JTM45 British lead‑style · 1962–1966 · 45 W

Marshall's first amplifier, and the first time one maker's circuit crossed to another: the JTM45 is a close copy of the tweed 5F6-A Bassman, rebuilt with British parts. Fender's low-gain 12AY7 input valve becomes an ECC83 (the European 12AX7), the 5881/6L6 output pair becomes a pair of KT66 beam tetrodes, the mains-supply voltages run a little higher, and solid-state diodes handle the bias supply. Everything downstream — the four inputs, the direct-coupled cathode follower feeding a treble-middle-bass tone stack, the long-tailed-pair phase inverter — is the Bassman, almost part for part. The higher-gain input valve and the KT66 bottles are what turned a clean bass amp into the seed of British rock tone. Produced from 1962; the drawing here is the mid-1960s revision shared across the tremolo combos and the 1987 head.

Circuit walkthrough (short form)

Normal (V1A) and bright (V1B) channels, two jacks each (a 68 kΩ stopper per jack, the 1 MΩ leak at jack 1's tip) → V1 ECC83 (100 kΩ plates, shared 820 Ω cathode with 250 µF bypass) → 0.02 µF couplers → 1 MΩ volume pots (the 100 pF bright cap sits on the volume V1B feeds) → 270 kΩ mixers, the bright one bridged by a 56 pF capacitor → V3A ECC83 (100 kΩ plate, 820 Ω cathode) → V3B cathode follower, DC-coupled (100 kΩ cathode load) → TMB tone stack (56 kΩ slope; 270 pF treble, 0.02 µF bass and 0.02 µF middle caps; 250 kΩ / 1 MΩ / 25 kΩ pots) → 0.02 µF → long-tailed-pair PI: 82 kΩ and 100 kΩ plates, 470 Ω + 10 kΩ tail, both 1 MΩ grid leaks returned to the tail junction → 0.1 µF couplers straight onto the grids (the drawing has no grid stoppers) → KT66 pair, fixed-biased through 220 kΩ leaks, their screens fed through one shared 1 kΩ · 2 W and then a 470 Ω · 1 W per screen → output transformer, 27 kΩ negative feedback into the tail foot with the 5 kΩ presence control and its 0.1 µF wiper-to-ground capacitor.

Power: 360-0-360 HT → GZ34 (0.05 µF from its cathode to the return) → standby → +450 V reservoir → 20 H choke → +440 V node carrying the output transformer's centre tap (plates, chart 430 V), the screen feed, a 32 µF filter and the 8.2 kΩ → +380 V PI → 10 kΩ → +310 V preamp. Fixed bias, as drawn: the HT winding end that also feeds a GZ34 anode → a series resistor → a diode → an "8" filter → 16 kΩ → the grid-line node, which carries a shunt to the return, an unvalued electrolytic and both 220 kΩ leaks.

The one part of the drawing not carried over here is the tremolo: on the factory sheet an extra ECC83 (its "V2") drives a 2G374 transistor that shunts the V3A mixer/grid signal through a 0.1 µF capacitor, 10 kΩ, the 50 kΩ depth control and 33 kΩ. This is signal-amplitude modulation, with no connection to the output-valve bias line. The plain head omits the entire tremolo branch, which is why the valve numbering skips from V1 to V3.

Lineage

The JTM45 is drawn from the 5F6-A, and the two schematics line up stage for stage. The differences are exactly the ones that give the JTM45 its voice: an ECC83 rather than a 12AY7 at the input (more front-end gain), KT66 output valves, a shared 1 kΩ screen feed added ahead of the 470 Ω each screen keeps, and a stiffer, higher HT rail.

The tone network, as the drawing wires it

The Marshall drawing wires the stack exactly as the Fender 5F6-A sheet it copies: the 270 pF treble capacitor and the 56 kΩ slope resistor both leave the cathode-follower output; one 0.02 µF runs from the slope foot to the node shared by the treble pot's lower lug and the bass pot; the bass pot is a rheostat — the drawing loops its wiper to the foot lug — in series down to the middle pot's top lug; the 0.02 µF mid capacitor feeds the middle pot's wiper; and the stack's output is the treble pot's wiper alone, into the 0.02 µF phase-inverter coupler. The textbook redrawing of these parts joins the treble and bass wipers at one output node and hangs the mid capacitor on a rheostat-wired middle pot instead; the schematic, the board diagram and the tone-stack lab all follow the drawing.

Verification — against the printed factory chart

Simulation matches all 12 quantitative chart nodes within their tolerances. The phase-inverter cathode is about 42 V against the printed 40 V. An earlier redrawing grounded the 10 kΩ tail and incorrectly treated the chart value as disputed; the factory drawing ends that resistor at the presence/feedback junction, whose 5 kΩ track and 27 kΩ feedback resistor both return to ground. The largest deviation is 6.5% at the V3A plate, within the chart's ±20% tube-pin tolerance. The tail junction carries no numerical chart value: the drawing marks the inverter grids only “+”, so it is reported informationally.

The drawing letters the input valve's sections, and that fixes which channel is which: V1A's plate feeds the plain volume and V1B's plate feeds the volume carrying the 100 pF, so V1B is the bright channel. V1A's printed load reads "1■0K", its middle digit a solid blob that cannot be told from a 0 or an 8 on this drawing. The corpus keeps 100 kΩ because the chart prints both input plates at 220 V on one shared cathode, which a 180 kΩ / 100 kΩ pair could not give; that rules 180 kΩ out, it does not read the digit.

The screen feed is drawn as one 1 kΩ · 2 W from the node after the choke to a junction, then a 1 W resistor to each screen. The upper one reads "470"; the lower one's first digit is not legible, and 470 is kept because 670 and 870 are not standard values, which is elimination rather than a reading.

What the drawing does not settle: the bias

The bias network's values are only partly legible. The series resistor prints as "1■0K" with a middle digit that could be 5, 6 or 8, and the grid-line shunt as "5■K" with its second digit overprinted by a wire. The electrolytic on the grid line carries no value, and the "8" and "16K" are probable readings. No type is printed at the diode; the drawing's "1N4007" note is about replacing the GZ34. The only voltage printed on the network is a bubble reading "10 5V" on the grid-line node, and it cannot be the quiescent bias. The cans there are drawn minus-up, so the node is negative. Read as −105 V, it is far beyond the KT66's cutoff of about −46 V at the chart's 440 V screens: the valves would pass no plate current, and the plates could not read 10 V below the centre-tap node, as the chart prints them. Read as −10.5 V, the output pair would draw several hundred milliamps and drag the 440 V rail to about 380 V. The chart's own cell for the KT66 grids is an illegible handwritten note.

So the DC model holds the grid line at an ideal −50 V. Nothing on the drawing supplies that figure; it is the corpus's earlier assumption, kept so the model has a bias at all. The gated rails depend on it through the plate current the choke carries. The 0.05 µF that the earlier model treated as a bias filter is on the rectifier side of the standby switch, from the GZ34 cathode to the return, and the 25 µF it listed is not printed anywhere.