Class D amplifiers for Headphones

This topic is sparked by a post from @GoldenSound in another thread but which has me curious. So I would like to know more info on the subject and thought it may also be beneficial to others as it is a topic that I am not familiar with, so maybe I am not alone :slight_smile:

The post, or section of the post, I am referring to is:

Could you explain further this subject?

I have multiple class D amps, including some Fosi ones (with a new one awaiting trial and review), so I am interested to know what the details are behind this.


Class D amplifiers are not load invariant. This means that with higher than expected load impedances, they will behave abnormally.

Class D amps use an output filter, and the driver itself forms a part of that filter. If the impedance is different than the intended load, you will have issues.

The Fosi V3 for instance seems to be set up for 4 ish Ohm speakers. And even at 8 Ohm it’s already over-amplifying high frequency content quite a bit, including quite significantly in the 30-50khz region

Here’s a simulation of a class D output filter with various loads. With a 155 Ohm load of tungsten, you could be creating a couple dB boost in the upper treble, with over TWENTY DB increase at 40khz.
L = 11.26uH
Cbtl = 0.703uF

This could mean that at best you’re colouring the sound as content under 20khz is affected, and at worst could potentially cause damage to your headphones or hearing if your amp is pumping high level high frequency content through. And you might not even realise it’s happening, because you can’t hear 40khz, but could still incur hearing damage.

Purifi stuff is apparently load invariant through the use of a different feedback approach, though I don’t know whether the amp will actually be performing as well with a 155 ohm load


I have several Aiyima Class D amps that I use with Klipsch KG 2.2’s & Radio Shack Optimus PRO77’s. The speakers are nominal 6 Ohm If I Remember Correctly. No ill effects observed, but I don’t think either set of speakers would have any significant output at 30 - 40khz.

No Class D’s being used with headphones.

Mark Gosdin

1 Like

Thanks for the explanation.

Before asking a few more questions, I just want to point out the obvious, I am not an engineer nor de I have much engineering knowledge at all (at least outside of basic design in the field I work in).

Chatting to a couple of engineers who do work in the amp design field (not headphone related, loudspeaker related), they were saying that their amplifiers use filterless modulation that mitigates the need for filters on the output, and also serves to reduce the EMI that was a major issue also with previous designs.

They obviously have no idea how the Fosi amplifiers are designed and implemented (nor are they interested in knowing) but they do say that if they are a competently designed amp, then it should not be an issue.

They did go on to speak about how it worked (saw waves or something) but I was way out of my depth by that moment :slight_smile:

Aren’t the vast majority of amplifiers on consumer devices (phones, audio receivers, dongles, etc.) all class D amplifiers that drive the headphone outputs?

1 Like

Hard to say without knowing who that was or what in particular they’re referring to.
There are as mentioned some like Purifi which are apparently load invariant, but generally most class D amps with normal LC low-pass filters (and lacking the feedback design purifi uses) will not be load invariant.


I bought and returned a Fosi TDA7498E a few years ago. I heard screeching, raspy, saw-tooth, digital treble. I’d been hoping to save space versus my old full-size speaker amp. Nope. Not a chance with that one. The treble artifacts were immediately obvious.

I’m thinking some of that excess treble indeed colors the audible frequency bands. High end Class D amps sound pretty high end, but I’m giving the lower price tiers a few more generations before trying again.


Interesting thought. Anti-pest devices are sold that are supposed to put out about 100-105 db in the 40-50K range. They are sometimes effective in keeping rodents away. I’ve had them in the house for years.

Last year when my Sansui Integrated amp for the office speakers - 6 ohm nominal - was being repaired, I replaced it temporarily with a Topping PA3S class D. I also tried and returned a Fosi TB10D class D. The Fosi was sold with an undersized power supply, so it could not produce the rated output, and it simply didn’t sound as good as the Topping. The Fosi had tone controls, the Topping volume only.

There were no headphone outputs, just banana/wire speaker outputs. The Topping did not seem headphone friendly either.


That explains everything about @pennstac’s personality and audio taste, and a whole lot more. This fact has nearly as many implications as 42.


Favorite song:
Roach Hotel California? Things like that?


This is pretty easy… look at the second graph and notice the 4ohm FR/Gain drops off and the 8ohm FR/Gail raises … so the sweet spot is between 4ohms and 8 ohms… so today I put together two of the Fosi Za3 amps and the Hifiman HE-Adapter box… with the load box mounted to the output stage of the amps the amp stage is presented with a 10 ohm load resistor in parallel with the output and with a headphone connected to the load box then by ohms law resistors in parallel the resistance will be lower than the least resistance… so the load presented in the case of a Tungsten will be slightly smaller than 10ohms… the FR/Gain at that resistance is very slightly elevated in gain and the Frequency is extended further out in a flat projectory… in other words with the ZA3 Amp and the load box connected the amp is being presented with the proper load and there is no issue… I didn’t have a Tungsten during this test so I used one other hard to drive headphone… the Hiflman HE6sev2 and with that headphone and the HE load box of 10 ohms I can tell you that I have never heard the HE6sev2 sound any better… the volume knob was set at the 12 oclock position and the source was used for the volume changes… it was quite an experience with that much power/headroom and how the notes were fast and tight with lots of slam… the FR was very well balanced across the range… it was amazing sounding actually… I am planning on tuning the amps a bit more via the opamps from Sparkos Labs and see if the amps will scale even more… should be interesting… if your looking for a lower cost solution to drive the Tungsten then I think the ZA3 amps connected to the Hifiman HE-Adapter box should do the trick quite well… btw for the source and easy hookup I just used an Iphone via Apple Music and a wireless connection to the BTR7… it was a easy solution for the test today… but any source with variable output will work great with this setup


As the happy owner of a big 2-channel class D amp (Wyred 4 Sound ST-500) + quite a few headphone amps/preamps and headphones, I’m just going to put this out there: why bother with class d for headphones at all?

Here’s why the question:

  • The primary advantages to class D are size and lack of heat

  • However, in many if not most headphone applications (where the amps are generally small and lower power than trad 2-way amps), neither of those advantages really hold up.

My best SS amps are the Wells Milo, the Violectric V281, and the humble but great-sounding Cavalli Liquid Carbon v2. I also have 2 tube amps (Woo WA3/OTL and Icon Audio HP8/TC). All are compact (the LC to an extreme) and permit balanced headphone cables. These devices fit well even on my cramped desktop, so I’m not among those who truly need something smaller. The tube amps put out some head, but I generally have only one on at a time, and only for short periods, so heat isn’t a big problem here.

The W4S amp is another matter. There is no way I could physically accommodate a class A/B amp in the 200+ WPC range, and the heat output would be problem, as well. But I need 200+ WPC to properly drive either of my big desktop 2-ways (ATC SCM12 Pros and the current pair, KEF 103.2s), at least on those occasions I really crank them. These 2-ways are both sealed/acoustic suspension designs with the great bass of that design type, but they soak up power. It was class D or nothing. The W4S amp sounds great and puts out so little heat that (with Wyred 4 Sound’s permission) I installed it resting on one side–the only way I could fit it IMS.

No doubt there are headphone users with even less space and heat tolerance than I have here, and for them class D headphone amps might make perfect sense. But I suspect that’s not a large group…


Favorite book:
Kafka’s “The Metamorphosis”


I’m thinking of your diet.


Sort of gives new meaning to diet of worms.

1 Like


Yeah, we know … it literally says that on the line immediately above …

They’re not remotely the same thing.

In the case of your AVR-X2600h, it does 95W into 8 ohms. Its headphone output is tapped off the speaker outputs, and then passed through a pair of 470 ohm resistors, which would yield a theoretical ~1,600mW … assuming the headphones had ZERO impedance.

Not really.

Assuming the double-sided Tungsten (which require less power that the single-sided for a given SPL), they have an impedance of 155 ohms. So this drops your available output to ~1,200mW, BUT the headphones also form a voltage-divider with the headphone output, which would result in 75% of that power being “lost” (unavailable to the headphone) … meaning you’d get just 300mW of power to the headphone.

That’s enough for a peak level of 102 dB/SPL, at the amplifier’s maximum output (which is a long way from where it sounds its best … a low bar for this unit in the first place). Which for an average listening level of 80 dB doesn’t leave much for musical peaks. Unless you listen at lower-than-typical levels, or only have a diet of low-dynamic-range music, it’s not going to be a satisfactory solution.


Torq…Then a question I have for you is if the DAN CLARK E3 is a compatible headphone to the DENON X2700H receiver? (more than something like the TUNGSTEN)? Is there any audiophile headphones that Torq or anyone can offer a suggestion to use through my source? Thank u. :small_blue_diamond:

They’ll all technically “work” … the question of “how well” (in quantitative rather than qualitative terms) comes down to how loud you listen, how much power the headphone you are using needs for a given volume level, and how much headroom you need for the type of music you’re listening too.

For the DCA E3, you’d have 85mW available after you factor in the HUGE voltage divider issue that a low-impedance headphone like the E3 is going to yield with your AVR’s ridiculous output impedance*. That’d get you to 109 dB/SPL … at maximum output (again, not where amplifiers tend to shine).

I wouldn’t do it.

You’d be vastly better off with any competent $100-$150 headphone amplifier. But you can’t pair those properly with your receiver since it doesn’t have proper line-outs (it has pre-outs, but they have constraints on signal routing … as they’re intended for a powered sub-woofer and a 2nd zone power-amp).


Is it just me that has a feeling that I have read this before…


@Cyan69: Do this. Use the right tool for the job.

There simply isn’t an “audiophile headphone” that’s a good match, because no designer of a $2k headphone is concerned about how it will sound on an amp with about a quarter of a watt of available power, virtually no headroom, and a low-pass filter on its output.


  • use a (relatively inexpensive) headphone that doesn’t make many demands on its amp (think HD58X / Koss Porta Pro) with the AVR. It might be more “audiophile” than you would guess.


  • If you just have to have that $2k headphone, find a few hundred more to get a proper amp for it.