What the Presence Knob Really Does

Why does my Twin Reverb feel so much tighter than my friend's Marshall, when the Marshall has a presence knob and mine doesn't?

That question, or something like it, comes up constantly, and the honest answer surprised me when I actually worked out the numbers rather than trusting the folklore. The blackface Fenders — the amps everyone assumes are the "loose, feedback-free" ones because they have no presence control — actually run some of the deepest negative feedback of any circuit in this family. The Marshalls people call tight and controlled mostly aren't, by comparison. What they have is a knob.

Let's take the whole thing apart properly.

The Short Version

• Presence is not a treble boost. It is a control that progressively removes negative feedback at high frequencies. Turning it up doesn't add treble — it lets treble the amp was already suppressing back through.
• Negative feedback is what makes a "tight" or "controlled" amp tight and controlled. It lowers output impedance, raises damping factor, and tames whatever the speaker's impedance curve throws at the amp.
• A blackface Fender AB763 runs about 12 dB of feedback and has no presence control at all. A stock 1970s Marshall Plexi or JCM800 runs about 6 dB — half as much, in the log sense — and needs a knob to manage even that.
• "JTM45 = Bassman" is a myth at the power amp, even though it copies the Bassman's exact circuit values. Because it takes feedback from a different transformer tap, the JTM45 actually runs about 15 dB of feedback — deeper than either amp above.
• A Vox AC30 has zero feedback. No loop at all. That's not a limitation the amp is compensating for — it's a deliberate, different design philosophy, and it's why an AC30 reacts to speaker and cab changes far more than anything else in this article.
• The JCM800's presence pot is commonly quoted as 5 kΩ. On the actual 1988 factory drawing it's a 22 kΩ rheostat. The 5 kΩ figure belongs to an older, different circuit.

What Everyone Gets Wrong

The word "presence" does real damage here, because it sounds like "more of the top end," and every other control on an amp panel that raises a frequency band works by adding it. Presence works by subtracting something else.

The loop, in one picture

Every amp in this article except the Vox closes a loop the same way. A sample of the output — taken from a speaker-jack tap on the output transformer — is fed back to the phase inverter, where it is subtracted from the incoming signal before amplification. That's negative feedback: the amplifier compares what it's about to do against what it actually did, and corrects.

speaker/OT tap ──R_nfb──●── feedback node
                        │
      phase-inverter ───●── shunt to ground:
      tail (from PI         fixed resistor, OR
      cathode)               a presence pot

The feedback resistor sets how much of the output signal reaches that node. The shunt to ground — fixed in a plain design, adjustable in a presence-equipped one — forms a voltage divider with the feedback resistor, and that divider ratio, called β (beta), is what actually reaches the phase inverter's feedback input.

More feedback reaching the loop means more correction, which means:

• lower output impedance at the speaker jack
• higher damping factor — the amp fights back harder against whatever the speaker's impedance curve does to it (this is exactly the mechanism in why reactive loads sound different from real cabinets)
• less distortion, tighter transient response, a more "controlled" feel

Less feedback means the opposite: looser, more speaker-interactive, generally perceived as bigger and airier but less precise.

So what does the presence pot actually do to that loop?

It changes the shunt leg — and every version of it does the same fundamental thing: it lets more high frequency signal escape the feedback divider at high presence settings, which means less high-frequency feedback reaches the loop, which means the amp's closed-loop response is allowed to rise back toward what it would do with no feedback at all.

Turn presence to zero: maximum feedback across the whole band, including highs. The amp is at its most controlled, and — because feedback flattens and tames a rising response — often at its darkest.

Turn presence to maximum: high frequencies drain out of the feedback path before they can be corrected, so at those frequencies the amp behaves closer to open-loop, which for most output stages means more gain and less damping right at the top of the band.

That is the mechanism. Not a boost circuit. A hole cut into the correction.

Running the Numbers

This is where it gets genuinely interesting, because the actual feedback amounts — worked from real factory schematics — overturn almost everything the folklore says.

The method, briefly

Feedback amount in decibels is:

$$\text{NFB (dB)} = 20 \log_{10}(1 + \beta A)$$

where $A$ is the amplifier's open-loop forward gain from the phase-inverter input to the feedback tap, and $\beta$ is the divider ratio set by the feedback resistor and the shunt leg. Both were worked out per-amp from the phase-inverter topology, the power tube transconductance, the output transformer's turns ratio, and the actual resistor values on the factory drawings — a documented, reproducible method, not a guess. The resulting figures carry roughly ±1.5 dB of uncertainty from tube-to-tube gm variation, which doesn't change any of the comparisons below.

The table that upends the folklore

| Amp | NFB resistor | Feedback takeoff | Shunt / presence | β (mid) | Feedback | |---|---|---|---|---|---| | Fender Deluxe Reverb AB763 | 820 Ω | 8 Ω tap | 47 Ω fixed | 0.054 | ≈7 dB | | Fender Twin Reverb AB763 | 820 Ω | 4 Ω tap | 100 Ω fixed | 0.109 | ≈12 dB | | Fender Princeton Reverb AA1164 | 2.7 kΩ | 8 Ω tap (cathode inject) | 47 Ω | 0.017 | ≈5.5 dB | | Marshall JTM45 | 27 kΩ | 16 Ω tap | 5 kΩ pot | 0.114 | ≈15 dB | | Marshall 1959 Plexi ('68–69) | 47 kΩ | 8 Ω tap | 5 kΩ pot | 0.068 | ≈12 dB | | Marshall 1959 Plexi ('70s) | 100 kΩ | 4 Ω tap | 5 kΩ pot | 0.034 | ≈6.5 dB | | Marshall JCM800 2203 | 100 kΩ | 4 Ω tap | 22 kΩ rheostat | 0.032 | ≈6 dB | | Vox AC30 (any) | — none — | — | — | — | 0 dB |

Three findings out of that table are worth sitting with.

The Fenders are not the loose ones

The Twin Reverb runs about 12 dB of feedback with no presence control whatsoever. That is deeper feedback than a stock 1970s Marshall Plexi. What the Twin lacks is not the loop — it's the knob. There is no way to dial that feedback out from the front panel; it is fixed at the factory value, all the time.

So "Fenders are loose and open, Marshalls are tight" — as a blanket statement about the power amp — has the physics backwards. It survives because Fender clean tones, played the way most people play a Twin, don't push the power stage into the territory where feedback amount is audible as "tightness." The character difference people are hearing is overwhelmingly the tone stack (see what a tone stack actually does) and the preamp gain structure, not the power amp's feedback loop.

JTM45 ≠ Bassman, even with identical parts

This is the single best example of why component values alone don't tell the whole story. The JTM45's power amp is, resistor for resistor, the same circuit as the Fender 5F6-A Bassman it was built to copy — same 27 kΩ feedback resistor, same 5 kΩ presence pot.

But the Bassman takes its feedback from the output transformer's 2 Ω tap, and the JTM45 takes it from the 16 Ω tap. Transformer tap voltage scales with the square root of impedance, so that's a ratio of $\sqrt{16/2} \approx 2.8\times$ more signal reaching the same 27 kΩ resistor. The result: the JTM45 runs around 15 dB of feedback against the Bassman's roughly 7 dB — more than double, in the sense that matters.

The "tight, punchy KT66 sound" people attribute to the JTM45's tubes is, to a meaningful degree, actually a feedback quantity that has nothing to do with what tubes are in the amp.

The Marshall story is a two-step loosening, and it's about 9 dB

Follow the Marshall lineage through the table and a clean historical story appears: JTM45 (≈15 dB) → late-'60s Plexi (≈12 dB) → early-'70s Plexi and the JCM800 (≈6 dB). Marshall progressively moved the feedback takeoff to lower-impedance transformer taps and raised the feedback resistor to compensate only partially, and the net result across roughly a decade was about 9 dB of feedback quietly removed from the power amp — the famous "looser, more aggressive" evolution of the Marshall sound, told entirely in one resistor and one transformer tap, no tube changes required.

The 2203's presence pot is not what people say it is

Forum lore consistently quotes the JCM800 2203's presence pot as 5 kΩ — the same value as the earlier Plexi-era circuit. The actual 1988 factory CAD drawing shows a 22 kΩ linear rheostat, wired in series with a capacitor as a branch across the fixed 4.7 kΩ shunt resistor, not as a pot sitting directly in the feedback path the way the older 5 kΩ design does.

That wiring difference has an audible consequence beyond the value mismatch: because it's a series-RC branch rather than a straight shunt pot, the coupling capacitor's reactance at 1 kHz works out to roughly 1.6 kΩ — comparable to the fixed 4.7 kΩ shunt it parallels. The 2203's presence control measurably thins the midband, not just the top end, which is not what "presence" implies and not what the simpler Plexi-era circuit does.

The AC30 isn't hiding a loop

Genuinely worth stating plainly: the Vox AC30 has no global negative feedback loop at all. Not a small one, not a hidden one — none. Every mechanism described in this article simply does not exist in that amp.

Which means an AC30 is maximally exposed to whatever the speaker and cabinet do to it — there's no correction pulling the response back toward flat. That's precisely the finding in why reactive loads sound different from real cabinets: a zero-feedback amp is the one where swapping the load changes everything, and the AC30 is the textbook case.

What This Means on the Bench

If you want a tighter power amp and your amp has a presence control, turn it down — not up. Minimum presence is maximum feedback, which is the tightest, most controlled the power section will ever be.

If your amp has no presence control, it is not automatically loose. Check what circuit it actually descends from before assuming. A Deluxe Reverb clone with no presence knob can be running more feedback than a Marshall with one.

Modding a feedback resistor changes more than "how tight it feels." It moves output impedance and damping factor, which changes how the amp interacts with your specific speaker and cabinet — see why reactive loads sound different for exactly how much that matters, and how much it varies amp to amp.

A tremolo circuit riding on the feedback loop is a real, documented thing — the Fender Princeton Reverb's optical tremolo bug shunts the feedback node directly, so its intensity control modulates loop gain and damping, not just volume. Worth knowing before you go hunting for a "volume-only" trem mod on that circuit; you'd be fighting the amp's actual mechanism.

Try It Yourself

Every number in this article is a live, adjustable value in Feedback Lab. Load the Vox AC30 preset — genuinely zero feedback, by design — next to a Marshall JCM800 2203, and watch the response at the speaker move dramatically on one and barely twitch on the other as you swap cabinet models. Then load the JTM45 beside the Fender Bassman 5F6-A and see the same resistor produce two completely different feedback amounts, purely from which transformer tap it's wired to.

For the mechanism this whole article depends on — how a fixed impedance turns into a frequency-dependent response at the speaker — why reactive loads sound different from real cabinets works through the voltage-divider maths from the other side.

Sources

• Per-amp feedback resistor, takeoff tap, shunt/presence values and calculated feedback-in-dB figures are schematic-verified data developed for Feedback Lab, read directly from factory drawings for each amp: the Fender 5F6-A Bassman, AB763 Twin/Deluxe Reverb, AA1164 Princeton Reverb, Marshall JTM45, the 1959 Super Lead across its '68–69 and '70s revisions, and the JCM800 2203 (1988 factory CAD, issue 4/6).
• The JCM800 2203 presence pot's 22 kΩ value and series-RC wiring were read directly off the 1988 factory CAD sheet, correcting the commonly repeated 5 kΩ figure, which belongs to the earlier Plexi-era shunt-pot circuit.
• The Marshall feedback-resistor progression (27k at 16Ω → 47k at 8Ω → 100k at 4Ω across the JTM45 through 1970s-production era) is documented Marshall circuit history, corroborated by the factory drawings themselves.
• Feedback-in-dB figures carry an estimated ±1.5 dB uncertainty from power-tube transconductance variation at idle versus datasheet test conditions; the network topology and divider ratios themselves are exact, taken straight from the schematics.
• The Princeton Reverb AA1164's optical tremolo circuit shunting the negative-feedback node is a documented feature of that specific circuit, distinct from the power-tremolo (bias-modulation) designs used elsewhere in the Fender line.