The spread of the world wide web has sadly brought with it many misconceptions about audio and electronics. If you've ever looked on the internet for information about headphones, you will have surely come across this argument before: high impedance headphones (300 ohms and up) are "very hard to drive", while low impedance headphones (less than 100 ohms) are "easy to drive".
So you get a shiny new pair of 32 ohm headphones and plug them into your device. They're very low impedance, your device will have no problem driving them. Right?
Actually your device will be screaming bloody murder, and will likely make its upset known in the sound.
There are three things that can go wrong when driving headphones:
Headphones have an impedance rating rather than a resistance rating. If headphones ran from DC, then we would simply measure the DC resistance of the voice coil inside the headphone (depending on how thick the wire is, and the length of the wire), and we would declare the headphones "300 ohms" or whatever.
But DC is not audible. Sound is AC! Headphones are a fairly complex load, the effective resistance they will present to an amplifier depends highly on the frequency being fed to those headphones - and even what shape your head is. You see, the diaphragm itself has a resonant frequency, as does the air pocket between the diaphragm and your head. There may even be multiple resonant frequencies in this complex system.
So if you plot the effective resistance of the headphones versus frequency, you will get a very wobbly graph which varies from one set of headphones to another (and even from one head shape to another):
R| e| /\ s| | | i| | | s| / \ /\ __ t| / \/ \ / \ a|---\ __/ \ / \ n| \/ \/ \ c| e| +------------------------------- Frequency
To ensure a perfectly flat frequency response, the amplifier must present a fixed voltage to the headphones regardless of the changing impedance of the headphones. In order to accomplish this, you don't want to add any additional resistance in between the amplifier and the headphones - the amplifier needs direct control over the headphones without a bottleneck in-between.
Unfortunately, any practical amplifier has an effective "output impedance". For all intents and purposes, this is exactly the same as putting a resistor in series with the amplifier's output. Designers try to minimise this resistance, but can never eliminate it entirely.
If it's a dedicated headphone amplifier, you may be looking at an impedance of 1 ohm up to a handful of ohms - or higher if it's valve-based. If it's an amplifier designed for speakers that also happens to have a headphone socket, the output impedance could be as high as 100 ohms (as they typically use a resistor divider to bring the speaker output down to headphone-compatible levels).
As the headphone's impedance drops towards the output impedance of the amplifier, the output impedance forms a very effective voltage divider. Let's say the output impedance is 10 ohms, and you have a 10 ohm set of headphones. Half your power ends up lost in the output impedance, and half actually makes it to the headphones.
But remember, the effective resistance of the headphones varies with frequency. Let's say the headphones are 100 ohms at one frequency, and 10 ohms at another. At the 100 ohm frequency, almost all of the power ends up going to the headphones. But at the 10 ohm frequency, only half makes it there. Your frequency response isn't remotely flat now!
If you hear an immediate difference between one amplifier and another, this is usually the culprit.
We can reduce this problem by using high impedance headphones. When your headphones are 300 ohms or more, a few ohms of output impedance will have bugger all effect. You might still hear a slight change between a dedicated headphone amplifier and the headphone socket on a speaker amplifier (where the output impedance is frankly stupid). But even then it's tolerable!
In electronics, a "hard to drive load" is a low impedance load. Just because you're an audiophile does not mean the laws of physics can be suddenly suspended!
To understand the problem with low impedance loads, take a look at one side of a (very simplified) output stage for an amplifier:
+ Supply ---------+
|
B |/ C
Signal >--------|
|\> E
|
+--->
Headphone
+--->
|
Ground -----------+
The transistor in this configuration is called an emitter follower. Whatever voltage you apply to the base terminal (B), you will also see appear on the emitter (E) terminal. This makes it an ideal choice for an amplifier. The signal is first increased in voltage in a previous stage of the circuit. Then it passes to the emitter follower where the current is amplified to be able to drive the headphone.
Unfortunately, the emitter follower circuit has some limitations. The voltage you get at the output is actually slightly lower than the voltage you apply to the input.
Some of this difference is a static offset of about 0.6 volts - but it varies depending on temperature. This can be corrected for and isn't a show-stopper.
But unfortunately, there is also a dynamic component. The voltage difference depends the amount of current flowing into the headphone.
This is really bad news, because it's not even remotely linear. Instead of just changing the volume at different frequencies (which is bad enough), it will change the volume depending on the current flowing in the headphone at that instant. A sine wave becomes horribly misshapen. This is distortion in one of its ugliest forms. The ear is incredibly sensitive to distortion since it creates additional harmonics.
| /\ /\ | /| /| | / \ / \ | / \ / \ |/ \ / \ / --> || | | | | | \ / \ / | \ / \ / | \/ \/ | \| \| +---------------------- +----------------------
Even worse, it allows inter-modulation distortion to occur - all the frequencies in the music can interact with each other in highly non-musical ways.
Worse still, some of these problems don't even show up at their full extent when the amplifier is driving a simple dummy load resistor - they need the complex load of headphones. So amplifier manufacturers get to claim outstanding figures on the bench, while in the real-world performance may fall short.
Because this distortion problem is so serious, almost every practical amplifier incorporates some negative feedback into its design. The output voltage is compared with the input voltage, and a correction is made until it's proper.
This goes a long way towards making this amplifier family - consisting of so-called Class A, Class B, and Class AB - practical for audio amplification. But negative feedback is not a total cure-all. By necessity, the amount of negative feedback must be slightly less than is required to totally correct all errors. If you try too hard for absolute correction, the amplifier becomes unstable and tends to oscillate.
So no matter what, some distortion slips through.
How can we reduce the distortion?
Generally speaking, make the transistor(s) work less hard. Less current must flow through the transistor(s), higher bias currents must be used, and the transistor(s) must be over-rated for the task.
This is why high impedance headphones are "easy to drive".
The amount of current high impedance headphones require for a given amount of power is a fraction of that required by low impedance headphones. This allows the amplifier to attain much better distortion figures, as the output transistors are barely tickling the headphones with minuscule current.
But if you drop from 300 ohms down to 32 ohms, suddenly your currents are 10x higher. This will impact the performance of the amplifier, and the only real way to correct it is to essentially over-build the amplifier. If it's something like a portable music player, you may be stuck with whatever output stage is built into the headphone amplifier IC. There simply isn't room inside the case for a discrete output stage with over-rated transistors. Even if there was room, an over-built amplifier will pull more bias current at idle. Your battery will go flat really fast.
Ah, and now we come to the true reason the "hard to drive" misnomer about high impedance headphones started!
Nothing in this world comes for free, and power is voltage times current. Since we already established that high impedance headphones requires much less current to drive, this also means they require much more voltage to drive to a given volume - all else being equal.
For a sensibly-designed mains-powered audio device, you will have +/- 15 volt power rails available. This is enough voltage to ignite any set of headphones and then some.
But a mobile device running from a 3V lithium battery, that could be a problem. You might want a capacitance multiplier before the amplifier to remove digital noise from the supply rail. Lose a volt, you have 2V to play with. There isn't enough space inside the case for big output capacitors, and you've only got a single positive supply rail. So you need to have a third amplifier driving a "virtual earth" that sets the headphone ground at the mid-way point - 1V. So now you can only swing 1V in each direction. On top of that, any practical amplifier design won't be able to swing completely to each power supply rail. It might not even perform well beyond +/- 0.5V.
So you can see how in some power-starved mobile devices, you may find a high impedance set of headphones could be on the quiet side. But still, +/- 0.5 volts is actually not too bad for driving headphones to a decent volume.
But not all headphones are made equal. Ultimately, it all boils down to the sensitivity of the headphones. Given a fixed amount of power, one pair of headphones can be nearly deafening while another pair can be barely loud enough. If you get unlucky and have a high impedance set of headphones which are particularly insensitive, you will have problems. But if you get lucky and have incredibly sensitive high impedance headphones, you will likely find them too loud at maximum volume on any device.
In practice, I have found a pair of 300-ohm Sennheiser HD600s to be capable of nearly deafening volume on any mobile device I have tried. Your mileage may vary.
But let's not go around saying they're "hard to drive". It might be difficult to get enough volume on some devices, but realise that they are going to be easier on your amplifier - and the audio quality will thank you for it.
I'm sure in many cases people complain about headphones being "too quiet" on shaky ground. Of course if you're going from a 32 ohm set of headphones to a 300 ohm set of equal sensitivity, swapping them over is going to need a volume adjustment! Some people will feel they're too quiet just because they have to use the upper end of the volume control's range - and they're not used to doing that. Or perhaps you're listening to music which is recorded too quiet in the first place and your playback device has no way of adding gain to correct for it. Or worse still, your device has speakers which are muted when inserting the headphone jack, and no independent volume controls for headphones and speakers. Good luck matching the two!
These problems can be solved with better software. ReplayGain, independent volume settings for headphones and speakers, and volume controls which allow actual gain. Some devices already do it.
But you should seriously ask yourself "is it loud enough?"
Volume in headphones is very hard to judge. We don't have ambient sounds to compare to, and there are no whole-body bass vibrations from a big set of speakers. Studies have suggested time and time again that people listen to music much louder in headphones than they do with speakers - even if they're not aware of it.
Worse yet, some people assume that when they plug headphones into a device with speakers, the same volume will be sent to the headphones. They may not even realise there is 10x the sound pressure hitting their eardrums because their headphones are more sensitive than the device's designer bargained for.
If you're not struggling to hear it over normal indoor background noise, it's probably loud enough. Remember, you only get one set of ears. They will last most of your 80 years if you keep volumes down to ordinary indoor noise levels. At any elevated levels whatsoever, those ears of yours will burn out early.
Oh but that's the plan isn't it? The headphone collection becomes the pension after you go deaf.
When switching between speakers and headphones, always lower the volume to zero first. Then slowly ramp up until a comfortable level is found.
If you're one of those weird audiophile types, you've probably shunned these anyway.
Class D amplifiers do behave more linearly into low impedance loads than class A/AB/B. The transistors are either fully on or fully off - no in-between section where the linearity is at its worst.
But they're still not completely linear. All else being equal, a class D amplifier should perform better into a high impedance load.
There may be exceptions - there always are. The output filters in your amp could perform best with a low impedance load. There may be other non-linearities that creep in with high impedance loads.
But any good digital headphone amplifier will be designed to run loads between 16 ohms and 600 ohms - the common range of headphone impedances. Without knowing anything about the specific amplifier, I would aim for the middle of this range for best performance.
Sometimes it's handy to be able to plug headphones directly into a line output - if you don't have an amplifier to hand.
Now look who's the quietest! Yes, those 32 ohm headphones are barely audible, while the 300 ohm set produce a comfortable volume.
This is because the output impedance of a line output is typically 100 to 2000 ohms. A 32 ohm pair of headphones absolutely swamps this, and you have a fraction of the original signal left. Worse still, they pull so much current that the output opamp will surely distort.
But with a 300 ohm pair of headphones, you can often get away with it. Most opamps can drive a 300 ohm load without much distortion. And that 100 ohm output resistance, that's not going to affect it much. Even at 2000 ohms you might still be able to hear it well enough.
I don't recommend running headphones from line outputs. But the higher impedance set is the one to use if you have to.
Sometimes your electric kettle is broken and you need to make a cuppa fast. The 300 ohm set of headphones immersed in water will dissipate about 200W per channel when connected straight across the mains. But your 32 ohm set will dissipate a healthy 1.8kW per channel. In this application, I recommend the 32 ohm set of headphones. Nobody wants to wait for a cuppa, and the 32 ohm set isn't much use outside boiling water.