Tube substitution & the gain ladder

The order is not simply the µ order. A 5751 has the second-highest µ (70) yet lands below the 12AT7 (µ 60) into 100 kΩ, because the 12AT7's plate resistance is tiny — 10.9 kΩ against the 5751's 58 kΩ — so it keeps a far larger share of its µ. µ tells you the ceiling; r_p tells you how close a given load gets to it. Shrink R_L in the tool above and the low-r_p tubes climb past the high-µ ones entirely — into a light load, transconductance wins.

For everyday purposes the two useful landmarks are the ends of the ladder. The 5751 is the drop-in that trims a 12AX7's gain to about three-quarters (0.72×, −2.9 dB) while sitting at almost the same current — the standard "tame a hot front end" move. The 12AU7 is the other end: about a quarter of the 12AX7's gain (0.26×, −11.8 dB), the choice when a stage needs to run clean with headroom to spare.

In a cathode-biased stage the grid returns to ground and the cathode floats up to V_k = I_a · R_k. A 12AX7 drawing 1.2 mA through the stock 1.5 kΩ cathode resistor sits at 1.2 mA · 1.5 kΩ = 1.8 V — which is essentially the −2 V grid its datasheet anchor assumes. That is why 1.5 kΩ is the value you find under a 12AX7 first stage.

Change the tube and the resistor sets the point automatically: a tube that wants more current develops more voltage across the same 1.5 kΩ, which biases it back down. The 100 kΩ plate resistor helps — off a 300 V node it caps the current near (300 V − V_plate) / 100 kΩ, a couple of milliamps at most, holding the plate around mid-supply (1 mA drops 100 V across it, so V_plate ≈ 200 V). The 5751 lands almost exactly where the 12AX7 did — same current, same 1.8 V or so of bias — the true drop-in. The lower-µ tubes (12AT7, 12AU7) reach a given current at a less negative grid, so they settle at slightly more current and pull the plate down a little, trading a touch of headroom for the far bigger change, which is the gain. No swap in this family needs the bias re-set; only the plate voltage drifts.

6V6 ↔ 6L6 / 5881 — dissipation and bias current. A 6V6GT idles and dissipates at roughly half the level of a 6L6-family tube (12 W plate rating against 23–30 W). In a fixed-bias amp — the 5F6-A, 5F4, or JTM45 — the bias supply must be re-set for the tube type: more negative for a bigger tube, less negative for a smaller one. Drop 6L6s into a bias set for 6V6s and they run cold and thin; drop 6V6s into a bias set for 6L6s and they red-plate. In a cathode-biased amp like the 5E3 the shared cathode resistor self-corrects to a degree, but it was sized for the tube fitted, and a much larger tube will overheat it.

Plate voltage is the hard limit. A 6V6GT is rated to about 315 V on the plate; a Bassman or JTM45 runs its output tubes near 430–450 V, well past that. The big tubes tolerate a small amp's voltage happily, so up-sizing (6V6 → 6L6) is limited by heaters and bias, but down-sizing (6L6 → 6V6) is limited by voltage — a 6V6 in a big-bottle amp is over its plate and screen ratings.

5881 vs 6L6GC. The 5881 is a rugged 6L6 with lower ratings — 23 W and 400 V against the 6L6GC's 30 W and 500 V. Those 5881 limits are the design-center maximums printed on the Tung-Sol 5881 datasheet ↗: 23 W plate dissipation, 400 V maximum plate and screen. A 6L6GC stands in for a 5881 with margin to spare; a 5881 asked to do a 6L6GC's job at high B+ has less. This archive models the 5881 as 6L6GC-class at tweed voltages, where the two are interchangeable.

KT66 in a 6L6 socket. The KT66 is 7AC-compatible (its pin 1 is an internal connection where the American tubes leave pin 1 unused) and slots straight in, but it draws a heavier 1.3 A heater and gives a firmer, louder British voice. The JTM45 is exactly this substitution made permanent — the 6L6/5881 Bassman topology fitted with a pair of KT66s. Comparing those two amplifiers is the clearest illustration of the swap in this corpus.

Heater winding. A pair of 6V6s asks 0.9 A of the 6.3 V heater winding; a pair of 6L6/5881s asks 1.8 A; a pair of KT66s asks 2.6 A. A power transformer wound for 6V6s cannot necessarily feed the bigger tubes — check the winding's rating before up-sizing.

Output-transformer load. The primary impedance is chosen for the tube and its voltage — roughly 8 kΩ plate-to-plate for a 6V6 or KT66 pair, near 4–6.6 kΩ for a 6L6 pair. A family swap detunes that match: the amp still plays, but maximum power and the feel of the damping shift.

Lower drop, higher B+. The GZ34 drops about 17 V, the 5U4 about 50 V, the 5Y3 about 60 V. Swap a 5Y3 for a GZ34 and roughly 40 V of supply comes back; swap a 5U4 for a GZ34 and about 30 V does. (The exact figure scales with how much current the amp draws, since the drop rises with current.) More B+ means more clean headroom and output, a hotter idle everywhere downstream, and bias points that have all moved — worth a re-check on a fixed-bias amp.

Sag is the point, on a tweed amp. Forward drop plus internal resistance means the supply dips under a hard transient and recovers a beat later — the compression and "give" those circuits are prized for. The 5Y3 sags most (it is also the softest and lowest-current, 125 mA), the 5U4 sits in the middle with a stiffer 21 Ω source impedance, and the GZ34 sags least — tight and solid, the closest of the three to a silicon rectifier, which drops almost nothing and does not sag at all.

Warm-up. The directly-heated 5Y3 and 5U4 have the filament for a cathode, so they conduct almost as soon as the heaters glow. The GZ34 is indirectly heated and comes up slowly, delaying B+ until the rest of the tubes are ready — gentler on the filter caps, and a genuine reason to prefer it.

Filament current — check the 5 V winding. The rectifier is fed by the transformer's 5 V winding, and the three tubes load it very differently: 5Y3 at 2 A, 5U4 at 3 A, GZ34 at 1.9 A. Moving up from a 5Y3 to a 5U4 adds a full amp to that winding — a transformer wound for a 5Y3 may not stand it. The GZ34 draws the least of the three, so it never overloads the winding on that count.

Socket wiring. The 5Y3 and 5U4 share the 5T base exactly (filament = cathode, B+ taken from the filament pins, plates on 4 and 6), so they interchange directly. The GZ34 uses the 5DA base: its cathode is separate, brought out on pin 8, with the heater on pins 2 and 8. Where an amp takes B+ from pin 8 the GZ34 drops straight in; where an amp bridges both filament pins to B+ — as some directly-heated designs do — the socket needs a small rewire (lift pin 2 off B+) so the 5 V winding is not compromised. Confirm the socket before assuming a drop-in.