 V all the way to 40 V, with no change in peak amplitude whatsoever. Now recompute the power:

$$P = \frac{V^2}{R} = \frac{40^2}{8} = 200\ \text{W}$$

Twice the power, from the same amplifier, with the same rail voltage, into the same speaker. Nothing was upgraded. The waveform simply got fatter.

On the bench: 170 W into a dummy load from my 1976 Marshall Superlead, cranked. Not the theoretical 200 W — the power supply sags under that kind of draw, and a real guitar signal never becomes a perfect square wave. But comfortably more than the 100 W on the badge.

That gap between 170 and 200 is worth respecting rather than rounding away. Real amps sag. Real signals are complex waves, not textbook squares. The 200 W figure is the ceiling; where a given amp actually lands depends on its power supply, its output transformer, and how hard you are hitting it.

What This Means on the Bench

The 3 dB problem

Right. So a cranked Plexi makes 170 W instead of 100 W. Surely that is why it is so loud?

Barely. Here is the arithmetic that ruins everybody's day:

$$\Delta L = 10 \cdot \log_{10}\!\left(\frac{P_2}{P_1}\right)$$

Going from 100 W to 170 W:

$$\Delta L = 10 \cdot \log_{10}(1.7) \approx 2.3\ \text{dB}$$

Two and a bit decibels. That is a small change — noticeable side by side, nothing like the difference people describe. And the full theoretical doubling to 200 W only gets you 3 dB.

Worse: perceived loudness does not track decibels one-for-one either. The rough psychoacoustic rule is that about 10 dB is needed before most listeners will call something "twice as loud." Ten decibels is ten times the power. To make your 100 W amp subjectively twice as loud, you need a 1000 W amp.

Rule of thumb: Doubling the watts gets you 3 dB. Ten times the watts gets you "twice as loud." Amplifier power is a terrible loudness control.

So why does the Marshall win?

Because loudness in a room is decided almost entirely by what happens after the amplifier.

Speaker sensitivity is the big one. It is quoted in dB at 1 W measured at 1 m, and guitar speakers vary a lot — roughly 96 to 100 dB across common 12″ models. That 4 dB spread does not sound like much until you convert it back into watts: 4 dB is a factor of 2.5. Swapping to a speaker 4 dB more sensitive is worth more than swapping to an amp with two and a half times the power, and it costs a fraction as much.

Cabinet size is the other one. Four twelves radiate far more effectively than one, both from the sheer area and from the drivers coupling to each other at low frequencies. In practice a 4×12 runs roughly 6 dB ahead of a 1×12 on the same amp — which, in power terms, is the same as multiplying your amplifier by four.

Add those up and the picture is clear. A Plexi through a Greenback-loaded 4×12 versus a solid-state 100 W combo through a single mediocre speaker is not a 100 W versus 100 W comparison at all. It is closer to a ten-to-one fight, and the watts contributed almost none of it.

The tube-versus-transistor myth, settled

Which brings us to the last one.

Myth: Tube amps are louder than solid-state amps of the same rating.

Verdict: Not inherently. They are louder in the way we actually use them, because they stay usable past the clipping point and most solid-state amps do not.

Everything in the square-wave section above applies to any amplifier. A 100 W transistor amp driven into hard clipping also heads towards 200 W. The difference is that nobody does it, because most solid-state power stages sound genuinely bad clipped — their designers went to real trouble to keep them clean right up to the rail and then stop.

Tube amps got pushed past that line very early on, largely by accident, and it turned out we liked it. Then designers started building for it deliberately. So we compare a tube amp running several dB into its clipping region against a transistor amp running below it, call the difference "tube loudness," and reach for a mythological explanation for something that is RMS arithmetic plus seventy years of habit.

We are not comparing amplifiers. We are comparing use cases.

Try It Yourself

The load side of this story — the part that actually decides your volume — is what Feedback Lab is for. Load a Marshall-style preset, switch the cabinet model between a closed-back 4×12 and an open-back 1×12, and watch what happens to the response at the speaker. Same amplifier. Nowhere near the same volume.

If you want the other half of the power equation — how hard the tubes are actually working at idle, and how much of your supply that is eating — that is Bias Bench, and the arithmetic is laid out step by step in my article on biasing a fixed-bias push-pull power amp.

And on your own bench, with nothing but a multimeter that reads AC volts: connect a dummy load of known resistance, feed the amp a 1 kHz sine, and read the voltage across the load. Square it, divide by the resistance, and you have your amp's real power. Then turn it up until the wave squares off and read it again. The second number is the one you play at.

Careful: Never run a tube amp without a load. Do this into a proper dummy load rated well above the amp's output — not a handful of resistors out of a drawer — and remember that a 100 W load turns 100 W into heat.

Sources

• Fender Musical Instruments, Blues Junior combined service diagram, drawing 0057275000 Rev D. The 15 W / 10.95 V_rms / 5 % THD output test specification quoted above is printed on the schematic itself.
• Author's own measurement: 1976 Marshall Superlead into a resistive dummy load, cranked, roughly 170 W.
• Speaker sensitivity figures are the manufacturer-published dB/W/m ranges for common 12″ guitar drivers. Check the specific model rather than trusting the range — the spread within a single brand is wider than you would expect.
• The "10 dB sounds twice as loud" figure is a psychoacoustic rule of thumb out of equal-loudness work, not a precise constant. It shifts with level and with spectrum. Treat it as an order of magnitude, not a number to calculate with.

One thing I have deliberately left alone here is amplifier architecture — output transformers and impedance matching, and what a solid-state amp's "minimum load" rating really means. That is a whole article of its own, and it will get one.

Until then: measure it, don't guess it. Happy chugin'.