Loudspeaker impedance, reflection and return loss
The number on the back of a cabinet is a convention, not a measurement. Here is what the driver actually presents to the amplifier: the reflection coefficient it creates, the power it really absorbs, and the tolerance test that decides whether a driver is within spec.
Impedance, reflection and return loss calculator
Enter the impedance as a magnitude and a phase, or leave the phase at zero for a purely resistive driver. The line impedance is the reference the reflection is measured against — for a direct connection it is the amplifier's output impedance.
The reflection calculation assumes a lossless line with the source matched to it, which is what a passive impedance analyser measures. Power is dissipated in the resistive part alone, so a driver with a large phase angle absorbs less than √(√V²/R) suggests. The last row compares the delivered power against the same source into a perfectly matched load: a lower impedance draws more power, a higher impedance less, and only the reactive case loses to the naive figure.
Three numbers that are not the same thing
A cabinet carries a number on its back, and it is almost never a measurement of anything. The nominal impedance is a single figure chosen to sit within about 20% of the lowest point on the magnitude curve, for historical reasons that have more to do with amplifier design in the 1950s than with physics. The minimum impedance is that lowest point, measured across the whole band, and it is the figure that decides how hard the amplifier has to work. The DC resistance is what a multimeter reads, and it is always lower than either.
Those three diverge because a loudspeaker is not a resistor. The voice coil is an inductor with a resistance in series, and the mechanical suspension and the acoustic load add a second, frequency-dependent term on top. The impedance magnitude rises with frequency because of the coil inductance, dips sharply at the resonant region where the mechanical terms cancel it, and settles towards the DCR at the top.
| Quantity | What it is | Why it matters |
|---|---|---|
| Nominal | The printed number, e.g. 8 Ω | Nothing technical — amplifier pairing convention |
| Rated minimum | Lowest point on |Z| − 20%, per IEC 60268-5 | The tolerance test the manufacturer actually runs |
| True minimum | Lowest point on |Z| | Peak current, and therefore amplifier sizing |
| DCR | DC resistance of the voice coil | Contradictions, open driver versus voice-coil short |
Reflection: the mismatch made visible
Put a load ZL on a line of characteristic impedance Z0 and part of the wave comes back. The reflection coefficient Γ is that returned fraction, as a complex number:
Γ = (ZL − Z0) / (ZL + Z0)
Its magnitude runs from 0 to 1 for any passive load. Zero means the load is matched and nothing returns; one means everything returns, which is an open circuit. The sign of the real part tells you which way the mismatch goes: a load heavier than the line gives a positive Γ and reflects in phase, a lighter one negative and reflects in antiphase.
For a purely resistive pair this collapses to the familiar ratio, and the symmetry is worth noticing. A 50 Ω line into 25 Ω and the same line into 100 Ω both give |Γ| = 0.333 and the same standing wave ratio of 2.000. They are different faults with the same voltage cost, which is why SWR alone cannot tell you which one you have.
| Load on line | Γ | |Γ| | SWR | Return loss |
|---|---|---|---|---|
| 8 Ω on 8 Ω | 0 | 0 | 1.000 | −∞ |
| 4 Ω on 8 Ω | −0.333 | 0.333 | 2.000 | −9.54 dB |
| 8 Ω on 4 Ω | +0.333 | 0.333 | 2.000 | −9.54 dB |
| 2 Ω on 8 Ω | −0.600 | 0.600 | 4.000 | −4.44 dB |
| 16 Ω on 8 Ω | +0.333 | 0.333 | 2.000 | −9.54 dB |
| open circuit | +1 | 1 | ∞ | 0 dB |
| short circuit | −1 | 1 | 1.000 | 0 dB |
Return loss is the same information in decibels, and its sign convention trips people up. It is the reduction in delivered voltage relative to a matched load, so it is zero when matched and negative otherwise. That means you add it to the source level, never subtract it. The −9.54 dB on a 4 Ω load is not noise, it is signal you paid for and did not use.
Where the power actually goes
The mistake that costs the most money is assuming the amplifier delivers V2 / ZL into whatever it is connected to. Two corrections apply, and the first one is the reason a 4 Ω speaker does not simply get twice the power.
The amplifier is a voltage source with a non-zero output impedance. Across the load the voltage divides between the two, so with a 24 V amplifier of 0.05 Ω output impedance driving 8 Ω, the load sees 24 × 8/8.05 = 23.85 V and dissipates 71 W, not 72 W. Into 4 Ω the same amplifier delivers 23.4 V and 137 W — 87% more power, which is exactly what the −1.5 dB return loss predicts.
The second correction matters when the load is reactive. Power is V2 × Re(Z) / |Z|2, not V2 / |Z|, because the reactive part stores energy every cycle and returns it. A driver at 6 Ω with a 40° phase is |Z| = 7.21 Ω but only 6 Ω of that is real, so it absorbs 17% less than a naive calculation suggests. The tool above reports the resistive part separately for exactly this reason.
The DCR tolerance test
The cheapest useful measurement you can make is DC resistance across the terminals with the speaker disconnected, and it catches most failures immediately. Because the coil inductance is bypassed at DC, a healthy 8 Ω driver reads between 6.0 and 7.2 Ω — the band IEC 60268-5 sets at 75% to 90% of the rated figure.
The two failure modes sit on opposite sides of that band. A reading below the minimum is a shorted voice coil, and the driver is dead. A reading above the maximum is either an open coil or, more often, a broken lead or a failed solder joint. Either way the driver is not reproducing, and no measurement of its frequency response is worth taking until the DC reading is plausible.
Five mistakes that survive in production
- Pairing by nominal impedance. Two 8 Ω cabinets can have minima of 4.2 and 11 Ω. In series that is 15.2 Ω of variation across the pair, and the crossover network designed for the nominal figure will not flatten the combined response.
- Reading SWR as a fault indicator. SWR of 2 is common in a healthy multi-driver cabinet, because crossover components and the enclosure itself sit on the line. SWR describes a load; it does not diagnose a fault.
- Computing power as V²/Z for a complex load. This overstates dissipation by up to 20% in the midband and is the usual explanation for a driver that runs hotter than predicted.
- Adding return loss instead of subtracting it. It is defined as a reduction. Applied the other way it inflates the predicted level by 9.5 dB for a doubled or halved load.
- Testing at one frequency. Impedance is a curve. A driver can present 5 Ω at 40 Hz while reading 40 Ω at 2 kHz, and only the sweep shows which of those the amplifier has to survive.
Formulas worth keeping
| Quantity | Expression |
|---|---|
| Reflection coefficient | Γ = (ZL − Z0) / (ZL + Z0) |
| Standing wave ratio | SWR = (1 + |Γ|) / (1 − |Γ|) |
| Return loss | −20 · log10 |Γ| |
| Load voltage | VL = Vs · ZL / (Zs + ZL) |
| Power into the load | P = VL2 · Re(ZL) / |ZL|2 |
| DCR tolerance band | 0.75 × Znominal to 0.90 × Znominal |
Summary
Treat the number on the cabinet as a label. Measure the curve, find its minimum, and size the amplifier for that. Compute the reflection coefficient to know what the mismatch costs you, remember that return loss is negative, and check the DC resistance before you trust anything else. A 2 Ω driver on an 8 Ω amplifier is a -4.44 dB return loss and nearly twice the current, which is the combination that activates protection.