The circuit that made the tweed Bassman famous, and the one revision short of
the amp everybody copied. Introduced in July 1957, the 5F6 brought in the
three-knob treble/middle/bass tone stack, the direct-coupled cathode follower
that drives it, the long-tailed-pair phase inverter and the fixed-bias 5881
pair — the whole architecture of the 5F6-A, arriving a year early behind a
mercury-vapour rectifier. Produced 1957–1958.
Circuit walkthrough (short form)
Bright + normal channels (1M leaks, 68k stoppers) → V1 12AY7 (100k plates,
shared 820 Ω cathode with 250 µF bypass) → 0.02 µF couplers → 1M volume pots
(100 pF bright cap) → 270k mixers → V2A 12AX7 (100k plate, 820 Ω cathode
with 25 µF bypass) → V2B cathode follower, DC-coupled (100k cathode load) →
TMB tone stack (56k slope, 250 pF treble, two 0.02 µF caps; 250k/1M/25k pots)
→ 0.02 µF → long-tailed-pair PI: 82k (5%) and 100k (5%) plates, 470 Ω + 10k
tail, both 1M grid leaks returned to the tail junction, 47 pF across the plates
→ 0.1 µF couplers → 5881 pair, fixed-biased at −48 V through 220k leaks and
1.5 kΩ grid stoppers, 100 Ω screen resistors → output transformer into four
speakers, with 27k of negative feedback returning into the tone stack's ground
leg.
Power: 325-0-325 (PT 8087) → 83 → standby → +432 V plates (two 20 µF)
→ choke (14684) → +430 V screens → 4.7 kΩ → +385 V PI (20 µF) → 10k →
+325 V preamp (8 µF). Bias supply: selenium rectifier, 15k/56k, two
8 µF/150 V → −48 V.
The tone network, as the drawings wire it
The published 5F6 schematic and its factory layout sheet agree on a tone-stack
wiring that differs from the textbook redrawing of these same parts, and the
schematic and layout here follow the sheets:
- The 250 pF treble capacitor and the 56 kΩ slope resistor both leave the
cathode-follower output.
- One 0.02 µF capacitor runs from the slope resistor's foot to the node shared
by the treble pot's lower lug and the bass pot — the treble pot's cold end
sits on the far side of that capacitor, not on the slope foot.
- The bass pot is a rheostat: the layout sheet straps its wiper, and the pot
stands in series between that node and the middle pot's top lug.
- The other 0.02 µF runs from the slope foot to the middle pot's wiper, so
the Middle control slides the capacitor's injection point along a 25 kΩ leg
that never leaves the circuit.
- The stack's output is the treble pot's wiper alone; the middle pot's foot
returns to ground through the 5 kΩ presence pot.
The textbook form of this stack ties the treble pot's cold end to the slope
foot, joins the treble and bass wipers at one output node, and hangs the mid
capacitor on top of a rheostat-wired middle pot. The two networks share every
part value and differ audibly at the stops: as drawn, the fixed 25 kΩ leg keeps
the stack from ever going fully silent with Bass and Middle at zero, where the
textbook network's output falls to ground. The tone-stack lab plots this
circuit with the wiring the sheets draw.
The 83, and what a mercury-vapour rectifier does to a supply
The 83 is the only mercury-vapour tube in the corpus, and it behaves nothing
like the vacuum rectifiers around it. A 5Y3 or a 5U4 is a space-charge device:
its forward drop climbs steeply with current, so the B+ falls away on loud
notes — the sag that tweed amps are loved for. The 83 instead strikes an arc in
mercury vapour, and once struck the drop is pinned near 15 V and stays there
from idle to the tube's full 225 mA. Its datasheet says as much in one line: the
tube supplies d-c current at essentially constant voltage in spite of rather
wide variations in output current.
The practical result is a stiff supply. It is also a demanding one, and the
datasheets say so in their ratings rather than in prose. The tube is specified
for vertical, base-down mounting. Its condensed mercury has to sit between
20 °C and 60 °C. And the supply must present at least 50 Ω per plate — more
if the first filter capacitor exceeds 40 µF — to keep the peak charging current
inside the 1 A per-plate limit. The chassis carries a standby switch ahead of
the first filter capacitor, which lets the filament come up before the high
voltage does. Every one of those conditions goes away with the 5F6-A's
indirectly-heated GZ34.
What changed on the way to the 5F6-A
Both drawings print the same rails (+432/+430/+385/+325), the same −48 V bias,
the same transformer set (8087 power, 14684 choke, 45249 output) and the same
preamp values, so the differences are narrow and specific:
- Rectifier. The 83 gives way to the GZ34.
- Presence. On the 5F6 the 5 kΩ presence control sits in the tone stack's
ground leg, below the middle pot, with the 27k feedback resistor landing on
that same junction and a 0.1 µF cap from the presence wiper to ground; the
phase-inverter tail's 10k returns straight to ground. The 5F6-A moves the
presence pot, its cap and the feedback return down to the phase-inverter
tail, and runs the middle pot's foot straight to ground instead.
- Output stage. The 5F6's 1.5 kΩ grid stoppers and 100 Ω screen resistors
become no stoppers and 470 Ω 1 W screen resistors.
- Second-stage cathode. The 5F6 bypasses V2A's 820 Ω cathode resistor with
25 µF; the 5F6-A leaves the same resistor unbypassed. It costs the later amp
gain at the bottom of the band before the tone stack ever sees the signal.
- Output transformer secondary. Both sheets print the same 45249 and the
same 2 Ω secondary for four 8 Ω speakers.
One thing to watch when reading the two parts lists together: the resistor and
pot designators are shared, but the capacitor numbers are not. This amp carries
two capacitors the 5F6-A does not — the 25 µF on V2A's cathode and the 47 pF
across the phase-inverter plates — so the C run diverges from C4 onward and the
same number names a different part on each page. Compare capacitors by role.
Verification — against the printed factory chart
The F-EG drawing prints a full voltage chart, and simulation matches all
fourteen gated nodes: rails within 0.8 %, every tube pin within 9.6 % against
the chart's own ±20 % convention. Two printed values are excluded, and the
reason is arithmetic rather than judgement. The chart gives both phase-inverter
grids (+22 V and +23 V) alongside the tail junction they return to (+32.5 V) —
but each grid reaches that junction through a 1 MΩ leak and nothing else, so
with no grid current both must sit at the junction's own potential. The printed
figures are 68 % and 71 % of it, the fraction a 1 MΩ source reads into a meter
of roughly 2 MΩ input resistance. They are the signature of a moving-coil
voltmeter loading the node, not of a circuit that behaves that way.
The one place the drawing invites a second look is the phase-inverter tail. It
is 10k with a 470 Ω bias resistor, and the chart's own +32.5 V junction figure
confirms it: 32.5 V across 10k is 3.25 mA, and the printed plate drops
(385→235 V through 82k, 385→230 V through 100k) imply about 3.4 mA through the
pair — the same current, inside the chart's own precision.
