The negative feedback loop in a Marshall samples a fraction of the speaker output and feeds it back (with inverted phase) to the phase-inverter's unused LTP input — the cathode side of the long-tailed pair, NOT the signal-input grid. (The presence pot sits in the NFB return path as a frequency-dependent shunt to ground.) The amount of feedback determines:
- Power-amp damping factor: how tightly the amp controls the speaker cone. More NFB = tighter, less NFB = looser/spongier.
- Power-amp distortion onset: more NFB linearizes the power tubes; less NFB lets them clip earlier and contribute more saturation.
- Frequency response of the power-amp + speaker system: NFB flattens the response; less NFB lets the speaker resonance peaks come through more.
Stock 1959 (early Plexi): 47kΩ from the 8Ω speaker tap → high feedback, tight Plexi feel. Stock 1959 (late-1969+): 100kΩ from the 16Ω tap. Stock 2203: 100kΩ from the 8Ω tap → slightly less feedback than the early 1959 (due to the higher resistor) but same tap. Jose: 100kΩ from the 4Ω tap → ~√2 (≈0.707×) of the feedback voltage compared to the 8Ω tap, since speaker-tap voltages scale with the square root of the impedance ratio (transformer windings sized for power matching, not voltage). Net effect: looser power amp, more spongy feel, shifts the perceived distortion source forward into the preamp.
The Jose mod is NOT "no feedback" — it's reduced feedback. Players who want the Plexi tightness with Jose preamp work should leave the NFB stock; players who want the canonical Brown Sound feel under-the-fingers should do the NFB mod.