L-pad calculator
Every L-pad calculator assumes the tweeter is an 8 Ω resistor. It is not: its impedance runs from its DC resistance to a peak at resonance, and the pad's two resistors turn that swing into a level change the pad itself adds. This page sizes the pad, rounds both resistors together, and shows what it does across the tweeter's real impedance.
The defaults are a typical 8 Ω ferrofluid dome: 6 Ω of voice-coil resistance and a resonance peak of 14 Ω. Take Re from the datasheet and the peak from its impedance plot. The amplifier power is into the design impedance; music puts roughly a tenth of it above a 2–3 kHz crossover.
| Tweeter impedance | Crossover sees | Level at tweeter | Versus design |
|---|
Each row is the same pad, built from the catalogue parts above, in front of a different tweeter impedance. The highlighted row is the design impedance.
How this is calculated
An L-pad is two resistors: R1 in series with the tweeter and R2 across it. Written in terms of the voltage ratio A = 10−dB/20 and the impedance Z the pad is designed for, the two values are
R1 = Z·(1 − A) R2 = Z·A / (1 − A)
They are chosen so that R2 in parallel with a load of exactly Z comes to Z·A, and R1 plus that comes back to Z. Two things follow at once. The tweeter gets the fraction A of the voltage, and whatever feeds the pad — normally a crossover's high-pass section — still sees Z, so the filter it was designed around keeps its frequency and shape. For any other tweeter impedance Zt the page evaluates the network directly:
Zin = R1 + (R2 ∥ Zt) level = (R2 ∥ Zt) / Zin
That is the whole calculation, and it is exact for a resistive load. A tweeter's impedance also has phase — it is inductive above its resonance and swings through both signs around it — so treat the figures between Re and the peak as a close guide rather than a measurement.
A tweeter is not 8 Ω
Both formulas hold only when the tweeter is exactly Z, and a real tweeter is exactly Z at perhaps two frequencies. Over most of its passband it sits near its voice-coil resistance Re, typically two-thirds to four-fifths of the nominal figure, and at its resonance it rises to a peak — modest in a ferrofluid-damped dome, several times nominal in a dry one. The pad handles this in two ways, one welcome and one not.
The welcome one: the pad flattens the load. R2 sits across the tweeter and swamps its variation, and R1 adds a fixed amount that does not vary at all. At the defaults the tweeter alone runs from 6 Ω to 14 Ω, a factor of 2.33; through the pad the crossover sees 7.36 Ω to 9.07 Ω, a factor of 1.23. The more attenuation, the flatter it gets, which is why a heavily padded tweeter is kinder to a crossover than a bare one.
The unwelcome one: the pad makes the level depend on the impedance. Fed straight from a voltage, a tweeter's level does not care what its impedance is. Behind a pad it does, because the fraction of the voltage that reaches it is (R2 ∥ Zt) / Zin, and that rises with Zt. At the defaults the tweeter gets −6.55 dB where it sits at Re and −4.88 dB at its resonance peak: a 1.67 dB rise the tweeter did not have before the pad went in, centred on exactly the frequency where it is least happy to play loud. It is the same thing the textbooks describe as a rise in the driver's electrical Q: the tweeter now looks back into R1 ∥ R2 — 2.64 Ω here — instead of the near-zero impedance of an amplifier, so less of its resonance is damped.
A plain series resistor, the other way people pad a tweeter, does both of these jobs worse. To lose 6 dB at 8 Ω it needs 8 Ω of its own; the tweeter then gets −7.36 dB at Re and −3.93 dB at the peak, a 3.43 dB rise, and the crossover sees 14 Ω to 22 Ω instead of the 8 Ω it was designed for.
Rounding the pair, not each part
The exact values are never catalogue values, and rounding each resistor to its nearest separately treats two coupled numbers as independent. The attenuation depends mostly on the ratio of the two and the load mostly on R1, so an error in one can cancel or compound an error in the other. The page searches every catalogue pair within a factor of three of the exact values and keeps the one with the smallest combined error, weighting a quarter of a decibel of attenuation the same as 5 % of load impedance — both at about the limit where you could hear or measure the difference. At the defaults the exact values are 3.99 Ω and 8.04 Ω and the E24 pair is 3.9 Ω and 8.2 Ω, which gives −5.86 dB into 7.95 Ω.
How much power the resistors take
Into its design load, a pad splits the power it is fed in a fixed proportion: R1 takes a share of (1 − A), R2 takes A·(1 − A) and the tweeter takes A². At −6 dB that is about half, a quarter and a quarter. So for any real attenuation the series resistor dissipates more than the tweeter does — twice as much at 6 dB, and more as the pad gets heavier. Where the tweeter sits at Re the pad draws more current than at nominal, and R1 takes more still.
The page sizes the resistors from the share of the amplifier's power that music puts above the crossover, taken at whichever tweeter impedance is worst for each resistor, and asks for a rating at least twice that — a power resistor at its full rating runs hot enough to drift. At the defaults R1 dissipates 5.75 W, so a 20 W part; R2 dissipates 3.17 W, so a 10 W one. A full-power sine above the crossover — a feedback howl, a test tone, a dropped phono cartridge — would put ten times that into each part. No sensible resistor rating survives it, and neither does the tweeter, which fails first.
When the level bump matters
The rise is confined to the region around the tweeter's resonance, and a crossover set well above that resonance — twice it or more, which is good practice for a tweeter anyway — has already rolled the signal off by the time it arrives there. It matters when the crossover sits close to resonance, when the tweeter is a dry dome with a tall peak, or when the pad is heavy. The fixes are the ones that flatten the impedance before the pad sees it: a series LCR notch tuned to the resonance, or a Zobel against the voice-coil inductance; the crossover calculator sizes the Zobel and the high-pass section that sits in front of the pad. To see what the attenuation is buying, the speaker SPL calculator turns sensitivity differences between drivers into the decibels a pad has to take off, and the decibel converter turns any figure here back into a voltage ratio.
FAQ
How do I calculate an L-pad?
Turn the attenuation into a voltage ratio, A = 10−dB/20. The series resistor is Z·(1 − A) and the resistor across the tweeter is Z·A / (1 − A), where Z is the impedance the crossover was designed for. For 6 dB on 8 Ω that is 3.99 Ω and 8.04 Ω. Then check the result against the tweeter's real impedance rather than its nominal one, which is what the table above does.
Does an L-pad change the crossover frequency?
At the design impedance, no — that is what the shunt resistor is for. But the crossover sees the pad's input, not the tweeter, so if the tweeter is not at the design impedance the filter sees what the pad makes of it: less variation than the bare tweeter, but not none. Where the tweeter sits at Re the default pad presents 7.36 Ω, so a high-pass section designed for 8 Ω moves up slightly in frequency.
What wattage should L-pad resistors be?
The series resistor takes the most power, (1 − A) of what the pad is fed, which is half of it at 6 dB. For home hi-fi 10 W to 20 W cement or wirewound parts are the usual choice for R1; the page's figure is at least twice the dissipation on program material at the worst tweeter impedance. Wirewound resistors are inductive, but at a few ohms and a few microhenries the effect stays far above the audio band.
Should I use a variable L-pad?
The knob-type attenuators sold for speakers are the same circuit on a dual-gang wirewound control, and they are fine for finding the right level by ear. Once found, measure the two resistances and replace them with fixed parts: the contacts of a wirewound control are a variable resistance in the signal path and they corrode.
Why is my tweeter's impedance lower than 8 Ω?
Because the nominal rating is a class, not a measurement. An "8 Ω" driver's voice coil typically measures 5.5 Ω to 6.5 Ω of DC resistance, and its impedance only reaches or passes 8 Ω near its resonance and again at high frequencies, where the coil's inductance takes over.