Power load lines

A power tube output stage drawn on the tube’s plate curves. Pick single-ended or push-pull, set up the screen supply, bias, driver and output transformer, and load it with a resistor or a speaker. Then drive it with a sine and read the power, the distortion and how hot each tube runs.

Put it in a power amp

A beam power tube, the usual output tube of the larger Fender amps and many others, in pairs or fours.

Swipe sideways for the whole drawing. Tap a part to jump to its value. The red numbers are DC voltages to ground at idle.

Plate curves

At the speaker, one cycle

outputthe fundamental alone

Currents

Harmonics at the speaker

evenodd dB below the fundamental, 2nd to 10th

All values

Reading the chart

One tube. Each gray curve is the plate current at one grid voltage as the plate voltage changes. With pentode or tetrode connection the screen voltage stays fixed. With ultra-linear or triode it moves with the plate. A single-ended stage idles at the red dot and swings along the load line: the reflected speaker load plus the primary’s resistance. The steep red line is the DC load line through the primary (and the supply resistance, if the supply sags). In push-pull, each tube runs along the class A line (half the plate-to-plate load) while both conduct, then the steeper class B line (a quarter of the load) once the other cuts off, and sits at zero for the rest of the cycle.

Both tubes. In push-pull the two grids are driven opposite ways: when tube A’s grid comes up a volt, tube B’s goes down a volt, and the plates swing opposite ways. The composite curves are tube A’s current minus tube B’s at every point of that swing, with a perfect center tap. The pair runs along one straight line through the idle plate voltage at zero current, sloped at a quarter of the plate-to-plate load. Where the curves bunch up or spread out along it, the stage distorts. A kink near the middle at cold bias is crossover distortion.

Speaker loads. A speaker’s current and voltage don’t peak together, so the path opens into a loop. Near resonance the load is several times the rated impedance and the loop is long and flat. Up high, the voice coil’s inductance tips it the other way. With a speaker, the distortion figures and Find full power go by the speaker’s current, which pushes the cone. The voltage reads higher, since the impedance is higher at the harmonics than at the note. Both are shown. For your own speaker, pick Measured curve from a file in the Load menu: a DATS .zma file, or text with frequency, ohms and phase on each line (example file).

Warnings and notes. Red marks what wears a tube out fast or makes the numbers doubtful: plate or screen dissipation over the rating (at idle or averaged with the signal), grid current pushing the bias around, and the solver not settling. A dotted underline and a plain note mark what’s past the datasheet but common in guitar amps: plate or screen voltage over the rating, and more grid-to-cathode resistance than the sheet allows. Ratings are the datasheet’s. Later sheets list higher ones for several types, and you can type those in.

Feedback. Negative feedback sends part of the speaker voltage back to the driver, as most guitar amps feed the secondary back to the phase inverter. The driver here is ideal, so you see what feedback does to the output stage alone: less distortion, a steadier output as the speaker’s impedance rises, and at cold bias what’s left crowds into the crossover region. A real inverter clips and runs out of bandwidth, so feedback does less in a real amp.

Full power. With Clipping picked, Find full power turns the signal up until the stage runs out of room and shows the power just short of that. There are three limits. The grids reach 0 V and start pulling current (with a DC-coupled driver, which can push them past 0 V, the limit is where grid current loads the driver down). The plates come down into the knee, which we put where the plate resistance has dropped to match the load, leaving the grid half its usual pull on the current. A single-ended tube heads into cutoff on the low half of the swing, which we put where its gain there has dropped by half. The first one reached sets full power. Pick a THD figure instead and the signal goes up until the distortion reaches it. With no feedback, the EL34, KT88 and KT66 sit at 5 to 7 % over a wide range and pass 5 % long before they clip, one reason amps are rated with feedback on. It takes a few seconds, and the signal is left at the level found.

Supply sag. A real B+ drops as the stage pulls more current, so the page starts with a sagging supply: a power transformer sized for the tubes, with the bias off a tap on its HT winding, as in a blackface Fender. B+ comes from a source behind the supply resistance, with the reservoir cap at B+, set so idle B+, bias and idle figures stay where you put them. Only the signal’s effect changes. A class AB push-pull stage pulls more current as it’s driven harder, so B+ drops, the screens drop further, and full power comes down. A class A stage pulls about the same current either way, so B+ barely moves, though the signal current still runs through the reservoir cap. Only the average is modeled, not the 100 or 120 Hz ripple. The supply resistance stands for the windings and rectifier together. To measure it, compare B+ at idle and at full power into a load.

Core saturation. The output transformer’s iron holds only so much flux. Flux builds over each half cycle of the primary voltage, so a lower note or a bigger swing means more. Past the knee the primary’s inductance collapses: the primary pulls a burst of current on each peak, the plates see much less load, and a low note at full power thins out and distorts. A standing current (single-ended, or an unmatched push-pull pair) starts the core part way up, closer to the knee on one side. The knee comes from the primary’s turns, the core’s area and the steel’s knee flux density, the figures a transformer winder uses. The starting values are placeholders, not a real transformer. The readouts show the peak flux against the knee and the lowest note at this swing before the core gets there. Hysteresis and losses are left out, and the inductance holds at the primary inductance up to the knee.

The 6L6GC

A beam power tube, the usual output tube of the larger Fender amps and many others, in pairs or fours.

Curves. The beam law from the bench, fitted to the Tung-Sol sheet (1960): plate curves at 250 and 400 V screen, its three screen-current charts and its two grid-current charts; GE’s 400 V and RCA’s 300 V plate curves (1959).

Expected error. Plate current within about 3 to 4 mA rms of Tung-Sol’s curves, 5 mA of GE’s and 4.5 mA of RCA’s (7 mA near the knee), screen current about 2 to 3 mA, grid current within about 3 mA at 400 V screen (up to 11 mA low at 250 V screen, and low where the plate swings under about 100 V). The Tung-Sol class A1 table reads 6 to 9 % above the curves on the same sheet, and the fit is to the curves; JJ’s current sheet reads about 4 % below it. Real tubes of one type vary by 20 % or more, so the tube in your amp can be further off than this. The sheet’s curves cover screen voltages from 50 to 400 V. Past that, the figures are extrapolated, and you get a note about it.

Ratings used. Plate 500 V, screen 450 V, plate dissipation 30 W, screen dissipation 5 W, grid circuit 100 kΩ with fixed bias, 500 kΩ with cathode bias (Tung-Sol, design maximum). You can change the voltage and dissipation ratings under Bias and ratings, for a later sheet or a different make.

Starting stages. The push-pull stage starts at 450 V B+ into 4 kΩ plate to plate, with fixed bias and the idle at 70 % of the plate rating. The single-ended stage starts at 350 V into 4.2 kΩ, with cathode bias and the idle at 90 %. Both stages use 470 Ω screen stoppers, 100 nF coupling caps and 1.5 kΩ grid stoppers, as in many Fender amps. These are common setups for the type at typical guitar-amp voltages, not copies of any one amp. Where they run past the sheet’s voltage or grid-circuit ratings, as guitar amps often do, you get a note about it.

How this tool works. The curves use the circuit bench’s tube models, fitted to each type’s datasheet charts and published operating points. We solve the stage with the bench’s engine, every part in place: the tubes with interelectrode capacitance and grid current, the screen supply, the bias network and the driver’s output resistance. The output transformer is coupled windings with their resistance, magnetizing and leakage inductance, and winding capacitance. The sine runs at least 0.3 s so the caps settle, and the last few cycles are drawn. The speaker loads use the same parts as the reactive load tool: a generic open-back or closed-back shape, or your own impedance measurement (a DATS .zma file, or any text file of frequency, ohms and phase) fitted in your browser. The file isn’t uploaded anywhere. Ripple, heater and hum effects, and noise aren’t modeled. The power tubes’ grid current follows a typical law not fitted to any sheet, so class AB2 figures are rough.

Spot an error, or want a circuit added? Email info@swellamps.com. A schematic or a clear photo of one helps. Anything you send may be used to improve these models and our desktop app.

© 2026 Swell Amplification LLC. All rights reserved. The simulator and tools, their circuit models, drawings, presets and text are the property of Swell Amplification LLC and may not be copied, reproduced, redistributed or used to build another product without written permission. The original circuits are historical designs documented in public sources, and no claim is made to them. Everything here is provided as is, without warranty of any kind, for study and personal use. Results are simulations of typical parts, not measurements of any real unit. Don’t rely on them for the safety of any build or repair. Tube amplifiers carry lethal voltages.

Product and company names are trademarks of their owners. They are used here only to identify the circuits these models are based on; this site is not affiliated with or endorsed by any of them.

From Swell Amplification, swellamps.com.