Massdrop X KOSS ESP/95X Electrostatic Headphone Review

[Blake Klondike]

This is review and detailed measurements of Drop's X KOSS ESP/95X electrostatic headphone. It is on kind loan from a member and when it was available, cost US $500. Not sure what is different with it from the original KOSS model which I think cost $1000.

Typical of most electrostatic headphones, there is a required amplifier/driver ("energizer"):

View attachment 88822

The headband separates from the earphones using a clever and simple button on each side. The rest is not so impressive as the plastic is what $10 toys would be made out of. Same applies to the amplifier which looks metal in the picture, but is very cheaply made plastic. The plastic has one advantage: the headphones are super light. Wearing them was comfortable for the 2 hours or so that I used them for testing.

The headband friction fit is too loose so the position of the headphones does not stay the same. Pull them on and off and they change.

The retainers on each side of the earphones creeks as you move around.

The most major flaw is the volume control. They copied the one from Stax where you have the left and right volume on the same shaft. On Stax, you adjust the two but then when you turn the very large knob, both channels adjust at the same time. Not so here. It is darn near impossibly to adjust the volume and not have one channel change more than the other. It was so frustrating that that I gave up on using it and set it to near max and used my DAC's volume control.

Massdrop X KOSS ESP/95X Measurements
I measured ESP/95X on GRAS 45CA standardized headphone measurement gear which is (still) on loan from the company. The drive and analysis was the Audio Precision APx555. Let's start with our new dashboard:

View attachment 88823

Distortion is quite low for low frequency source tone of 40 Hz (peak amplitude in sampling of music). We can see second harmonic at 80 Hz to the tune of -65 dB which is quite good.

Most important measurement though is frequency response:

View attachment 88824

The dashed blue line is most preferred frequency response for a headphone. Between 200 and 3 kHz, we have decent agreement. Below that we have significant deficiency which is the case with many headphones. We do have drooping in lower treble as indicated by the cursor lines near 5 kHz. There is also some extra energy around 1 kHz. Lack of bass and highs would indicate a sound that would be rather flat and unexciting.

While I have become comfortable with the above comparison of measurement versus expected, some people may want to see the two substracted from each other and look at the deviation. A flat line here then indicates 100% compliance with the preference curve:

View attachment 88825

This looks a lot worse than many other measurements out there. The reason is that I use 20 dB span and others use many multiples of this. Increase the range vertically enough and you can make any frequency response look flat!

Anyway, we see the shortfall in bass from 200 Hz down. There is that strange shelving which may be some kind of electronic filtering in the amplifier.

As usual, ignore stuff at or above 10 kHz.

Somehow I lost my IMD vs level test from the last review. Until I recreate it, let me treat you to other distortion graphs:

View attachment 88826

For some reason this AP graph only shows second and third harmonic. Note that this is pure THD and has no noise in it.

Comparison to Stax SR-303 Headphones
I thought it would be good to perform a quick comparison to one of my Stax headphones, the SR-303 driven by their lower end SR-313 amplifier:

View attachment 88827

This response has a much better adherence to our preference curve. We have matching bass down to 50 Hz compared to 200 Hz in ESP/95X. And then slopes down gradually, kind of like a speaker would. This looks like some kind of high pass filter.

Massdrop X KOSS ESP/95X Listening Tests
Out of the box the sound was decent. I think having that first peak in the 3 kHz help to not have a hollow sound. That said, lack of bass is obvious, and there is little sparkle. I heard no "speed" people talk about. Once again I think people take the lack of bass and equate that with 'speed." The large earphones do provide a taller presentation which as a fun exercise, you can adjust up and down as you move the cups.

I decided to use EQ to improve things and this was more time consuming than I expected. Here is where I landed:

View attachment 88829

I had to build a custom bass boost for it as you see above. It responded very well to this.

There was some extra energy around 1 kHz. I went back and forth on this a fair bit but at the end, I decided to take it out using that teal Band 1 filter.

This last bit was the most complicated. I tried to boost what was missing around 5 kHz, but not let the overshoot at 6 kHz make it too bright. I think a more proper automatic filter generation may work better here than my eyeballing it.

When it was all said and done, the ESP 95/X went from rather unexciting and dull to a "hifi speaker." We now had bass, more highs and less midrange emphasis. Indeed I enjoyed it for a couple of hours listening to a lot of music.

Then out of the corner of my eye I saw my Stax SR-303 and decided to listen to that. The tonality of the Stax out of the box was way more pleasant than the ESP/95X. The Stax material was also in a completely different class. This prompted me to then measure the Stax as you saw in the previous section, confirming what I was hearing.

There was a difference though: the ESP/95X could barely get loud enough for my music selection. The Stax misses that mark. Turn it up beyond a certain spot and the drivers start to crackle.

Conclusions
I don't know why people are so fascinated with how some speaker/headphone is made. What matters are the results. If full frequency response is not delivered faithfully, I don't care if you bring a live orchestra to my room. I won't like it. We have that problem here to some extent. The frequency response deficiencies are there and are significant but are not show stoppers. Correction is mandatory unless you want something plain sounding with little dynamics, bass and sparkle in the highs.

The ESP95/X didn't light up any magic for me that people talk about other than what the large size of the driver brings.

If you are itching to hear what electrostatic headphones are about, you can opt opt for these and I won't chase you out of town. Otherwise, I suggest looking for a headphone with better response. So while I can't recommend the Massdrop X KOSS ESP/95X, I am not opposed to them either in a sea of other headphones, many of which are worse than it.

-----------
As always, questions, comments, recommendations, etc. are welcome.

Appreciate any kind donations using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/

Is there an electro-static you've found that does measure well?

 

[bobbooo]

you can get the Koss esp 95x quite easily near the harman target with eq. eq setting

Yep, and they sound fantastic once EQed with Oratory's settings (96/100 preference rating), and still quite listenable without EQ (72/100), especially mids-centric acoustic tracks. They're actually going for $450 at the moment, and regularly go down to $400, which isn't too unreasonable considering that includes the amp/energizer. (I actually managed to get them for just $270 on a brief clear-out sale.) With low distortion comparable to the previously measured Audeze LCD-X, yet a much more palatable stock frequency response, and without the Audeze's deep cancellation notch which limits its EQ-ability, plus lower weight and better comfort, a much lower price which includes an amp, and a lifetime warranty, the ESP/95X seems to be a decent deal in comparison really.

 

[PeteL]

What is with "the most preferred response for headphones" and criticizing headphones for flat frequency response? I want ruler flat response, period. Just because a bunch of people like inaccurate bass response? Who cares? Come on. Accuracy is paramount! What people prefer is irrelevant. This reminds me on another Tooleism - that early speaker room reflection is good because people prefer it because it enhances "space". The fact that boosted bass and early reflections color the sound is fine for some, whatever floats your boat. Give me the most accurate representation of the recording please, not what people "prefer". I can EQ it myself, or build in some switchable optional EQ if you want.

I prefer my headphone neutral. A set of headphone with flat response would be quite colored. without eq. Flat is the target for speakers, not headphones.

 

[amirm]

Is there an electro-static you've found that does measure well?

I have only recently started to measure them so don't know.

 

[Degru]

We hear bass frequencies better when the volume is high.
Thus depending on your listening volume you will require a different bass balance along with personal preferences for bass balances.
Additionally I would have thought that the upper trebble in the Harman curve for headphones (not the loudspeaker curve) is rolled off as we can hear those frequencies very well when in the ear (vs poorly when in a room due to tiny wavelengths) and high volumes with them can be very, very damaging.

Even if that were the case theres no way to reliably design a headphone that follows that treble curve on both measurement rig and real ears. It's too heavily dependent on reflections and such to tune it not by ear.

 

[Jimster480]

I always wondered what these were about. Now I know. Glad I didn't put down money for them as they seem to be mediocre.
I'm a big planar fan and I am happy with what I have at the moment.

 

[restorer-john]

It is darn near impossibly to adjust the volume and not have one channel change more than the other. It was so frustrating that that I gave up on using it and set it to near max and used my DAC's volume control.

Aren't the two channel shafts friction clutched together like proper concentric bass/treble rec level of the old days?

 

[Robbo99999]

Do we have a 'What headphone do you want tested next?' thread yet?

Sounds like Nuraphone should be added..

Ha, well on that topic I'd like to have AKG K702 tested! I hear noises that it's soundstage/imaging is pretty much as good as the Sennheiser HD800s, I tend to believe that to be possible given that I own & use the K702, it's my favourite headphone.

 

[SpaceMonkey]

Great. I suspect results can be epic.

I suspect that results will be acceptable in terms of THD (ESL headphones have low THD even with energizers used), frequency response might be tailored depending on transformers used (or built in filters) and biggest limit will be max voltage swing that these can push out.

 

[bigjacko]

Aren't the two channel shafts friction clutched together like proper concentric bass/treble rec level of the old days?

No, the two actually are not coupled, there is no friction between the two knob.

 

[Severian]

I've been interested in these for quite a while, but one thing that holds me back is wondering whether the energizer has enough power to get loud on quiet recordings after taking into account a substantial negative preamp gain to accommodate EQ.

 

[BYRTT]

.....I decided to use EQ to improve things and this was more time consuming than I expected.....

Thanks all the pro work and graphs published, prediction of EQ string for left channel on paper modeled in below animation including your new posted cool model of Harman target compensated to flat..

 

[bobbooo]

I've been interested in these for quite a while, but one thing that holds me back is wondering whether the energizer has enough power to get loud on quiet recordings after taking into account a substantial negative preamp gain to accommodate EQ.

They're plenty loud enough for me, even with the -7.4 dB preamp gain I use as per Oratory's excellent EQ settings, and my Samsung Galaxy S10e (1V max output) feeding the energizer. Any louder and they'd likely be approaching hearing-damage territory.

 

[Severian]

They're plenty loud enough for me, even with the -7.4 dB preamp gain I use as per Oratory's excellent EQ settings, and my Samsung Galaxy S10e (1V max output) feeding the energizer. Any louder and they'd likely be approaching hearing-damage territory.

That's good to know (and I have the same phone - bless the headphone jack).

My holy grail headphones are the Stax L700 but I'm never going to spend that, so I've always been eyeing these to have some electrostats on the cheap.

 

[The Jniac]

How would the Etymotic ER-4XR perform then since it's a sub bass boosted version of the ER-4SR?

Well, the problem with the ER4-XR is that while there is a bass boost, it is the result of a very broad, but fairly low amplitude peak in the lower mids. While that is probably not a huge auditory issue since most people are pretty bad at detecting broad but weak variations in frequency response, it is still probably not an accurate representation of what we hear.

The LCD4 gives huge rumbling sub bass that everyone says is intense, powerful and controlled yet the harman target would tell you it needs a 6db boost

The placebo effect is a powerful thing. There is also the fact that Audeze tend to tune their headphones to really show off their bass extension, often by making the rest of the spectrum a bit withdrawn (or just horribly mangled in the case of their closed headphones). It is notable that people praised the bass of the LCD-4 far more than they did other headphones with very similar bass extension.

 

[The Jniac]

This is not what is desired though.

I know. I was assuming that a flat-at-listening position was the response that he wanted based on his initial comment.

That said, I am a bit of outlier in that I, personally, like the idea of headphones that sound like a pair of flat-at-listening-position speakers. This is mostly because I have this rather weird thing were I would want my speakers to be flat anechoically, so that the sound that they produce in a given room would be identical to the sound one would get if they had the initial performance that was recorded performed in that room. However, I want my headphones to sound like a pair of flat-at-listening position speakers because, and please try to bear with me on this as I know it sounds crazy, with headphones I want to essentially be inside the recording.

However, I am curious about this "typical room" thing. Do you know how this might differ from an "ideal room"? Is there any sort of consensus about what would constitute an ideal room rather than a typical one?

 

[Blujackaal]

Well, the problem with the ER4-XR is that while there is a bass boost, it is the result of a very broad, but fairly low amplitude peak in the lower mids. While that is probably not a huge auditory issue since most people are pretty bad at detecting broad but weak variations in frequency response, it is still probably not an accurate representation of what we hear.

It quite noticeable if you tried the ER4SR in a A/B test.

The placebo effect is a powerful thing. There is also the fact that Audeze tend to tune their headphones to really show off their bass extension, often by making the rest of the spectrum a bit withdrawn (or just horribly mangled in the case of their closed headphones). It is notable that people praised the bass of the LCD-4 far more than they did other headphones with very similar bass extension.

Expection bias seems very strong with planar/Estat fans. Many are trying to downplay that Open back Stax's are junk for bass, LCD gimp the mids/highs to give impression of big bass. But then lash out when the ER4SR has +10.5db more at 20Hz, Then run too get the KSE1200(£1200) or mod/EQ them which is bit stupid for a £1600. When the Stax L300LTD & L700 has the less bass than any Grado headphone.

 

[Atratusnex]

I've been interested in these for quite a while, but one thing that holds me back is wondering whether the energizer has enough power to get loud on quiet recordings after taking into account a substantial negative preamp gain to accommodate EQ.

i dont get what amirm means, that this ones are to quite.... With k tone board / motu m2 i cant even nearly max them out,
im able to get far to loud with them even after preamp gain (-7,4) for harman like eq.
Its painfully loud.

 

[MadMan]

I have a very simple question.
What this Gras Harman Target is:
A. You have to use it because of the properties of the used Gras "artificial ear"?
B. Most people preferences it as the most pleasant for them.

It's called the CONSUMER preference for a reason - it's what your average bigbox store CONSUMER likes. So, for mass market manufacturers who want to sell the most units to the most people, sure it's great. Harman's own research shows that trained/experienced listeners do not have this same preference. Also, it's stupid to want all headphones to have very similar fr curves, a big draw of headphones is the ability to have more variations much more easily than with a speaker setup.

 

[bobbooo]

It's called the CONSUMER preference for a reason - it's what your average bigbox store CONSUMER likes. So, for mass market manufacturers who want to sell the most units to the most people, sure it's great. Harman's own research shows that trained/experienced listeners do not have this same preference.

This is false. From Dr Sean Olive's (of Harman) 2018 paper on the over-ear target and preference formula (my emphasis):

Trained listeners preferred the Harman target in all tests. Untrained listeners preferred the Harman target in two of the tests, and rated it about equal to four other headphones in three of the tests.

Headphones received lower preference ratings as their frequency response deviated further from the response of the Harman target.

The slight difference between trained and untrained listeners is in fact opposite to your claim that the latter prefer the Harman target more than the former, but is anyway likely just due to untrained listeners finding it more difficult to discriminate between the top headphones and the Harman target. From the paper:

It is clear that untrained listeners generally had more difficulty choosing a favourite among the top two or three headphones