That 3.5mm or 6.25mm socket on the front panel of most home theatre receivers is ignored by most equipment reviewers, despite there being a large market for headgear. Stephen Dawson explores why they aren’t as straightforward as they seem.

Unfortunately, my much-loved ASUS Zenbook notebook computer semi-died last year while I was overseas. On my return, I replaced it with a recent Lenovo Yoga, which was only a little chunkier.

Then it became apparent that I needed portability not available from my Mac Mini, given its external monitor and keyboard. I replaced it with a MacBook Air M4.

So?

Well, I’m about to finish off a review of the FiiO JT7 budget planar magnetic headphones. They are astonishingly good for the price, and they work magnificently with the MacBook Air, but they would be a very poor purchase choice for the Yoga.

Why would all that be?

Impedance 

It’s all to do with impedance. It’s a simple enough figure, but it is very rarely specified for the headphone outputs of amplifiers… unless the particular electronics is specifically designed for headphones. It can have a huge effect on the performance of headphones, most of which also have an impedance that varies significantly across the audible frequency range.

Consider loudspeakers. They also have a variable impedance across their frequency range, but everyone kind of expects that. And their “nominal” impedance is almost always somewhere between 4-8Ω. Amplifier outputs are designed with that in mind, and the great majority are designed with a low output impedance to minimise any influence on the frequency performance of the speakers.

By contrast, the output impedance of many devices driving headphones is unknown, but often remarkably high, while the nominal impedance of headphones and earphones ranges from as low as 8Ω to, sometimes, in excess of 300Ω, and their variation around the nominal value is typically unknown to the purchaser.

If, in the absence of this information, the result of matching any particular set of headphones to any particular amplifier sounds really nice, this can be considered a fluke. Because we aren’t talking about subtle “audiophile” differences in sound here. We’re talking about huge differences that everyone can notice.

Let’s go back to 2018 

Even though I’d been reviewing gear for a couple of decades, I hadn’t paid too much attention to headphone outputs. I had an office out the back of the house and could use my many speakers without too much concern for volume levels. So, I rarely bothered with headphones.

But I had a Marantz NR1609 slimline receiver in for review, and for some reason, I had to use headphones. It sounded awful with them. By that point, I’d owned those Sennheiser HD535 headphones for over twenty years. And they shouldn’t have sounded so poor.

Trying to work out what was going on, I measured a bunch of things, including the output impedance of the unit’s headphone socket. That turned out to have an impedance of 468Ω.

The HD535 headphones have a nominal impedance of 150Ω, or one-third of the output impedance.

As basic electrical circuit theory tells us, when you apply a voltage across a circuit with several components in series, it is divided across those components in proportion to their resistances. So, an output of, say, 20V from the Marantz across the nominal impedance of the headphones would see more than 15 of those volts going as heat in the output’s internal resistance, and less than 5V powering the headphones themselves*.

But the resistance in the headphones varies by frequency. According to my measurements, because of that varying resistance, with the Sennheiser HD535 headphones, the voltage into the headphones was 4.5dB higher at 110Hz than it was at 1.5kHz. That made them boomy, without giving a real sense of true bass depth.

See Figure 1.1 for what the Sennheiser headphones look like. That’s from a logarithmic sine wave sweep from 20Hz (left) to 20,000Hz (right).

Let me be clear: this isn’t the sound delivery of the headphones. This is the voltage delivered to the headphones, expressed in decibels. That’s what the headphones start with.

The problem was worse with my Focal Elear headphones (Figure 1.2). They have a nominal impedance of 80Ω, and they had a 100Hz bass 7dB higher than the midrange.

Okay, you might think, a bit of bass boost is generally enjoyable! True. But what if you use a pair of Audiofly AF180 Mk2 in-ear monitors (i.e., high-quality wired earbuds)? With that Marantz-like output impedance, they are provided with a 12dB boost at 3.1kHz compared with the bass, and another at 5.8kHz. They sound truly dreadful as a result.

While I’m at it, there are the same graphics for (high-ish end) Final Audio B3 IEMs in Figure 1.4.

But if I plug the Audiofly IEMs, or the Sennheisers or the Focals or the B3s into a FiiO M21 or M11 Plus digital audio player, or a FiiO DM13 portable CD player or an iFi Zen Can headphone amplifier or any of several Astell&Kern DAPs or a Topping A30 Pro headphone amplifier or even a MacBook Air M4, they all sound fine!

That’s because those other devices all have a low output impedance of 2Ω or less. Even the Apple Lightning dongle for headphone output from an iPhone, or the equivalent for an old Google Pixel 2 XL were fine!

Marantz wasn’t alone! A NAD and a Denon from the same period had a similarly high output resistance. My ancient NAD 1010 preamplifier was around 230Ω. My Denon 90’s CD player was similar to that.

I could carry on. My spreadsheet has about 80 entries for this value.

In general, portable digital audio players and dedicated headfi gear have low output impedance, and most stereo amplifiers and home theatre receivers have a higher one, sometimes 100Ω, often way more.

And, typically, the consumer has no idea. And no idea how this is going to affect the sound quality of their headphones.

Why 

Why do so many headphone outputs have high output impedances?

Well, there are no standards to begin with, so there’s that. But in my estimation, the industry is gradually bending towards low impedances.

Meanwhile, without being able to examine the intentions of equipment designers, here’s what I reckon. First, if you are making a stereo or multichannel device to drive loudspeakers, why not save a bit of money and have the loudspeaker outputs drive the headphone socket as well? Oh, you can’t put tens of watts into headphones, but if you stick a whopping big resistor in line with the output, that problem is solved.

Or, you might have a source device such as a streamer, CD player or such. Providing a headphone output capable of providing high current outputs might be a bit too expensive, since the headphone isn’t really the main game here. So, you use a lower quality output amp and protect it by putting in an in-line resistor. The preamplifier I use for my system costs more than $5K, yet it has 100Ω inline on the headphone output.

But products that are devoted to use with headphones or earphones – headphone amplifiers and portable digital audio players, primarily – do choose electronics with low output impedances, and whatever necessary output protections are required.

Headphone types 

The two headphones I mentioned above are dynamic models. They represent the great majority of headphones. They are, essentially, little loudspeakers with coils on a diaphragm. The two IEM’s use multiple balanced armature drivers and crossover networks between the drivers. It is these designs that have an impedance that varies by frequency. And it is that which creates the weird signals when they are connected to high impedance headphones.

You can avoid all these problems by choosing planar magnetic headphones. They have other virtues as well, but for our purposes, their main one is that they have an even impedance across their frequency range. For example, Figure 1.5 uses the same graph as shown earlier, except for those FiiO JT7 headphones I mentioned at the top of the article.

The variation in voltage delivered to them across the audible bandwidth is less than 0.1dB. I’ve measured several other planar magnetic headphone models from several makers, and they are all similar. That makes them immune to frequency variation according to the output impedance of the source.

If you want to buy headphones to use with an amplifier, and you do not know its output impedance, as is very often the case, go for planar magnetic.

But that doesn’t necessarily solve all the problems. Such as with a Lenovo Yoga computer.

Levels 

When I tried the FiiO JT7 headphones with the Yoga, they sounded perfectly adequate, but I could not get a respectably high level of sound out of them. The headphones had worked fine with plenty of other equipment. I plugged them into the MacBook, and voila, plentiful sound.

Again, it’s the impedance.

I measured both computers to find out what was going on. The output resistance of the MacBook was 0.3Ω. That for the Yoga was 69Ω. The FiiO JT7 headphones have a nominal – and since they’re planar magnetic, actual – impedance of 18Ω. While the MacBook could provide 158mW of power into 16Ω headphones, the Yoga could manage only… 2.5mW. These headphones are rated at 92dB sensitivity for 1mW of input. As a result of their low impedance, a maximum of just 4dB above sensitivity – 96dB that is – was available on even the most powerfully-recorded music.

But the MacBook could push the headphones to a massive 22dB or more above sensitivity: 114dB SPL.

That 69Ω killed the output voltage of the Yoga into low impedance headphones. Into an open circuit, it could deliver 1.1V RMS. Into 300Ω headphones, it could still deliver 0.9V. But into a 16Ω load, its output voltage dropped to just 0.2V. Thus, the low output.

The MacBook Air delivered almost 3V RMS into an open circuit, almost 3V into 300Ω, and 1.6V into 16Ω (I had to turn it down to avoid “clipping” distortion).

Most planar magnetic headphones have a higher impedance than the JT7s. So, they are probably okay even with a Yoga. And, of course, the workaround for the Yoga is to use a decent USB Audio DAC. An AudioQuest Dragonfly Red, for example, can manage better than half a volt into 16Ω, which means 14dB (+92dB for up to 106dB).

Rounding up 

I’d love to offer a nice bit of buying advice for all this, but there is no such thing. Yes, I’d recommend buying planar magnetic headphones for their ability to retain their tonal balance, despite poor headphone outputs. But that does restrict the choice of headphones considerably, and mostly they’re relatively expensive.

In the end, the performance of headphone outputs on all manner of gear remains extremely uncertain. Output impedance is typically not specified. How that affects any particular set of dynamic headphones, which is most of them, also remains uncertain.

If things don’t sound as they ought to with a set of headphones and a particular amplifier, then perhaps the only thing to do is to acquire a dedicated headphone amplifier. Some are reasonably affordable. And they will often provide superb performance. Just check that they actually do specify their output impedance, and that it is a low value.

*The voltage across the headphones equals 20V times 150Ω, divided by 150 + 468Ω, equals 4.85V. Also, the difference between resistance and impedance is that the latter is resistance plus reactance or capacitance. When talking about DC or one frequency, they’re the same.