How to Bias a Fixed‑Bias Push‑Pull Tube Power Amp

Every tech has a bias method they trust, and most of the arguments about which one is best are really arguments about which one you happen to have got away with so far.

Mine is the output-transformer resistance method, and I want to make the case for it properly — not because it is clever, but because it is the one that gives you the number you actually want, without asking you to put a multimeter in series with 450 volts.

That last point is not a small one. The traditional shunt method works. It also asks you to break a plate lead on a live amp and carry the full plate current through your meter and your probes. I have seen what happens when somebody does that with the meter still set to volts, and I would rather write this article than have that conversation again.

The Short Version

• Measure the DC resistance of each half of the output transformer primary, centre tap to plate lead, with the amp off and the caps discharged.
• Measure the DC voltage drop across that same half, centre tap to plate lead, with the amp on and idling.
• Ohm's law gives you the current. Divide by the number of tubes on that side. Multiply by plate voltage for dissipation.
• The current you get is true plate current — the screens are fed from a different node and their current never passes through the primary. This is the method's biggest advantage and it is the thing most write-ups get backwards.
• Aim for roughly 60–70 % of the tube's maximum plate dissipation at idle in a fixed-bias Class AB guitar amp.
• Your meter never goes into the high-voltage current path. Volts mode when the amp is live, ohms mode only when it is dead.

Careful — Read This First

Tube amplifiers run at lethal voltages. Plate and B+ rails sit anywhere from about 300 V in a small amp to 550 V and beyond in a big one, and the filter capacitors hold that charge after you unplug it. This is not a "be careful of shocks" situation. It is a "this can stop your heart" situation.

• Always unplug the amp before opening it. Switching it off is not the same thing.
• Discharge the filter capacitors through a suitable bleeder resistor, and then verify with your meter that B+ has fallen to a safe voltage. Do not trust that a bleeder did its job.
• Keep one hand behind your back when probing a live circuit. The reason is specific: it stops a fault current from crossing your chest.
• Use insulated probes, and clip-on grabbers wherever you can. Fit the clip with the amp off, then power up. Fingers should not be near high voltage while it is live.
• Always have a speaker or a proper dummy load connected when the amp is on. Running a tube amp open-circuit can arc the output transformer and destroy it.

If reading that list made you uncomfortable rather than bored, that is useful information. Take the amp to somebody who does this every day.

What You Need

Almost nothing, which is the other reason I like this method.

• A digital multimeter that reads resistance and DC volts to at least 600 V — check the CAT rating and the probe rating, not just the display range.
• Insulated clip leads or grabber probes.
• An insulated screwdriver for the bias trimmer.
• The datasheet for your power tubes, or at minimum the maximum plate dissipation figure.

A meter with a relative / REL / zero function is worth seeking out here. You will be measuring resistances in the tens of ohms, and a typical pair of test leads contributes 0.2–0.5 Ω of their own. On a 70 Ω winding that is up to 0.7 % of error before you have measured anything. Short the probes together, hit REL, and it disappears.

Step 1 — Measure the OT Primary Resistance

Amp off. Unplugged. Caps discharged and verified.

The primary of a push-pull output transformer has three connections that matter: a centre tap (CT) fed from the high-voltage supply, and two ends going to the plates of the "left" and "right" tube groups. Because the primary is just a long piece of copper wire, each half has a small but perfectly measurable DC resistance.

1. Identify the primary wires. The CT normally runs from the transformer to the first B+ filter capacitor. The two plate leads run to the power tube plates — pin 3 on an octal socket, pin 7 on a nine-pin EL84.
2. Set the meter to ohms, short the probes, and zero the leads.
3. Measure CT to plate lead A. Call it $R_A$.
4. Measure CT to plate lead B. Call it $R_B$.

A small difference between the two is completely normal, and it is not a fault. One half is wound closer to the core than the other, so it uses a slightly shorter length of wire. Expect a few percent.

For a typical 50 W amp you might see:

$$R_A = 70\ \Omega \qquad R_B = 72\ \Omega$$

Write both down. The whole method rests on these two numbers, so if the meter gives you an unstable reading, chase that down now rather than later.

Step 2 — Measure the Voltage Drops

Now the amp comes on.

1. Connect a speaker or a dummy load.
2. Make sure every power tube is in its socket.
3. Clip your black probe onto the OT centre tap before you power up.
4. Power on and let it idle for at least five minutes. Idle current drifts while everything comes up to temperature, and biasing a cold amp gives you a hot amp twenty minutes later.
5. Set the meter to DC volts on a range that covers your B+.

Then measure directly across each half of the primary:

• Black probe on the CT, red probe on plate lead A → this is $V_{\text{dropA}}$.
• Black probe stays on the CT, red probe moves to plate lead B → this is $V_{\text{dropB}}$.

Measuring the drop directly matters more than it looks. The obvious alternative is to measure CT-to-ground and plate-to-ground and subtract, but you are then subtracting two numbers of around 450 V to find a difference of around 5 V. Every bit of meter error in both readings lands in that difference, and B+ moves slightly as you adjust the bias anyway. Measure across the winding and the problem disappears.

While you are in there, note the plate voltages to ground, $V_{PA}$ and $V_{PB}$. You need them for dissipation in a moment.

Step 3 — Calculate the Current in Each Half

Ohm's law, nothing more:

$$I_A = \frac{V_{\text{dropA}}}{R_A} \qquad I_B = \frac{V_{\text{dropB}}}{R_B}$$

Worked through with the numbers from above, and drops of 5 V and 4 V:

$$I_A = \frac{5\ \text{V}}{70\ \Omega} \approx 71\ \text{mA}$$ $$I_B = \frac{4\ \text{V}}{72\ \Omega} \approx 56\ \text{mA}$$

$I_A$ is the total current drawn by every tube on side A; $I_B$ is the same for side B.

Why this reads true plate current

This is the part I want to make a fuss about, because most descriptions of this method — and an earlier version of this very article — hedge it with "ignoring the small screen current," and that hedge is wrong.

In a normal pentode or beam-tetrode output stage, the screen grid is fed from its own supply node, downstream of a dropping resistor from B+. It is a separate branch. Screen current flows out of that node, through the tube, and back to ground via the cathode. It never touches the output transformer primary.

So the current you just calculated through the primary half is plate current, and nothing else. That is exactly the quantity you need for plate dissipation, and it is exactly the quantity the alternatives blur:

• A 1 Ω cathode resistor measures $I_p + I_{g2}$ together, because both return through the cathode. At idle the screen current is small, but "small" is doing real work in that sentence — in some tubes it is a few percent, in others more, and it climbs as the tube ages.
• The shunt method measures plate current too, but at the cost of putting your meter in the high-voltage path.

The output transformer method gets you the clean number and keeps the meter out of harm's way. That is the whole argument.

One exception worth knowing: in an ultralinear output stage, the screens are tapped off the output transformer primary itself. In that topology screen current does flow through part of the winding, and this method needs adjusting. Guitar amps are rarely ultralinear, but hi-fi amps often are — check before you assume.

Step 4 — Divide by the Number of Tubes

How you split that side current depends on how many tubes share each half of the primary.

Two tubes total (typical 50 W). One per side, so the side current is the tube current:

$$I_{\text{tubeA}} = I_A \qquad I_{\text{tubeB}} = I_B$$

Four tubes total (classic 100 W). Two in parallel per side. Assuming they are reasonably matched:

$$I_{\text{tubeA}} \approx \frac{I_A}{2}$$

So an $I_A$ of 120 mA is about 60 mA per tube, and an $I_B$ of 116 mA is about 58 mA per tube.

Six tubes total. Three per side, so divide by three.

That word assuming is carrying weight. This method tells you what a side is drawing, not what an individual tube inside that side is drawing. Two tubes on one side, one pulling 80 mA and one pulling 40 mA, look exactly like two well-matched tubes pulling 60 mA each. If you have any reason to doubt the matching, measure cathode current per tube as a cross-check, or swap tubes around and see whether the imbalance follows the tube or stays with the socket.

Step 5 — Calculate Plate Dissipation

Plate dissipation at idle is plate voltage times plate current:

$$P_{\text{tube}} = V_{\text{plate}} \times I_{\text{tube}}$$

Strictly, $V_{\text{plate}}$ means plate-to-cathode voltage. In a fixed-bias amp the cathode sits at ground, so the plate-to-ground reading you already took is the right number. In a cathode-biased amp it is not — you would have to subtract the cathode voltage, which in a Vox-style circuit can be 10 V or more.

Maximum plate dissipation comes from the datasheet. The common ones:

• EL84 / 6BQ5 — 12 W
• 6V6GT — 14 W
• EL34 — 25 W
• 6L6GC — 30 W
• 6550 — 35 W
• KT88 — 42 W

Choosing a target

For fixed-bias Class AB guitar amps, the working convention is to idle at about 60–70 % of maximum plate dissipation. That range comes out of Randall Aiken's writing on the subject and has been the default in amp tech circles ever since.

$$I_{\text{target}} = \frac{k \cdot P_{\text{max}}}{V_{\text{plate}}}$$

For a 6L6GC at 450 V, targeting 70 %:

$$P_{\text{target}} = 0.7 \times 30\ \text{W} = 21\ \text{W}$$ $$I_{\text{target}} = \frac{21\ \text{W}}{450\ \text{V}} \approx 47\ \text{mA}$$

Rule of thumb: Cooler (50–60 %) buys tube life and headroom. Hotter (toward 70 %) reduces crossover distortion and thickens the midrange. Past 70 % you are mostly buying red plates.

Two things this convention does not cover. It is for fixed bias, Class AB — a cathode-biased amp idles far closer to its maximum by design, and the percentage rule does not transfer. And it assumes the tube is running at or below its rated plate voltage; a 6V6 at 450 V in a Bassman-style circuit is already outside the datasheet before you start.

Step 6 — Adjust, Then Measure Again

Now the loop.

1. Find the bias trimmer. Confirm its range first by measuring the negative grid voltage at the power tube control grids — typically somewhere between −30 V and −60 V for octal tubes, and around −10 to −15 V for EL84s.
2. Move the trimmer a small amount.
3. Re-measure $V_{\text{dropA}}$ and $V_{\text{dropB}}$, and the plate voltages.
4. Recalculate current, then dissipation.
5. Repeat until you land on your target.

Step 3 is the one people skip, and it is why their bias never quite settles. Both terms in the dissipation equation move together. Bias hotter and the extra current pulls the supply down, so B+ sags and plate voltage falls. Bias cooler and B+ rises. If you set current from a fresh reading but calculate dissipation using the plate voltage you wrote down ten minutes ago, you will be off — and always in the direction of running hotter than you think.

When you are happy with the number, let the amp idle for another five or ten minutes and check it has stayed there.

When the Numbers Look Wrong

The example above is a good teaching case precisely because it is a bad result. Look again:

$$I_A \approx 71\ \text{mA} \qquad I_B \approx 56\ \text{mA}$$

That is a 27 % imbalance between the two sides, and it is not something to shrug at. Almost every guitar amp has a single bias adjustment serving all the tubes, so you cannot trim that out — turning the pot moves both sides together and one of them will always be wrong.

A few things it usually means:

• Badly mismatched output tubes. By far the most common. Swap the tubes between sides. If the high reading follows the tube, you have your answer.
• A gassy or failing tube. A tube drawing far more than its partner at the same grid voltage is often on its way out.
• A fault in the bias supply. If one side's grid is sitting at a different voltage from the other's, measure both grids directly. Bias networks with separate feed resistors per side can develop this.
• A leaky coupling capacitor from the phase inverter, putting positive DC on one grid and pushing that tube hot. This one is worth ruling out early, because it will destroy a tube and possibly the output transformer.

An imbalance under about 5 % is normal and not worth chasing. Above 10 %, find out why before you set the bias.

The other reading that should stop you: a voltage drop that will not settle while you watch it, or one that climbs steadily. That is a tube running away, and the correct response is to power down rather than to keep adjusting.

The Whole Thing on One Page

1. Amp off, unplugged, caps discharged and verified. Zero your meter leads.

2. Measure CT-to-plate resistance on each side: $R_A$, $R_B$.

3. Amp on, loaded, warmed up five minutes. Measure CT-to-plate voltage drop on each side: $V_{\text{dropA}}$, $V_{\text{dropB}}$. Note plate voltages.

4. Side currents:

   $$I_A = \frac{V_{\text{dropA}}}{R_A}, \qquad I_B = \frac{V_{\text{dropB}}}{R_B}$$

5. Divide by tubes per side — 1, 2 or 3.

6. Dissipation per tube:

   $$P_{\text{tube}} = V_{\text{plate}} \times I_{\text{tube}}$$

   Target roughly 60–70 % of $P_{\text{max}}$.

7. Adjust the trimmer, then re-measure voltage as well as current, and iterate.

Try It Yourself

Steps 3 through 6 are arithmetic, and arithmetic is exactly the thing that goes wrong at the end of a long day with a live amp open in front of you. That is what I built Bias Bench for: put in your transformer resistances, your voltage drops, your tube type and your tube count, and it does the current, the per-tube split, the dissipation and the percentage of maximum — including the screen-current and supply-sag behaviour that the hand method glosses over.

Use it as a second opinion rather than a replacement. Doing it by hand once is how you learn what a wrong number looks like.

If you want to know what that idle current is doing to the sound rather than to the tube, the related piece is Feedback Lab — bias sets where in the transfer curve the tubes sit, and the feedback loop decides how much the rest of the amp cares. And for what happens when you stop idling and start clipping, that is how amplifier output power is really measured.

Sources

• Tube maximum plate dissipation figures are from the manufacturers' datasheets. Use the datasheet for the tube you actually have; modern production of a classic type does not always match the original ratings.
• The 60–70 % of maximum dissipation convention for fixed-bias Class AB follows Randall Aiken's published work on bias setting, which is where most of the trade got it.
• The claim that screen current does not flow through the output transformer primary holds for conventional pentode and beam-tetrode output stages, where the screens are fed from a separate supply node. It does not hold for ultralinear stages, where the screens tap the primary itself.
• Voltage-drop and resistance figures used in the worked examples are representative of a typical 50 W amp, chosen to make the arithmetic legible. Measure your own; do not calculate from mine.

Hope that was useful. Thanks for reading, and stay safe on the bench.

Marko, Slightly Technical