[DJ The Rocket]

Short version: is there any reason it is or isn't a good idea powering high impedance headphones (50 to 300 ohm) off the speaker posts of a 45wpc amp/tuner made for 8 or 16 ohm loudspeakers? Besides accidentally turning the volume up and blowing the 'phones

Long version:
I'm currently getting the best sound I've ever had from my headphones using an old 45 wpc Kenwood KR-3070 tuner. I'm using a Q701, Alpha Prime, HE-400, LCD2, and DT880 (250) with great success. Anything much easier to drive than the (49 ohm) Alpha Primes starts to hiss loudly enough to interfere with the music.

There's nothing special about my tuner, I beat on it through high school AND college. So I figured that I could upgrade my sound further still with a better tuner, maybe a Denon, or Marantz, or Onkyo. There's a million great models out there, and I wouldn't have to spend more than a couple hundred bucks for a used one.

But when I started demo'ing units, I noticed that every single one I tried has a problem controlling the bass at the low volume levels headphones need. The bass gets badly, badly distorted. On some units the problem goes away with more volume, but we're talking too loud to listen to volumes.

What is going on here? I know the amps I'm using are designed for 8 ohm loudspeakers, but is a 50 or 300 ohm headphone load really any different, from the amp's perspective?

And if so, is there any reason why my Kenwood handles the sound perfectly at any volume, or did I just luck out? I tried my dad's Harman Kardon amplifier, which has had an easier life than my Kenwood (by far, lol), and it distorted the bass too. So I think I can rule out the possibility that all the ones I demo'ed had been abused somehow. I just don't understand enough about how this stuff works to understand what I'm hearing.

 

[ev13wt]

You want to keep a ratio of 8.1 impedance/output impedance to keep the hiss at bay and to not introduce frequency response shifts.

Possibly you are outputting juuust enough to stay in Class-A mode. (Class A/B Kenny?)

Your brain knows this amp, knows how it "sounds". Expectation bias comes into play as well with the "sounds good" part. I know because I've been there. Its the reason my dads gear that I grew up with just "sound great".

 

[castleofargh]

hard to say why and even harder to say what you find different/better.
without some measurements (ideally into a load equivalent to the headphones), I don't even have a guess.

 

[DJ The Rocket]

Possibly you are outputting juuust enough to stay in Class-A mode. (Class A/B Kenny?)

This is a really good guess that would elegantly answer all of my questions, if true!

And that's why it can't possibly be true, right? Since when does life work out like that? The Car Talk guys would roll over in their graves!

hard to say why and even harder to say what you find different/better.
without some measurements (ideally into a load equivalent to the headphones), I don't even have a guess.

You're right of course. I understand that I can't expect anything more than guesses based on your own experience, but often enough that method seems to work out

 

[ev13wt]

This is a really good guess that would elegantly answer all of my questions, if true!

On the other hand, most if not all headphone amps are also class A I think.

 

[Mambosenior]

Well, here's my yearly reveal for any to consider:

I own 10 headphones: from the humble Beyer 880 (600 ohm) and AKG 712 to the (less humble) Sennheiser 800 and 800S, and Hifiman HD-6 (I stopped the upgrade dragon at this point long ago). I use speaker amps exclusively except for the occasional foray to my La Figaro 339 when I'm in the watch-tubes-get-lit-up mood.

Only fried things I've encountered came from McDonalds.

 

[pinnahertz]

On the other hand, most if not all headphone amps are also class A I think.

Umm...I don't think. In fact, I have one right here that isn't class A. I would think most are A/B, but outside of things getting a bit hot in Class A I doubt it would matter much relative to other design parameters.

 

[pinnahertz]

You want to keep a ratio of 8.1 impedance/output impedance to keep the hiss at bay and to not introduce frequency response shifts.

Any speaker amp would already meet the impedance ratio. But that has nothing to do with noise or hiss, that's a gain issue. A speaker amp has far too much gain for headphones, which reach their rated SPL with 1mW applied, as compared to speakers which are rated at 1W/m. Connecting headphones to a speaker amp barefoot is just silly.

Possibly you are outputting juuust enough to stay in Class-A mode. (Class A/B Kenny?)

Your brain knows this amp, knows how it "sounds". Expectation bias comes into play as well with the "sounds good" part. I know because I've been there. Its the reason my dads gear that I grew up with just "sound great".

Doubtful the bass distortion problem is any of that, but without more information there's no way to know.

 

[pinnahertz]

To elaborate, now that I have a bit more time...

A speaker and a headphone have power requirements that are different by somewhere between 30 and 40dB or more, for IEMs. That means in order to operate a speaker amp volume control in a comfortable position for headphones you'd need to adjust the apparent gain of the amp by at least 20dB or more. It also means that relative to connecting a pair of speakers to the same amp, residual noise will be 30-40dB higher in the headphones. Because of this, many early amps with headphone jacks simply place a pair of build-out resistors at the amp output and drove headphones through them creating a sort of pad. However, that also significantly raises the output impedance seen by the headphones, so what's needed is a real attenuator, something like a 50 ohm resistor and a 1 ohm resistor, taking the headphone output across the 1 ohm. Seems like a lot of trouble to go through when using an actual amplifier designed to drive headphones would be much better from the standpoint of noise, gain control, and output impedance.

 

[71 dB]

To elaborate, now that I have a bit more time...

A speaker and a headphone have power requirements that are different by somewhere between 30 and 40dB or more, for IEMs. That means in order to operate a speaker amp volume control in a comfortable position for headphones you'd need to adjust the apparent gain of the amp by at least 20dB or more. It also means that relative to connecting a pair of speakers to the same amp, residual noise will be 30-40dB higher in the headphones. Because of this, many early amps with headphone jacks simply place a pair of build-out resistors at the amp output and drove headphones through them creating a sort of pad. However, that also significantly raises the output impedance seen by the headphones, so what's needed is a real attenuator, something like a 50 ohm resistor and a 1 ohm resistor, taking the headphone output across the 1 ohm. Seems like a lot of trouble to go through when using an actual amplifier designed to drive headphones would be much better from the standpoint of noise, gain control, and output impedance.

I use DIY headphone adapters connected to the B-speaker terminals of my amp to drive my headphones. Cross-feed (based on Linkwitz-Cmoy circuit) is incorporated. I started with a 33 Ω / 1 Ω voltage divider for effective output impedance of 1 Ω. This means about 30 dB attenuation. Then I added a 47 Ω resistor in series with the 33 Ω resistor with a switch to sort circuit either of the larger resistors or nothing. So, I had voltage divider options 33 Ω / 1 Ω, 47 Ω / 1 Ω and 80 Ω / 1 Ω. In the cross-feed circuit this translates into 3 different cross-feed levels between about -1 dB and -8 dB. This headphone adapter worked really well, but I wanted more cross-feed levels so I made a new improved version with 6 cross-feed levels.

Last year I studied the effect of output impedance on headphones. Too large output impedance causes curvy frequency response, inadequate damping and distortion. I analyzed these cases mathematically and found out that it different headphone models require different output impedances. My Sennheiser HD 598 requires very low (Rout < 5 Ω) output impedance for flat, adequately damped and distortion-free sound. For my older Sennheiser HD 580 the requirement is much more relaxed (Rout < 88 Ω). The old ”1/8” -rule gives 6.25 Ω for HD 598 and 37.5 Ω for HD 580, so it works pretty well actually. More sensitive headphones require smaller output impedance, because the sensitivity is achieved decreasing mechanical damping, which means more electric damping is needed. Headphones with curvy impedance response need smaller output impedances. It seems that "1/20" -rule works for ALL headphones in the world.

This year I build a "wide cross-feed" headphone adapter with 120 Ω / 2.2 Ω voltage divider (-35 dB) and only one -3 dB cross-feed level. The Linkwitz-Cmoy circuit is modified so, that the cut-off frequency is about 300 Hz (instead of the normal 700-800 Hz) and the time difference at low frequencies about 0.6 ms (instead normal 0.2-0.25 ms). The sound image is wide instead of deep, but it works nicely and I like it. The constant -3 dB cross-feed level works well with almost anything (that was my theory and it seems to work).