New Headphone Amplifier Measurement Page!

I'm in the midst of doing a review of the HeadAmp GS-X mk2 and ran it through my amplifier measurement routine. Then I remembered I had a couple of other amps lying in wait and figured I might as well measure them. Well, with a dozen graphs now done including the Big Sound 2015 amplifiers I figured it was time to start an amplifier measurement page.

I know it's not many measurements. I'll go bug the folks at HeadRoom to borrow a bunch of amps for measurement over the weekend. But it's a start! After all, my headphone measurement page started off with just a handful and there is now well over 800 headphones measured. I'll also be willing to accept amps from readers for measurement...but that will take a month or so to get rolling.

I'll also note that I am currently just measuring single-ended, solid-state amplifiers. Balanced amps, tube amps, and digital input measurements of DAC/amps will come along over time as I rewrite the measurement routine for them.

For now, you can view the amp measurement page here, and you'll also find it under the "Resources" tab at the top of the page. Of course, I'll also maintain a .pdf of all the amps measured, which you'll find here.

Lastly, a little word of caution: To my eyes, unlike headphones, it's almost impossible to have any idea what an amplifier will sound like from looking at the measurements. I think amplifier measurements are more a matter of, "Is it broken or not." Or, possibly, they may give you an idea of the quality of engineering employed. Be very careful however, sometimes distortion is a good thing. Tube amps have more distortion than transistor amps by a long shot, but they can sound great!


inventionlws's picture

Why not include the impulse response and step response, I think they'd be very good indicators of transient response, just as the case with headphones.

zobel's picture

Now we can actually hear gravity waves! Check this,

It is simply amazing.

castleofargh's picture

well the THD graph is super cool as it gives an idea of the actually usable power. given how unclear some manufacturers can be sometimes with that, not giving enough loads, not telling if the value is per channel etc etc... I believe in the long rung that graph will be of great value here.

the noise floor of course can also be good to know depending on our headphones. (I'm a hiss maniac)

and FR shouldn't matter too often, but it can still be interesting to find out how much of a coloration is due to FR and how much comes from distortions.

fr the rest, I agree tat it's hard to imagine translating the values into how it will sound. but I still happily take them all. ^_^

thanks for that.
a little something on the protocols maybe? a "how it's done" kind of post.

noob question: for the noise, what is the volt equivalent of a given value? and can I then from that extrapolate with the sensitivity of my headphone and how many volt I'll need to reach 90db, how far down the noise will be below my music? or am I imagining something too simple and the noise will rise when playing music?

detlev24's picture

unfortunately, it matters too often and frequency response is one of the 4 categories which affect fidelity of any audio gear. the other aspects are noise, distortion, and time-based errors (to quote ethan winer).

@Tyll: as always, thank you for your effort! this is really great so have as a reference.

detlev24's picture

*to* have as a reference.

tony's picture

Once again our Tyll is proving himself to be the Go-To person for understanding all things Headphone.

Twenty years from now we'll look back at all of Tyll's body of work and say he's been the keel under the Ship of Headphones.

If this were the English language, Tyll would be the creator of our Oxford English Dictionary.

I've purchased numerous headphone devices, those Tyll has measured, described and recommended have been consistent in their excellence.

Tony on the Campaign Trail with 'the Bern"

tony's picture

There is something significant about Amplifier Measurements.

My lovely Sennhiesers seem to require only .2 Milliwatts to achieve powerful listening levels.

Headphone Amplifiers seem far more powerful than useful, my own superb Asgard 2 is said to output 1 Watt.

So then, the very modest power output ( 35 milliwatts ) of the famed Mojo from Chord has ample power for most headphones.

Why then are we needing the huge power of those amps that deliver 6,8 and 10 Watts of output power?

Is it simply to satisfy the Hifiman products?

Tony in Michigan

Bob Katz's picture

I have an (as yet) unproven contention: That amplifier headroom beyond what is needed translates to a more open, impacting, and dynamic sound. It's hard to make such a comparison because how do you keep all other things equal? But at this moment I like to see at least 2 watts of power in my headphone amps into 32 ohms at least. Again I have no way of proving if this is really a good thing because Tony rightfully points out that we don't need that kind of power to deliver a good, clean sound.

One indication that my idea is full of crap could be my reaction to the Astell and Kearn portable player. It seemed to have all the clarity, impact and headroom I would want, but obviously, being battery powered, I doubt it delivers "watts worth" of power.

Another good test of the contention would be to take an amplifier and compare its balanced versus unbalanced performance. As far as I'm concerned, balanced with headphones is a stupid concept EXCEPT that it doubles the output capability, by 6 dB. So if you can make an absolutely matched comparison in level between an amplifier driven balanced versus unbalanced you can examine my contention about headroom.

castleofargh's picture

but balanced output may change a all lot more than just twice the max voltage. not that most specs like crosstalk, disto, and impedance will necessarily have an audible impact, but they also have really no reason to be less significant than max power.

wouldn't changing the gain come closer to a power test?
I really can't see why more power would do anything(all other things being equal) it's not like the load will "use" more power at a given loudness anyway. therefore it has nothing to do with driving the headphone, or the actual power used. so what could be the perk for the amp itself?

tony's picture

Didn't we use all the Power (we could control) to actuate our loudspeaker drivers?, I can recall advancing to Mono Amps for my better systems, even tri-amped systems.

Meridian came along and used a 35 watt Amp for each individual driver in their Active Loudspeaker systems, opening my eyes to engineered sound system designs. The Meridian designs ( that I owned & loved ) : the M3s, M2s and M10s all had incredible headroom.

Today, Genelec use (seemingly) modest Amplifier Power to drive their individual drivers yielding superb performance/$ values.

Still, all these loudspeakers have scads of Amplifier Power which seems to taint my thinking about owning capable Amplifiers, it's easy to maintain a similar headphone amplifier philosophy.

Which I've been happy to do with my successful Asgard2, until I discovered our own 'measurement Guru' : Stereophile's JA reviewing the 35 milliwatt Chord Mojo. Somehow the "modestly" powered Mojo successfully powers a wide range of Audiophile headphones! I went further to discover "owner" reports confirming JAs findings.

35 Milliwatts?, how can this be?

I went back to "our" Tyll's vast collection of headphone measurements to grasp the bedrock of important understandings : my Sennheiser's require only .17 Milliwatts to sing their full voice!

How did I miss this?, I misunderstood, I seemed to be reading and understanding the .17 as Watts not Milliwatts! My poor vision needed me to expand Tylls graphs to very large sizes in order to clearly read ( the finely printed ) milliwatts . Oh my !

Now, I calculate the diminutive 35 Milliwatts to be at least 150x more powerful than my Sennheisers require, which is scads of headroom ( compared to any loudspeaker system I've ever seen or heard of ) .

For me, A whole new Light is shining on headphone Amplifiers. I'm hoping our Tyll ends up clearing away the prevailing Amplifier misconceptions we all seem to be laboring under.

I may be able to power my headphones with Button Batteries ( as the Hearing Aid people seem able to accomplish ).

Chord may be onto something.

Hm, Meridian, Genelec and now Chord.

Tony in Michigan

ps. (0 degrees F), pining for Florida

Phoniac's picture

I don't know which Senn, but Tyll measures 90 dB SPL for said 0.17 mW. That is not 'sing their full voice'. That would be at 105 dB SPL. Now 15 dB more level means a factor of 31.62 for power, your 0.17 mW turn into 5.38 mW. Still ridiculous, I know.

tony's picture

Thank you for mentioning and confirming all this, I was amazed to finally read the fine print.

I'm suspecting the Dac output + a Volume Control ( as in the Emotiva ) might be entirely adequate.

This knowledge changes my view of Headphone Amplification. Am I missing something?

Now I'm getting the idea that the Mojo doesn't have an output amplifier. I'll have to ask the Chord Guy Franks who is a regular on Headfi forums.

Tony in freeeeeezing Michigan

Phoniac's picture

Well, yes, you are missing something. People's needs are different, very different. My SPL is more like 115 dB SPL and up, especially with sub-bass amplified. That needs not only a headphone with very low distortion as well as intermodulation distortion (a topic completely ignored on Innerfidelity), but also a lot more power. Suddenly amps with 1 Watt for 32 Ohms start to make sense...

zobel's picture

Using the same amp, balanced vs unbalanced. Do have an amp you could perform this test with Bob? You would need two of those amps though, so quick, blind A/B switches could be made between them. What headphones would you use to do this test? I think headroom might play a role, but again, how much headroom?, and through what cans?, are important factors to consider. Go for it! We would love to hear your results.

knghtwhosaysni's picture

Thanks for this :) Two requests for future measurements:
1. Schiit Magni, seems to be the most recommended amp along with the O2, would be curious to see if there's a significant diff in the measurements
2. Benchmark's HPA2, seems to be the only company as serious about measurements as NwAvGuy but we only have their measurements to go off of and they mostly focus on the DAC part, not the headphone amplifier

Phoniac's picture

As far as I could see in the amps 'leaflet' the DAC2 will blow away everything measured so far. Maybe that's why it isn't in there...

Phoniac's picture

I measn HPA2.

Grumpy's picture

How do I read the Frequency response graph? Some seem far from "flat" within the audible range. (The Violectric and Antelope Zodiac for example)

Bob Katz's picture

These two amp's frequency response graphs to correlate with what I heard at Big Sound.... artificially warm and veiled sound quality. If an amp is not flat to 20 kHz within 0.2 to 0.3 dB I would suspect a colored presentation. Also need to check into various loads to see if it affects that frequency response.

Another good test is called "dynamic headroom" but I don't know enough how to engineer such a test... it was originated I believe by Richard Heyser.

Grumpy's picture

But the Violectric graph seems to suggest the low end is rolled off, not the high end, wouldn't that create the opposite of a "warm" sound?
I'm just wondering how to read the graphs: Is the Violectric for example really down by 2dB at 20Hz, or doesn't it work like that?
It just seems like a pretty serious deviation from "flat" (And from violectric's specs)

Bob Katz's picture

Yes, It's actually the Antelope that has a severe high freq. rolloff. "Warmer" sound is not always due to HF rolloff... it can be due to harmonic distortion (largely second), soft power supply regulation, etc. It's harder to predict from the Violectric's response how it might sound. But both the Antelope and violectric sounded overly "tubey" (in a bad way) to my ears at Big Sound. I have no objections to tubes done right, by the way. And I believe both of these are solid state anyway.

Bob Katz's picture

I found a problem in my Burson audio Soloist below about 100 Hz at rated power by running a THD versus frequency check at different powers.

Rillion's picture

I believe the vertical axis on the DIM/TIM plot is mislabeled -- looks like it should be dB instead of percentages. It is probably a good a idea to fix this before dozens of amps get measured.

m8o's picture

Mr. Hertsens,

If McIntosh is not forthcoming and proactive in putting an MHA-100 in your hands, I am willing to send you mine for your analysis and review.

You have my email address in my profile. Drop me a line if you would like to pursue this.

At best, perhaps a McIntosh rep will read this and the idea a customer needs to offer up their own personal unit will embarrass them enough to spur them to action to send you an evaluation unit.

logscool's picture

I think it will be very interesting to see how these measurements and trends start to correlate to subjective sound quality impressions of measurements.

Just going off what I can see from the measurements the O2 and the GS-X mk2 look to be some of the best performers measurement wise but owning an O2 I can say that while it is certainly a great little amp it can definitely be bested by other amps with some headphones that are not as costly as the GS-X mk2 and probably don't measure nearly as nicely. Such as the cavalli liquid carbon, garage 1217 project ember, and finally the centrance dacport. The ultra-low noise floor of the O2 is really great for sensitive IEMs though.

Chiumeister's picture

If so, that's amazing for the price.

tiger's picture

Hi Tyll,

Thanks for your hardwork, the community really appreciate it.

However, mind checking on the Auralic Taurus mk2 measurements? The crosstalk looks outrageous!