2016 July/August NY area measurements by Serious

woops yes even I forgot about that.
Good point about the test equipment used.

 

HFM Edition S (open back, also @Maxx134's) coupler measurements

Frequency response

This measurement is indicative of a good seal. I found it hard to get a good seal on my head.

CSD
(in bright, flashy colors of cancer)



I had to raise the top of the CSDs to 125db instead of 120 for the other measurements. I still kept the bottom because the treble is much lower than the peak at 4kHz and because they looked very clean anyway.
The L CSDs actually look pretty good in the treble and they might actually be right. 8kHz null might be an artifact. This is also a common artifact with my IEM measurements, so I can't really say. Could be that it's there.

Harmonic Distortion
I didn't post these for the others because the HD was pretty much below the capabilites of my measurement rig. Not so with these.



A ton of bass distortion, especially even orders. 3rd order is much higher than 2nd. 3rd order is at more than 5% below 50Hz, 2nd order is higher than 1% throughout the whole bass region. Even the 5th harmonic is still much higher than the noise floor, but keep in mind that the noise floor with my UMIK-1 seems to rise a lot below 200Hz, seemingly in a pink noise fashion. I would say that it's below the noise floor from 6th order and up. I kept the higher orders in so you can better see the noise floor, but at the same time it makes the distortion harder to see. I might remove all but the 10th order.
The bass distortion seems worse than the 15$ Superlux that I modded.

 

Oppo PM2 modded by @Maxx134 coupler measurements

Frequency response


CSD




Technically the midrange distortion was above the noise floor, but the measurements are really too noisy. I would feel comfortable posting them had I taken a longer sweep in REW and with slightly better environmental conditions but even then the UMIK is pretty noisy. I'm sure that you could make a better microphone with the same money, but without the ADC. The UMIK is good because it's calibrated, cheap and works with USB, but it's pretty noisy.

These have lots of bass warmth. The treble could be smoother. I don't really care much for the Oppo cans.

 

Before I post the Elear measurements, here's one channel of my modded HD800. Don't worry, the channel matching is actually good, I was only too lazy to do both channels.

Sennheiser HD800X

Frequency response

CSD


Essentially, very close to a "perfect" result to me, in both the FR and CSD. The HD800 is weird in that, if it measures normally (like Strato's measurements), it will most likely sound bad. I'm not sure about the 11 and 16kHz ringing, that's likely an artifact as I have gotten better results before. Super clean - maybe cleaner than it sounds. I will try to find out more when I measure different coupling materials including dead animals. I can drop the noise floor to below -60db and it will still be clean, only limited by the CSD visualization and my microphone noise floor.
If only it had better bass extension and maybe about 2db less from 5kHz on.

 

Ok, here we go. I will not post the TH-X00 PH coupler measurements for now as it wasn't possible for me to get a seal with it on my coupler. I think my coupler is too lossy for closed headphones in general, but open headphones seem to work fine.
Compare to @Marvey's measurements, but keep in mind that we use different couplers. This is the exact same headphone that Marv measured.

Focal Elear

Frequency response

The channel imbalance is partly real and partly having to deal with squishy pads on a coupler, but it's mostly real. Marv's measurements also showed some channel imbalance but he uses a much more compressed Y-axis scale.

CSD



Ignore the two ridges at 2-3kHz on the left channel. That's just environmental noise. I would say my results are consistent with @Marvey's foam CSDs. Remember that I wanted to mimic the V1 coupler, but without losing all the bass.

Next up are the Elear IEM and Elear bass-reduction-adapter measurements. Very interesting will be the Elear IEM measurements. Some of you might already know what they look like, if you look for differences between IEM and coupler in the other ones.

 

Before we move on to the IEM measurements which subjectively capture the FR more accurately to me, I will post the results of the Elear EQ circuit that Marv made. His measurements are here: http://www.superbestaudiofriends.org/index.php?threads/focal-elear-eq-circuit-mod.2606/
Personally I didn't like it because it emphasized the shouty midrange for me, but it works well in removing the warmth. I only measured the right channel. Like I mentioned in my post here, the EQ circuit also has reversed channels.

Focal Elear Bass Shelf Circuit

Frequency response


CSD

Note that this CSD is less noisy than the first R channel one, but this is obviously not because of the EQ circuit.

Here's an overlay of with vs without:


I could look at the IR and phase response but I doubt that the effects will be large. I think that's just worrying too much.

 

Ok, on to the in ear microphone measurements. For now, one picture:

Focal Elear IEM

As you can see, this is both with and without the bass shelf circuit. Without the circuit the 1.4kHz region is nearly 2db forward compared to 500Hz. Doesn't sound like a lot but enough to make me hate a headphone. It doesn't help that the rest of the FR is pretty wonky, too. The Utopia looks more like 5-8db forward between 1 and 2kHz, but also seems to be less wonky in general.

BTW: Sharp resonances at 6, 8 and 10kHz. Only the 8kHz resonance should be there ideally. The 4kHz resonance actually doesn't look too bad on the head.

I should also mention that I tuned my HD800 to be 0.5db laid-back in the broad 800Hz to 4kHz region compared to sub 500H on my IEM measurements, which I oreferred over conpletely flat. Half a db over such a wide region makes a big difference.

 

Yup, will do! The differences between coupler and IEM tend to be similar across headphones (but there is some variance).
Oh, wait. That's not what you meant. Will do both ...

 

Elear, HD600, HD800X coupler overlay

Weird, these measurements don't seem to explain the midrange weirdness of the Elear at all. This is my own HD600, not the HD650. I'm using the same mods people use on the HD650. I have a noisy HD650 stock coupler measurement somewhere and it looked similar, except with less treble. I don't have modded HD650 coupler measurements I think.
This basically means that the Elear's weirdness can only be explained by a very different interaction with the ears than HD600 and HD800. Different wavefront?

More soon.

 

So the problem why it took so long is that my IEMic measurements from today look different than those from yesterday ... weird. Sometimes that happens and you wonder what happened. This is why the HD800X results look a little peakier than usual. Normally the region above 6kHz is in line with the 500Hz level, now it's 1db over that. And today they look rolled off, wtf. I also got a big 50Hz hum on my HD800 graph and the 1-4kHz region could be a little smoother. Oh, and I pet my cat and got distracted. Anyways...

Elear, HD600, HD800X IEM overlay

Still, overall the measurements are in line with what I heard. Comparing my HD600 and the HD800X, I would say that they are closer in treble quantity than the graphs would suggest with the main difference subjectively being around 10kHz, but otherwise the measurements look pretty much how I hear them. Also the HD800 sounds airier than the other two and I'm not sure if the big 6kHz spike on the Elear is accurate.

Update: Coupler - IEM difference

These are coupler measurements calibrated with the averaged IEM results. If the headphone measured like this on the coupler it would look flat on the head. I'm not too sure if the region below 200Hz is accurate on the HD800 plots and the HD600 measurement that I used is a little older, but they should give you a good idea. Elear seems to interact much more strongly with the ear around 1.4kHz, which was already obvious from the other graphs. The seal with the Elear on my coupler seems spot-on, while it seems a little too good on the Sennheiser phones. You can also compare this new graph against this old graph. This way you can see how my older coupler measurements compare to the newer ones.

Now the big question is which of the graphs, IEM or coupler, people find more realistic to their ears. OJ told me my coupler measurements looked spot-on to him, while I think the IEMic measurements more accurately describe what I heard.
Going by the graph above I would say that the best idea is to look at the IEM graphs up to about 4-6kHz and the coupler measurements from 5-7kHz on, while keeping in mind that they should be about 2-3db lower than the baseline by that point. The variance seems pretty low above that point and the IEM measurements just get way too noisy. However I do think that we can extract some information from how the headphones excite resonances in the ear. The HD600 seems to pretty much have no resonance at 8kHz, while the HD800 is closer to speaker in-ear-measurements in that regard.

 

I just found out that the Code-x actually shipped with Alpha Pads, which to me sounded congested and more bassy.
So actually the Audeze pads were an upgrade from stock Alpha Pads, which honestly, sound like garbage to me on both the Code-x and the Ether because of congested sound.

 

Inspired by the discussion in the other thread I will now share the Coupler - IEM difference measurements for other headphones. Note that none of these are FR curves and they are all calibrated for the FR.


First up, however, the difference between an omni mic and the ear canal measurements for speakers, calibrated for speaker response to the B&K target. These are essentially HRTF plots calibrated to the B&K response, but turned around for easier comparison with the "headphone coupler target" plots. It's not easy to get very meaningful data here as the room response masks some of the smaller effects the head and body has. Note that these are at the ear canal opening, not the eardrum reference point. I cut off the Riva S data below 60Hz as it was way too noisy, as the Riva S hardly has any ouput below that. You can argue forever if you should measure both speakers at the same time for these measurements or only one. To make the graphs easier to look at I averaged results from one speaker and both speakers running at the same time.

The Living room has the biggest distance between listening position and speakers with the speakers and the "Rema" measurement was probably the one with the least room reverb.
The main takeaway, at least that's how I see it, here is that the B&K target produces a very flat response from 20Hz to 2kHz and from 6kHz to 20kHz. This depends on many factors. I would say the Living room data is the least accurate from 100Hz to 1kHz as it was very hard to get similar measurements in the reverby room.
This (and subjective impressions) is what I base my hypothesis on that a relatively flat response at the ear canal opening for headphones will sound similar to speakers in a room that closely follow the B&K target. In other words the ear canal opening measurements should be comparable to speaker measurements calibrated with the B&K target.

I've had a lengthy discussion with @OJneg on why headphone measurements of neutral headphones don't look similar to the speaker measurements. At the ear canal opening you don't get a big ear gain peak from 2-6kHz with headphones. Instead I think they work without any compensation. I think this big difference is simply because of how different headphone and speaker listening is.


So what do the "Coupler - ear canal opening difference" plots look like for other headphones?


The Edition S and the PM2 both seem to interact less strongly with the ear.


The Code-X and HE1000 interact much more strongly with the ear. They have more distance between the ear and driver. Both are planars.


Also a comparison between B&K target speaker avg and HD800X and HE1000. I'm thinking that how much headphones interact with the ear will change the soundstage perception. In other words, I think this measurement essentially measures driver distance and the wavefront hitting the ear, among other things.



I'm interested in what other people have to say about this. Is the flat-at-the-ear-canal-opening-target a reasonable headphone target? Do you think the coupler measurements which do not take into account how headphones interact with the ears are more realistic in terms of FR? Tagging @Hands, @ultrabike, @OJneg and @Marvey to see what they think.

 

That's what the graphs above show. The biggest difference seems to be at 3-4kHz to me, at least with my coupler (and my ears). Still, one compensation can't work for every headphone for flat plate couplers. To me the ear canal opening measurements seem more accurate than compensated coupler measurements. The ear canal measurements don't seem to need a compensation (at least until 5kHz) for me to look accurate. I think @Hands said something similar before.

I think this depends on speaker directivity, room reverb and many other factors. I would say a BBC dip is good under most circumstances, but might not be an ultimate target. I know the Raidho speakers are voiced that way (but with a pretty big dip, 5db or more) and it really depends on the room if they can sound correct to me or not. I would say in that case it's overdone.

 

FWIW with the 4 people I measured the results didn't seem to vary much more except for the usual positional and fitment differences up to about 4kHz. From 4kHz on every person I measured showed more 6kHz relative to my ears. It would be interesting to measure the same headphone and speaker setup with multiple people and correlate the measured differences between headphone and speaker with their subjective impressions on the sound differences. Pretty sure that just because the headphone measurements looked similar, it probably doesn't mean that speaker measurements will look similar for different people.

Also I find your IEM measurements to reflect what I heard pretty well once you figure out how to interpret them. Especially the treble measurements can be difficult to figure out - which dips and peaks are actually there and which aren't. Like I said above I would personally use the ear canal opening measurements from 20Hz to about 6kHz or so and the coupler measurements for the treble, but the ear canal measurements do still give some information in the treble. You just have to take them with a grain of salt.
This obviously also depends on the particular coupler. I like my coupler for open headphones but it doesn't seem to work for closed ones.

 

Let me know when you guys invent the Rezolooshun Decombobulator and the Plankton-O-Matic Meter. Otherwise it's all ghey.