Hifiman HE400i Review (planar headphone)

[solderdude]

The reason the HD800S has the highest soundstage rating is because Sam used the HD800S results as reference because he liked the HD800S best for that aspect. (I had talked to Sam about this).
As Sam needed something to rate in numbers he went for that specific measurement as he can put a number on those measurements.

 

[the_brunx]

Many people who like AKG K701 also like the older AKG K500 and K501 Just the same if not more.

 

[Robbo99999]

The reason the HD800S has the highest soundstage rating is because Sam used the HD800S results as reference because he liked the HD800S best for that aspect. (I had talked to Sam about this).
As Sam needed something to rate in numbers he went for that specific measurement as he can put a number on those measurements.

Ah, I see, very good to know. I suppose that doesn't help the cause of finding out if the RTings rating for soundstage is accurate or not.....it would require quite a lot of subjective testing of different headphones. If it is accurate though, then it will be good news for my HE4XX purchase, but I'll probably reply in this thread with my impressions of that headphone (as it's quite similar to the 400i)....they'll be fighting with my K702 for my final headphone. (Well actually, if there's significant improvement in characterisation/measuring of soundstage in headphones in the near or far future, then it might not be my last headphone & I'd use that resource to choose a superior one, but for the forseeable future it will be my last one).

 

[cyph3r]

Great to see headphone review. Any plans for in-ears too? Cool would also be a comparison of noise cancelling headphone performance...

 

[solderdude]

I suppose that doesn't help the cause of finding out if the RTings rating for soundstage is accurate or not.

I have difficulties with determining 'soundstage' being wide or not. Most HP's are mediocre for me, a few stand out to me but maybe not for others.
It all has to do with one's brain, pinna angle, headphone itself etc.
To me there is no relation with Sam's idea of measuring and the ones I heard. The dip at 10kHz is simply in the test rig and depends on several factors. If that were really the issue then EQ'ing out 10kHz would lead to better stereo image.
The fun part is that the HD800 has a peak there which you cannot see because the Pinna lowers it. Lowering that 10kHz peak to lower levels removes some of the magic so it does have some effect.

Headphones that work for me are (filtered)HD560S, (EQ'd)HD800, (modified)X2HR. Those manage to deliver a steady and pinpointed stereo image with good instrument separation.
The HD800 also works well with simple crossfeed which 'stabilizes' the stereo image. Makes it less wide but equally accurate in pinpointing instruments. The HD560S and X2HR don't reach HD800 quality. The HE400i's were too 'smoothed'. I like a bit more forward and open/dynamic sound.

 

[Robbo99999]

I have difficulties with determining 'soundstage' being wide or not. Most HP's are mediocre for me, a few stand out to me but maybe not for others.
It all has to do with one's brain, pinna angle, headphone itself etc.
To me there is no relation with Sam's idea of measuring and the ones I heard. The dip at 10kHz is simply in the test rig and depends on several factors. If that were really the issue then EQ'ing out 10kHz would lead to better stereo image.
The fun part is that the HD800 has a peak there which you cannot see because the Pinna lowers it. Lowering that 10kHz peak to lower levels removes some of the magic so it does have some effect.

Headphones that work for me are (filtered)HD560S, (EQ'd)HD800, (modified)X2HR. Those manage to deliver a steady and pinpointed stereo image with good instrument separation.
The HD800 also works well with simple crossfeed which 'stabilizes' the stereo image. Makes it less wide but equally accurate in pinpointing instruments. The HD560S and X2HR don't reach HD800 quality. The HE400i's were too 'smoothed'. I like a bit more forward and open/dynamic sound.

Yes, I remember you saying months ago that your struggle with headphones sounding "out of your head", etc re soundstage, that's probably a personal HRTF thing, and maybe a psychoacoustic brain adaption thing (perhaps). I've got a pretty good imagination and have used headphones for 3D fps gaming using Virtual 7.1 surround processing so I'm always trying to use headphones to place sounds in different physical spaces around me, which I do to great effect when gaming, so that might be the side of the "psychoacoustic brain adaption" thing I was talking about. Might be a number of reasons why, but I'm sure personal HRTF comes into play.....although so far only the K702 has been able to do it properly for me.

Yes, so you can't verify that the RTings soundstage rating is accurate, yeah I wouldn't expect you or anyone to be able to....it would mean testing a lot of headphones and I'd imagine maybe having them EQ'd to the same curve to start seeing which headphones are truly able to provide the best soundstage. Soundstage is quite an elusive property, I'm happy to test the HE4XX when it arrives & if it's no good then K702 for life until headphone science truly identifies measurable soundstage parameters for ranking headphones for that criteria accurately.

 

[F1308]

I have calibrated my headphone fixture with 94 dBSPL test tones and decided that I use the same for headphone measurements. However, instead of using 1 kHz tone I have opted for 425 Hz. This seems to better match research data.

A very good and reasonable change indeed.
From 1000 to 425...Great.

On a piano using A4 (key number 49 on a 88 standard keyboard) tuned to 440 Hz (standard nowadays, Versailles used 392, Barroque 415 and Venecian 465, and some bag pipes around reaching 485), those 425 Hz are located close to G#4 (415.13 Hz) so it will be out of tune by 40 cents from G#4 and 60 cents from A4.
Anyway, we can match and get perfect tuning setting a key at 425 (G#4) by making A4 playing 450.27.
Other way might very well be setting A4 directly to 425, but as I said before, the tuning fork went from 392 to 465 and back to 440 for a reason: brightness. Lower tuning forks calls for sadness or solemnity, higher ones for happiness and enjoyment...but it seemed 465 was too much !!!!
So if G#4 @ 425 then A4 @450.27, and that seems reasonable.

Now we go and play using a 61 keyboard, or a 76 or even 88 keys, and D1-A6 for most of the time means a frequency range going from some 37.56 to 1801.08 Hz.
There are so many loudspeakers easily achieving that low...!!!!

 

[phoenixsong]

Then HD650 and HD600 are one of the worst.
You need something with considerable angled drivers.

I was wondering about those headphones without angled drivers but have rotary cups

 

[rodtor]

Headphones that work for me are (filtered)HD560S, (EQ'd)HD800, (modified)X2HR. Those manage to deliver a steady and pinpointed stereo image with good instrument separation.
The HD800 also works well with simple crossfeed which 'stabilizes' the stereo image. Makes it less wide but equally accurate in pinpointing instruments. The HD560S and X2HR don't reach HD800 quality. The HE400i's were too 'smoothed'. I like a bit more forward and open/dynamic sound.

Thanks for this. I have been thinking about getting the X2HR phones. I remember your favourable discussion of them on your site, but also noted your concern about quality control with them, that they seem inconsistent pair-to-pair. This is one reason why I am still hesitating about them.

 

[amirm]

Great to see headphone review. Any plans for in-ears too?

Yes. The fixture can do that but I like to get more experience with testing them before doing any formal reviews. Getting them to fit may be challenging. I already have some candidates here.

 

[DualTriode]

We know its there but 2nd harmonics are so low in level that they are masked.
Besides my measurements are only intended to look for unusually high amounts of distortion. In reality distortion is lower than I can measure.
Jude's distortion measurements are closer to reality as he measures in a special chamber.
It's all about audibilty. A 120 SINAD DAC and 100 SINAD DAC will both have no audible distortion at normal listening levels listening to music.

Besides... Harman research already told us distortion is not a sound quality determining factor.
I would add unless it reaches >1% in the mids.
With this driver distortion was audible... had to fix the driver (planar so easy to fix)

and this one also did not sound that great, the owner liked it a lot for piano

Hello All,

@soulderdude,

Thanks for making my point in your pedantic way. Ambient noise is important in the measurement of distortion.

For distortion to be separated from a noise floor of -80dBs the Harmonic needs to be above 0.01%.

Currently using the 45CA-9, I am measuring a noise floor near -90dB. 0.0032%, distortion harmonics just start to peek above the noise. For low frequencies the noise floor climbs steeply.

See the following example links. Masking can make over double digit % harmonic distortion auditory benign.



Geedes Ph.D. and Lee 2003 AES Paper

Auditory Perception of Nonlinear Distortion - Theory

http://www.gedlee.com/Papers/Distortion_AES_I.pdf

Geddes Ph.D. Text Book Chapter 10

Distortion

http://www.gedlee.com/downloads/AT/Chapter_10.pdf



@amirm,

Thanks for your insights.

Now that I look at it more and give it some thought; perhaps plotting the headphone frequency response and 2nd & 3rd Harmonics works well enough. If the THD is below some considered level of 2 or 3 % the 2nd and 3rd Harmonics will not be audible and the higher level harmonics will be below -90dB’s and well into the measurement noise floor.

Thanks DT

 

[bobbooo]

Nope. I register, page reloads, go to download and asks to register again.

I’m digitizing the ~30 in the paper, and I’m getting close to the scores in the paper that I digitized (I have done 15 of them thus far). So, that coupled with the fact that I can’t see where I could go wrong, leads me wondering how I am getting a few points off when digitizing Oratory’s HD800S & HE400i...

_____
It’s a little trickier with Amir’s data, as I have to use an interpolated target curve to generate the error curve, whereas Oratory and the paper both give the error curves. @amirm, could you export the error curve for the HE400i and/or HD800S, just so that I can make sure my interpolated curve isn’t causing issues?

I'll DM you the Listen Inc. preference formula spreadsheets then.

 

[bobbooo]

Ah, fair do's then on the lack of bass plate on your HE400i measured headphone. (So that version wouldn't measure any better than the one you measured, because that is the one you measured!)

I wouldn't be certain of this. That photo of this review unit with the grille on differs somewhat to the no-backplate unit measured by Rtings below (and others I've seen which look like the latter):

So I don't think it's 100% conclusive yet which version this review is of. It's not that easy to tell if there's a backplate just from a photo with the grille on really - a photo of the headphone with the grille off would be needed to be sure, which is fairly easy to do:

Now, even if this review unit doesn't have the backplate, there still remains an apparent difference in frequency response between it and most other HE400i measurements. This difference includes a dip at around 500/600 Hz which is not present in this review unit, but is in most others:

In addition to Oratory's measurements above, other measurements showing a similar dip include Crinacle's, Rtings', Dekoni's, solderdude's (2.5mm connector) etc.

This review unit also exhibits pronounced wiggles around 200-300 Hz which others don't (or have less of at least). The only exception I've seen are solderdude's measurements from 2017 of the older HE400i unit with SMC connector, which like this review unit, lacks the 600 Hz dip and exhibits a similar wiggle around 300 Hz. (2.5mm connectors are not a guarantee of no backplate by the way - there have been reports of some 2.5mm connector units with backplates. I haven't seen any SMC connector units without backplates however.)

So it could be possible that there were other driver changes (or housing or pad etc. - Hifiman are known for their stealth revisions) not overtly noticeable that have caused this stratification into what seems like two sub-versions of the (old, not including the 2020 remake) HE400i in terms of frequency response, and as the majority of other measurements do not match the ones in this review that well, this could mean the performance here does not accurately represent most users' units and so experience of their headphone (including distortion, considering solderdude's measurements show a significant THD difference between the two versions he measured).

There's nothing wrong with putting forward testable hypotheses based on suggestive evidence (that's what science is all about), but extrapolating this individual (possibly less common) HE400i unit's performance even further to the HE4XX based merely on the same initial model number and looking similar? Talk about random conjecture. Especially when the actual measurement data contradicts such an extrapolation, showing they are in fact very different headphones, in the form of not only frequency response (and the significant corresponding preference rating difference of 80/100 for the 400i, and 88 for the 4XX as calculated by Oratory), but the latter's low distortion (measured @90dB, which Jude of Head-Fi has told me was taken on a GRAS 45BB-12 which is more sensitive to low-level distortion and noise, in a purpose-built isolation chamber), and comprehensive differing measurements by the brilliant, legendary Tyll of Innerfidelity, including electrical and isolation:

HE400i / HE4XX
Voltage required to reach 90dB SPL: 0.162 / 0.271 Vrms
Impedance @ 1kHz: 45 / 49 ohms
Power Needed for 90dB SPL: 0.59 / 1.51 mW
Broadband Isolation in dB (100Hz to 10kHz): -2 / -5 dBr

 

[solderdude]

Thanks for this. I have been thinking about getting the X2HR phones. I remember your favourable discussion of them on your site, but also noted your concern about quality control with them, that they seem inconsistent pair-to-pair. This is one reason why I am still hesitating about them.

I have been getting reports from inconsistent results with the HD560S as well.
The main downside of X2HR is (so I am told) that pads are not available so one would have to use (sound altering) aftermarket pads as replacement.

 

[Robbo99999]

I wouldn't be certain of this. That photo of this review unit with the grille on differs somewhat to the no-backplate unit measured by Rtings below (and others I've seen which look like the latter):

View attachment 100623

So I don't think it's 100% conclusive yet which version this review is of. It's not that easy to tell if there's a backplate just from a photo with the grille on really - a photo of the headphone with the grille off would be needed to be sure, which is fairly easy to do:

Now, even if this review unit doesn't have the backplate, there still remains an apparent difference in frequency response between it and most other HE400i measurements. This difference includes a dip at around 500/600 Hz which is not present in this review unit, but is in most others:

View attachment 100624

In addition to Oratory's measurements above, other measurements showing a similar dip include Crinacle's, Rtings', Dekoni's, solderdude's (2.5mm connector) etc.

This review unit also exhibits pronounced wiggles around 200-300 Hz which others don't (or have less of at least). The only exception I've seen are solderdude's measurements from 2017 of the older HE400i unit with SMC connector, which like this review unit, lacks the 600 Hz dip and exhibits a similar wiggle around 300 Hz. (2.5mm connectors is not a guarantee of no backplate by the way - there have been reports of some 2.5mm connector units with backplates. I haven't seen any SMC connector units without backplates however.)

So it could be possible that there were other driver changes (or housing or pad etc. - Hifiman are known for their stealth revisions) not overtly noticeable that have caused this stratification into what seems like two sub-versions of the (old, not including the 2020 remake) HE400i in terms of frequency response, and as the majority of other measurements do not match the ones in this review that well, this could mean the performance (including distortion) here does not accurately represent most users' units and so experience of their headphone.

There's nothing wrong with putting forward testable hypotheses based on suggestive evidence (that's what science is all about), but extrapolating this individual (possibly less common) HE400i unit's performance even further to the HE4XX based merely on the same initial model number and a photo with the grille off? Talk about random conjecture. Especially when the actual measurement data contradicts such an extrapolation, showing they are in fact very different headphones, in the form of not only frequency response, but the HE4XX's low distortion (measured @90dB, which Jude of Head-Fi has told me was taken on a GRAS 45BB-12 which is more sensitive to low-level distortion and noise, in a purpose-built isolation chamber), and differing electrical and isolation measurements by the brilliant, legendary Tyll of Innerfidelity:

HE400i / HE4XX
Voltage required to reach 90dB SPL: 0.162 / 0.271 Vrms
Impedance @ 1kHz: 45 / 49 ohms
Power Needed for 90dB SPL: 0.59 / 1.51 mW
Broadband Isolation in dB (100Hz to 10kHz): -2 / -5 dBr

Hmm, ok, I was quite sure Amir showed a torn down photo of his headphone that clearly showed there was no backplate? That wouldn't explain away your subversion theory though. And yes, HE4XX is different again....so can't really extrapolate with certainty. I remain confident about my HE4XX purchase which I'm waiting to arrive, the only real question is gonna be soundstage in my eyes....which we talked about earlier.

 

[MZKM]

I suspect it is 1/3 octave or something like that.

I couldn’t find it in the papers, but Oratory and the Listen Inc document say to use 1/12. Speakers is 1/20, so maybe they had to go smoother due to treble inconsistency.
I’m almost done digitizing all the data (which gets like 20-30 PPO in the last octave), so I’ll see what Sheets comes up with as a formula to see how far off it is with this higher treble resolution.

 

[bobbooo]

Hmm, ok, I was quite sure Amir showed a torn down photo of his headphone that clearly showed there was no backplate? That wouldn't explain away your subversion theory though. And yes, HE4XX is different again....so can't really extrapolate with certainty. I remain confident about my HE4XX purchase which I'm waiting to arrive, the only real question is gonna be soundstage in my eyes....which we talked about earlier.

That torn down photo was taken from this user's post on Head-Fi.

 

[bobbooo]

I couldn’t find it in the papers, but Oratory and the Listen Inc document say to use 1/12. Speakers is 1/20, so maybe they had to go smoother due to treble inconsistency.

And/or maybe they found that the extra smoothing combatted the formula unfairly penalising very small high-Q frequency response variations across the whole range, which aren't audible.

 

[MZKM]

And/or maybe they found that the extra smoothing combatted unfairly penalising very small high-Q frequency response variations across the whole range, which aren't audible.

Could be.
I just finished comparing my digitized versions:

​

So, almost a constant decrease in scores, no doubt to the finer resolution in the treble.

I also get a formula: 119 - 10.2(SD) - 22.2(AS)
So less focus on SD and more focus on AS. Again, the finer resolution in the treble is probably why. Testing it on the data though leads to less accuracy in the >6/10 scores.
But since Amir’s data is fine (and I don’t want to do interpolation every single time), I’ll work with this finer data (even though it’s not the same as Amir’s), and see what separating it into bass/mids/treble (maybe 200Hz/1kHz/8kHz) gets me.

 

[BogdanR]

Yes. The fixture can do that but I like to get more experience with testing them before doing any formal reviews. Getting them to fit may be challenging. I already have some candidates here.

The proper fit for In Ears is absolutely paramount. I’ve just got a pair of Final E4000 and got to play with them extensively. Their frequency response varies wildly with insertion depth for instance. There’s an ideal depth I found and anything shallower or deeper causes either massive high frequency ”detail” loss or bass “reverb”. All the included tips negatively affected higher frequencies, fortunately I have foam tips which considerably alleviate these issues. Still, getting the proper fit is not easy.