Headphone impedance and power: why your amp's output matters more than watts

I had a reader email me last month convinced his new amp was faulty. He'd bought a decent desktop DAC/amp, plugged in a pair of high-impedance planars, and the bass sounded thin, almost polite, compared to the demo he'd heard in a shop on a different amp. Nothing was broken. He'd just run into the single most misunderstood spec in headphone electronics: output impedance, and the way it interacts with the headphone's own impedance curve. Watts had nothing to do with it.

This is a topic I get asked about constantly, because the marketing around headphone power is, frankly, a mess. Brands quote power output into 32 ohms because it's the biggest number they can put on a box, and most buyers assume bigger equals better, full stop. It doesn't. The number that actually predicts whether an amp will sound right with your specific headphones is rarely printed anywhere, and when it is, it's buried in a PDF nobody reads.

Why watts alone tell you almost nothing

Headphones are wildly more efficient than speakers. A pair of IEMs might need a fraction of a milliwatt to reach uncomfortable volume. Even a power-hungry planar like the Audeze LCD-X needs only a few hundred milliwatts to get loud in most rooms. Compare that to a floorstanding speaker chewing through tens of watts, and you can see why "how many watts" is the wrong first question for headphones.

The right first question is: what's the headphone's impedance curve, and what's the amp's output impedance? Those two numbers together decide whether the frequency response you hear matches the frequency response the headphone was designed to produce.

A headphone driver isn't a flat resistive load. Its impedance rises and falls across the frequency range, often quite dramatically. A dynamic driver headphone typically shows a big impedance peak somewhere around the bass and lower-midrange, tied to the driver's mechanical resonance, then settles into a flatter response through the mids and treble. Planar magnetic headphones behave differently again, usually with a much flatter impedance curve overall, which is one reason they tend to sound more consistent across different amps.

The voltage divider you didn't know was in your signal path

Here's the bit that actually matters, and it's simple electronics, not folklore. Every amplifier has some non-zero output impedance, even a good one. When you connect a headphone, you've created a voltage divider between the amp's output impedance and the headphone's impedance at each frequency. If the amp's output impedance is small relative to the headphone's impedance, the divider barely affects anything, and the amp drives the headphone flat, as intended. If the amp's output impedance is large relative to the headphone's impedance, the frequency response gets warped, because more voltage gets dropped across the amp's own output stage at the frequencies where the headphone's impedance dips lowest, and less reaches the driver there.

Practically, this shows up as a frequency response tilt that tracks the headphone's own impedance curve. On a dynamic driver headphone with a bass-region impedance peak, a high-output-impedance amp will actually boost the bass slightly at that peak frequency, because that's where the headphone's impedance is highest relative to the divider, and less of the signal is lost there. Elsewhere, where impedance dips, you lose relative level. The net effect isn't usually "louder bass" in the way people expect. It's a shift in the tonal balance and damping character that changes how the headphone integrates through the crossover-ish region where the driver's own mechanical behaviour is doing something interesting.

The consumer electronics standard, IEC 61938, recommends amplifier output impedance stay under one-eighth of the nominal headphone impedance for this reason, though plenty of gear on the market doesn't hit that target and doesn't advertise it either.

Why this bites harder with some headphones than others

Multi-driver, multi-BA IEMs are the worst-case scenario, and it's not close. A lot of high-end IEMs use passive crossover networks with multiple balanced armature drivers, and those networks produce impedance curves with wild swings, sometimes ranging from single digits to over 50 ohms across the audible band, with sharp peaks and dips. Pair one of those with a source that has meaningfully high output impedance, like some older phones, some cheap dongles, or tube headphone amps, and you can get audible frequency response errors of several dB in places. That's not subtle. That's a different headphone.

Dynamic-driver over-ear headphones are more forgiving because their impedance curves are smoother, generally one broad hump rather than several sharp ones. Planar magnetic headphones like the Audeze LCD-X (check price) or the Focal Clear MG (check price) are the most forgiving of all, since their impedance curves are relatively flat by design, which is part of why they tend to sound consistent whether you're running them from a proper desktop amp or something more modest.

Full-size headphones with genuinely high nominal impedance, the 250 and 300 ohm dynamic designs like the Beyerdynamic DT 1990 PRO MK II (check price) or the Sennheiser HD 660S2 (check price), sit in an interesting middle ground. High nominal impedance makes them less sensitive to output impedance mismatch in relative terms, since the divider ratio stays more favourable, but it also means they need more voltage swing to reach reference levels, which is where the "you need a proper amp for high impedance headphones" advice actually comes from. It's not that they need more raw power in the wattage sense. They need more voltage headroom, and a lot of phone and laptop headphone jacks simply can't supply it before clipping.

What actually decides whether you need a separate amp

I'd rather give you the honest decision tree than a blanket rule, because the internet's blanket rule ("always get a separate amp") isn't true and it costs people money for no audible benefit.

If you're running efficient dynamic-driver headphones or most IEMs from a reasonable modern DAC/amp or a decent phone, you very likely don't need more power. The limiting factor for you is going to be output impedance and noise floor, not voltage or current capability. Check the source's output impedance if you can find it, and if you're running sensitive multi-driver IEMs, treat anything above a couple of ohms with some scepticism.

If you're running high-impedance dynamic headphones designed with studio and audiophile use in mind, or planar magnetics with low sensitivity, that's where a dedicated desktop amp starts to earn its keep, mostly by supplying enough clean voltage swing to hit reference levels without compressing dynamics on peaks. This is the genuine use case for something like the iFi Audio ZEN DAC 3 (check price) or the RME ADI-2 DAC FS (check price), both of which publish honest output impedance figures rather than hiding them, which tells you something about how seriously the engineering was done.

I'll admit a mild bias here: I trust brands that publish an output impedance spec at all, because it means someone in the design team actually thought about the interaction rather than just chasing a power number for the box. FiiO does this reasonably well on gear like the FiiO K11 (check price), and it's one of the reasons I keep recommending budget desktop amps from that category over unnamed dongles with no published spec sheet at all.

Reading a headphone's impedance curve, if you can find one

Independent measurement sites publish impedance sweeps for a lot of popular headphones, and it's worth ten minutes with one before you buy an amp, particularly for IEMs. Look for how much the curve swings, not just the nominal number printed on the box. A headphone quoted at "16 ohms" that actually swings from 10 to 45 ohms across the band is a very different proposition to one that sits flat at 16 ohms throughout. The nominal figure on the spec sheet is often measured at 1kHz and tells you almost nothing about the shape of the curve either side of it.

If you can't find a measured curve for your specific model, the safest default is still to favour a source with low output impedance, ideally under 1 ohm for anything with a complex crossover, and not worry about it much further if your headphones are simple single-driver dynamics.

Where the desk fits into this

None of this happens in isolation from the rest of your desktop chain, and I've written before about getting the desk itself sorted before you start chasing amp upgrades, because a lot of perceived "amp problems" are actually gain-staging or noise-floor problems introduced upstream. If you're assembling a desktop rig from scratch, our guide to the best desktop DAC/amps covers a spread of options across this exact power-versus-output-impedance trade-off, and our broader look at the best audiophile headphones notes nominal impedance for each pick, which is at least a starting point even if it's not the full curve.

The reader with the thin-sounding planars, by the way, didn't need a more powerful amp at all. His DAC/amp's output impedance was fine; the actual issue turned out to be gain set too low and the headphone's efficiency being on the lower side, so he was running out of clean voltage swing before he ran out of level. A firmware update and a gain switch fixed it. No new hardware required, which is not the answer anyone wants when they've been eyeing off a new amp, but it's the honest one.

The number worth asking for

Next time you're comparing headphone amps, I'd skip the watts-into-32-ohms figure entirely and ask two questions instead: what's the output impedance, and does it publish a full power curve across the impedance range you actually own. Most manufacturers can answer the second if pushed, even if it's not on the box. The Bluetooth SIG and the audio industry more broadly have spent decades standardising connector and protocol specs; headphone output impedance disclosure remains stubbornly optional, which is exactly why it's worth checking yourself before you spend.