Tube Amplifier Power Transformer Calculator: B+, Heaters and VA
Introduction
The power transformer is the part that decides whether a tube amplifier behaves as intended or does whatever it wants. Missing the high-voltage secondary by twenty volts can be corrected; getting the VA rating wrong, or forgetting that the heater load can weigh as much as the B+ supply, means rewinding or replacing it. And that is expensive, because it is almost always a custom part.
This calculator starts from what you know — which tubes you are using and what plate voltage you want — and works out what you need to order or wind the transformer: secondary voltage, currents, wire cross-sections and total VA. It also does something that classic tube manuals usually do not: instead of multiplying by 1.4 or 0.9, it simulates the rectifier, because those two rules ignore rectifier-tube voltage drop and transformer resistance, which at these currents can account for tens of volts.
This page is written for class-AB push-pull output stages, which cover the great majority of DIY tube amplifiers. Single-ended class A draws nearly constant current and must be sized by different criteria; it will have its own calculator.
How this calculator works
The calculation engine is the same as in the rectifier and smoothing calculator: one mains period divided into four thousand steps, solved with an implicit method, with the steady-state condition found directly instead of waiting for the transient to settle. Here, however, the problem is reversed: instead of starting from the secondary voltage to obtain the DC output, the calculator starts from the DC voltage you want and searches for the secondary voltage that produces it.
Mullard provides a useful validation point for the GZ34: a table of DC output voltage versus AC input voltage, load current and limiting resistance. Feeding the same conditions into the simulator gives 387 V for 2×350 V at 250 mA with 60 µF, compared with Mullard’s published 380 V: about a two-percent difference from a 1958 manufacturer data point.
The GZ34 voltage drop that nobody publishes
There is a difference in documentation style between manufacturers that makes this number difficult to find. American manufacturers — RCA, Sylvania and GE — often publish it directly: the 5V4G drops 25 V at 175 mA, while the 5U4G drops 44 V at 225 mA. European manufacturers such as Mullard and Philips often do not publish a single drop value at all; instead they give output voltage as a function of input voltage, so the rectifier drop is embedded in the published result.
Because the simulator reproduces that table, the drop can be inferred in reverse: it is the value that makes the simulation match each row. The resulting values are 20 V at 250 mA, 14 V at 200 mA and 8 V at 160 mA, which is consistent with a voltage drop that rises with current. In the calculator it is explicitly marked as derived rather than published, because that is exactly what it is.
The model is resistive: leakage inductance is not included, the rectifier tube is represented by a voltage drop rather than a full nonlinear internal-resistance curve, and temperature is not modeled. As usual, treat the current peaks as a worst-case estimate and the voltages as a good engineering approximation.
Reference parameters
On a 50 Hz mains supply, a tube transformer designed for 60 Hz operates at higher flux density and runs hotter. Keep this in mind when reusing an American transformer.
Card 1 — Tube complement
Start by counting the tubes, because the heaters are one of the largest loads and the easiest part to forget. An amplifier with four EL34s and three ECC83s requires about six amperes at 6.3 V, nearly forty VA before the B+ supply is even considered.
A tube rectifier requires its heater on a separate winding rather than sharing the heater supply with the other tubes, because its cathode sits at the high-voltage potential and the heater-to-cathode insulation cannot safely withstand that difference if the winding is shared.
The 12-series dual triodes have center-tapped heaters and can be powered at 6.3 V with the two halves in parallel or at 12.6 V in series. The second arrangement halves the current and is often preferable with long wiring runs, but not every tube has both ratings explicitly documented. Where I could not verify the value from a primary source, the calculator says so instead of inferring it.
Card 2 — Plate current
The operating conditions are taken from published tube data, not invented: for each pair you get the plate voltage, idle current and full-power current. The two current values serve different purposes — the transformer is sized for the maximum, while regulation is evaluated from the difference.
The ratio between the two is one of the most useful numbers in this card. With cathode bias it is typically between 1.1 and 1.4, so current changes relatively little with signal. With fixed bias in class AB it can reach three or even four times the idle value. The tube may be the same, but the power supply sees two very different operating conditions, which helps explain why amplifiers using the same output tube can behave so differently on sustained notes.
Some tubes have more than one tabulated operating condition because manufacturers published different editions of the data sheet over the years, with different plate voltages and loads. This is not an error: they are all documented operating points and are worth comparing.
Card 3 — From the desired B+ to the secondary voltage
This is the card that answers the main question behind the page: what secondary voltage is required to obtain the B+ you want. It does not multiply by a fixed factor; it searches. The simulator tries secondary voltages until the idle DC output matches the requested value, including rectifier drop and series resistance.
Total series resistance is the sum of the secondary resistance, the primary resistance referred to the secondary, and any limiting resistance specified by the rectifier-tube manufacturer. This is not a minor detail: in a 250 mA supply it can account for tens of volts, and it strongly affects how much the supply sags under signal.
Choosing a choke-input filter changes the picture completely: the DC voltage moves toward about 0.9 times the RMS secondary voltage instead of approaching the peak, while the secondary current becomes almost continuous instead of strongly pulsed. For the same delivered power the required transformer can be smaller, and ripple drops sharply.
Critical inductance is the value below which choke current becomes discontinuous and the filter starts behaving more like a capacitor-input supply. It depends on load, so a supply that remains above the critical value at full power can fall below it at idle. This is one reason for using a permanent bleeder resistor.
Card 4 — Regulation and sag
The voltage drop — or sag — is the difference between the plate supply voltage at idle and at full power. It is not inherently a defect: in a guitar amplifier it is part of the sound, the natural compression heard when a note is struck hard. In a hi-fi power amplifier it is instead a loss of regulation and control, reduced by using a lower-resistance secondary and more filtering capacitance.
What surprises many people the first time they calculate it is how strongly the bias method matters. With fixed bias, current can triple between idle and full power and the supply voltage follows it downward; with cathode bias the current may change by only about a quarter and the supply remains much steadier. The transformer is the same — only the demand placed on it changes.
This calculation is purely resistive and does not model the time-dependent recovery of the power supply. Real sag also has a time constant set by the total capacitance; that is what is perceived as the supply breathing after a transient. That dynamic behavior is not modeled by this calculator.
Card 5 — Heater windings
It is worth being generous with heater windings. They are usually among the outermost and hottest windings, they carry several amperes, and a half-volt drop on a 6.3 V heater supply is not negligible: tube heaters have a relatively tight voltage tolerance, and undervoltage reduces emission and can shorten cathode life.
For this reason the default current density is lower than the value typically used for the high-voltage winding. The result should still be checked with the tubes hot, measuring at the tube pins rather than at no load; that is where you find out whether the chosen wire size was sufficient.
AC heaters: twist them, and sometimes elevate them
The two heater wires should be twisted together and routed along the corners of the chassis rather than through the middle. When twisted, the magnetic fields from the equal and opposite currents largely cancel; when separated, the pair forms a loop that radiates at 50 Hz exactly where the smallest signals are present.
If hum remains after the wiring has been done properly, the next step is often to elevate the heater supply by a few tens of volts DC above ground, using a divider from the B+ supply. This serves two purposes: it keeps the heater-to-cathode voltage within the tube rating when the cathode sits at a high potential, as in cathode followers, and it reduces leakage current that can become hum. Another option is a roughly 100 Ω hum-balance potentiometer across the heater winding with the wiper grounded, adjusted for minimum hum.
Card 6 — Negative bias supply
Fixed-bias output stages require a negative grid supply, and there are two common ways to obtain it. A dedicated winding is clean and independent: it adds another winding, which costs very little in VA but must be specified when the transformer is ordered because adding it later means rewinding the transformer.
A bias feed taken from the high-voltage winding costs no extra winding: it is taken from one end of the secondary through a diode of opposite polarity and then smoothed. It has one important consequence, however: the negative bias voltage follows changes in the high-voltage supply, so when the B+ sags under signal the bias shifts with it and the output tubes conduct a little more. In a guitar amplifier this can be part of the intended behavior; in a hi-fi amplifier it is an argument in favor of a separate bias winding.
Whichever scheme is used, the bias circuit is what keeps the output tubes under control: if it fails open, the grids can move toward zero volts and the tubes may conduct excessively. A fail-safe resistor toward the negative supply and periodic checks of idle current are worth more than any calculation.
Card 7 — Total VA and transformer size
This is where everything is added together, and the result often surprises people: the heaters can account for as much VA as the high-voltage supply, sometimes more. An amplifier with four EL34s can use around forty VA just to heat the tubes before producing a single watt of audio.
The high-voltage winding must be rated from the RMS current that actually flows in it, not from the DC current delivered to the load. With a capacitor-input filter the form factor is typically around 1.7; with a choke-input filter it falls close to one. That is the technical reason why a choke, which can look like a luxury, can reduce the required transformer VA.
The margin covers what the calculation cannot know: mains tolerance, operation inside a hot chassis, and the common habit of adding another tube later. Twenty percent is conservative without being excessive.
Card 8 — Where the hum comes from
This card is diagnostic rather than numerical. The most common hum components in a tube amplifier are 50 Hz and 100 Hz, and distinguishing between them immediately cuts the troubleshooting problem in half before the chassis is even opened.
One hundred hertz is twice the mains frequency and is the characteristic ripple frequency of a full-wave rectified B+ supply. Fifty hertz, by contrast, is usually not produced by the full-wave B+ rectifier, so it points instead to direct coupling: magnetic field from the mains transformer, AC heater coupling, or a ground path carrying return current.
An oscilloscope makes the distinction straightforward, although with experience it can also be heard: 100 Hz hum is fuller and steadier, while 50 Hz is lower and rougher. If the noise is neither, rule out oscillation before blaming the power supply.
Construction: placement, shielding and grounding
A well-designed power transformer mounted badly can cause more trouble than a mediocre transformer mounted well. The reason is magnetic leakage at mains frequency, in the same chassis as an output transformer whose job is to couple magnetic flux. The following rules are not superstition; they are geometry.
Distance and orientation: why 90 degrees can matter more than centimeters
At a distance, transformer leakage field behaves approximately like a dipole field, so it falls roughly with the cube of distance: doubling the spacing can reduce coupling by about a factor of eight. That is a strong improvement, but chassis space is limited and available centimeters disappear quickly.
Orientation costs nothing and can be even more RMSective. What matters is not simply the magnetic field that reaches the output transformer, but the flux coupled into its winding, which depends on the cosine of the angle between the winding axes. At 90 degrees that term is close to zero: rotating one transformer by a quarter turn can help more than adding ten centimeters of distance. If the chassis layout allows it, test the orientation before drilling by powering the amplifier with the output transformer temporarily positioned and rotating it while listening for the minimum hum.
Shielding: when it helps and when it is a placebo
Magnetic shielding works mainly by redirecting flux rather than absorbing it: a high-permeability material provides an easier path for the magnetic field. That means the shield must be ferromagnetic and reasonably continuous. An aluminum sheet does essentially nothing against a 50 Hz magnetic field regardless of thickness; aluminum is useful for electric-field and radio-frequency shielding, not low-frequency magnetic shielding.
A mild-steel cover around the mains transformer can help, and difficult cases may call for mu-metal. Mu-metal must be handled carefully because bending or machining it after its final heat treatment can seriously degrade its magnetic properties. Shield the source rather than the victim: redirecting the field where it is concentrated is much more RMSective than trying to stop it after it has spread through the chassis.
Grounding: the return path is an antenna
Fifty-hertz hum that does not change when the transformers are rotated often comes from grounding. The filter-capacitor charging currents are pulses of several amperes, and if their return current flows through the same section of chassis used as the reference for the input stage, that metal becomes a voltage source in series with the signal.
The practical rule is to keep the two current domains separate: return the power-supply charging current directly to the filter-capacitor reference point, then make a single connection from there to signal ground. Perfect-star-ground theories matter less than ensuring that pulsed charging current never shares a conductor segment with the input-stage reference.
Mechanical hum, the one you can hear directly
If the transformer itself hums mechanically, the problem is not in the audio signal path. Three common causes are loose laminations, core saturation because the mains voltage is higher than the transformer was designed for, or a DC component on the mains that shifts the operating point on the hysteresis loop.
The first two are easy to diagnose by measuring mains voltage and no-load current. The third is more subtle and can be caused by equipment using asymmetric or half-wave loads on the same supply line. Impregnation or tighter clamping can reduce audible noise but does not remove the underlying cause; a saturating transformer will still run hot.
Reference tables and sources
The data are not copied from other calculators: they come from tube data sheets, and where I could not verify a value from a primary source the calculator says so and leaves the value for you to enter rather than filling it with a plausible-looking number.
Output-tube heaters
| Tube | Current | Alternative value | Source |
|---|
Tube rectifiers
| Tube | Heater | I max | Voltage drop | Source |
|---|
The plate-current data for push-pull pairs come from Mullard, Philips-licensed Ei-RC, Sylvania and Genalex Gold Lion data sheets and, where explicitly indicated, from modern manufacturers such as Svetlana and JJ. Several values remain unverified from a primary source — including the 5U4GB voltage drop, the 5Y3GT maximum current and the 12.6 V heater connection of some dual triodes — and in those cases the calculator asks for the value instead of inventing one.