[durwood]

Will have to investigate. Our space between the mic end and the tip is 2 cm minus the tip length.

You may explain what the optimal distance is and why?

I am basing my distance from the AES link above and also the BK datasheet that calls out 1.26cm^3. Since my tube is 8mm ID, that works out to 2.5cm in length. Then I try to adjust to get my peak at 8khz like crinacle does. THen you have to realize anthying over this resonant point is inaccurate. I don't subtract tip length because that you decrease the volume.

https://www.bksv.com/-/media/literature/Product-Data/bp0262.ashx

I agree, the IMM6 is fine for hobby use as long as we know the limitations. Of course if you are in the business of making headphones you will want something that extends the higher frequency accuracy

BTW I didn't know the complaint was your 2-5Khz region was lower than other public measurements, mine is too, I was only seeing a much more drastic issue >5Khz. I have not compensated my measurements against a known good measurement, but I am curious that I I am lower by about 2-3db in that range. I don't know if compensation will correct this or if this is an isolated difference with the KP.

 

[Otto Motor]

I tried to find the basic physics behind that 8 kHz normalization -- but can't find any explanation on the net.

One guy in another forums writes: "...80% of peaks above 7-8kHz are the result of specific coupler / IEM interactions. Too misleading..."

Headflux.de compare their measurements between their old Veritas coupler and their new IEC 603-18 standard device -- and I can't see much similarity between them: http://headflux.de/neues-messequipment-fuer-inears-iec-60318-4/

What I need to understand:

1. Why does insertion depth move "resonance" peaks around? The resonance should be specific to each coupler and therefore fixed. I obviously miss something here.
2. How can an artificially created 8 kHz peak be real? And why is it created at 8 kHz?

BTW I didn't know the complaint was your 2-5Khz region was lower than other public measurements, mine is too, I was only seeing a much more drastic issue >5Khz. I have not compensated my measurements against a known good measurement, but I am curious that I I am lower by about 2-3db in that range. I don't know if compensation will correct this or if this is an isolated difference with the KP.

I realized that. You were concerned about the treble peaks. As said, our midrange looks rather similar.

 

[crinacle]

1. Why does insertion depth move "resonance" peaks around? The resonance should be specific to each coupler and therefore fixed. I obviously miss something here.

https://en.wikipedia.org/wiki/Acoustic_resonance#Cylinders

2. How can an artificially created 8 kHz peak be real? And why is it created at 8 kHz?

Consistency. Most other HATS rigs (Innerfidelity, Rtings) have half-wave resonances at around 10kHz but I'd say that's inaccurate as these rigs usually have their in-ear measurements at reference plane i.e. a canal bypass and usually only reserved for custom-moulded equipment, not universal earphones. I selected 8kHz as it was a resonance point that I could obtain very reliably, and I guess the rest followed suit to keep parity with my data.

 

[Otto Motor]

My co-blogger lets us know:
I don't know why this is even a subject of discussion. It should be obvious that there'll be differences due to different internal dimensions (all of volume, length and width) of the coupler, and the acoustic impedance is obviously different between a proper IEC one and a bit or bits of tubing. The links that @durwood posted give dimensions for the IEC711, which a lot of people try to imitate (or have, because cheap Chinese ones are now available). I didn't try to copy that or any other when I made our stepped couplers, I just used materials I had to hand that'd fit the end of the microphone and would conveniently fit medium tip sizes because, for the 100th time, I was interested in *relative* differences between mods to the same earphone or between different earphones that I had here, to see if I could correlate different measurement results with what I thought I was hearing. That the results so far seem to have been consistent with my perceived variations in tonality, including across different earphone types (with occasional surprises, eg UE900s, which is likely due to the odd venting), and not *massively* inconsistent with raw measurements from many other people, has been a bit of a surprise.

Different couplers will give different results. Instead of wondering why there are differences and picking at them, It would be more useful to *highlight* the differences so that people can mentally 'compensate' your measurements to those of others (and vice versa) for earphones they've heard.


Otto adds: I'll charm him into cutting a 2.65 cm tube for comparison reasons. This will result in a net 2.15 cm coupler.

 

[hakuzen]

Uh Oh... I could ask some compromising questions about Your calibration and why Your measurements differ etc. ...but I just don`t bother. Do what You want and how You want, not my to say.
I personally hope You get Your measurements more "credible", accurate and comparable to others, not just Your buddy.
But You could stay with Your own settings and start to create Your own "database" which can be used within Your site reviews. Latter is easier way.
Good luck with Your measuring gear and everything related to it

Btw, I would like to hear what hakuzen has to say about all this as he has spent some serious time for his evolving measuring gear.

yep, i'm one of these who started with iMM6+tube and ended with a 60318-4 rig, as told in my signature. the advantages are huge.
you can build your database using iMM6+tube and keep consistency between your own measurements. but if you plan to do many measurements, compare to others, and post them in a website, investing in a IEC 60318-4(711) rig is very recommendable

 

[castleofargh]

1. Why does insertion depth move "resonance" peaks around? The resonance should be specific to each coupler and therefore fixed. I obviously miss something here.

any container has its own resonance frequency, from beer bottle to empty beer bottle the internal volume of air changes and so does that resonance. for closed tubes, the simplified estimation is: speed of sound / (4 times the length of the tube).
so if my ear canal is... say 2.5cm=0.025m, and we decide on the speed of sound as 343m/s, we end up with 343/0.1=3430Hz as estimated first resonance. everything in this is an approximation that should change depending on circumstances, but just the usual equal loudness contour could suggest that we do get a boost around 3khz.

on this they suggest below 3kHz, but as people have different ear canal sizes and shapes, such variations are completely expected.

just like the next big resonance(we should only consider 5 on that graph), is going to change even more, depending on where the first one happens. usually we see ear measurements and ear simulators with such resonance somewhere between 9 and 15kHz I'd say.

My co-blogger lets us know:
I don't know why this is even a subject of discussion. It should be obvious that there'll be differences due to different internal dimensions (all of volume, length and width) of the coupler, and the acoustic impedance is obviously different between a proper IEC one and a bit or bits of tubing. The links that @durwood posted give dimensions for the IEC711, which a lot of people try to imitate (or have, because cheap Chinese ones are now available). I didn't try to copy that or any other when I made our stepped couplers, I just used materials I had to hand that'd fit the end of the microphone and would conveniently fit medium tip sizes because, for the 100th time, I was interested in *relative* differences between mods to the same earphone or between different earphones that I had here, to see if I could correlate different measurement results with what I thought I was hearing. That the results so far seem to have been consistent with my perceived variations in tonality, including across different earphone types (with occasional surprises, eg UE900s, which is likely due to the odd venting), and not *massively* inconsistent with raw measurements from many other people, has been a bit of a surprise.

Different couplers will give different results. Instead of wondering why there are differences and picking at them, It would be more useful to *highlight* the differences so that people can mentally 'compensate' your measurements to those of others (and vice versa) for earphones they've heard.

almost everybody participating here is an amateur or just starting to become curious about what the hell is going on objectively, so obviously we're considering a lot of stuff and try to find common references even when we understand that they're arbitrary or even maybe caused by the measurement rig itself. you mention the desire to show variations between gears, and I agree that it's the only thing worth analyzing, more is often going to touch over-interpreting something the data wasn't actually saying. but if our modus operandi greatly affects our measurements and introduces changes within frequencies that are subjectively more important than others, then IMO it's worth thinking about it and maybe settle somewhere else. or even if we don't mind that, finding such resonance and deciding to have the same on all measurements if possible, that becomes a measurement method to have repeatability and the idea of a reference plane. repeatability is always a good idea for measurement, but a reference plane is not very realistic compared to practical use of IEMs. different IEMs are made and will be used at different insertion depths, and then to make things even more complicated, you have to modulate all those variations with the size and shape of the listener's ear(or simulated ear).
so there won't always be a way to simply highlight the differences in our measurements. and not even always a clear answer to the question of how we should do something. but IMO it's important to discuss those and get a clear understanding, if only to make our own decisions.

 

[CoiL]

Ok. This is my final calibration. This time out of HO with low impedance:

Whole bass area is less "wobbly" compared to previous LO measurement. Peaks and dips also got little "correction".
Didn`t align resonant peak @ 8kHz. That graph is what I think I hear out of KPE... if 8kHz peak would be there, my ears would be in pain.
But for sake of comparison (for myself), I`m going to try to make "swappable" tubes - one as current and one for 8kHz peak. Will see how it works out.

Btw, did I miss it Otto, or You haven`t done review about KPE yet?

 

[Otto Motor]

Ok. This is my final calibration. This time out of HO with low impedance:

Whole bass area is less "wobbly" compared to previous LO measurement. Peaks and dips also got little "correction".
Didn`t align resonant peak @ 8kHz. That graph is what I think I hear out of KPE... if 8kHz peak would be there, my ears would be in pain.
But for sake of comparison (for myself), I`m going to try to make "swappable" tubes - one as current and one for 8kHz peak. Will see how it works out.

Btw, did I miss it Otto, or You haven`t done review about KPE yet?

No KPE review yet...I haven't even had in in my ears yet.

I re-iterate again: we at Audio Reviews create our own internally consistent database and don't compare to other measurements (see our disclaimer). The whole discussion started when somebody pulled the KPE graph out of our database and started bickering. This was followed by personal attacks paired with the usual lack of (scientific) rigour.

The cherry-picked corpus delicti in our MKP graph was the lack of an upper midrange plateau (we were talking 1.5 - 2 dB), but what had been missed in this overambition was our fewer treble peaks compared to the IEC couplers (attributed to lack of distance between iem and microphone). In the meantime, we increased the coupler length from 2.0 to 2.65 cm according to the scientific input by @durwood and inserted the tip to the end of skirt so that the actual distance between microphone and iem was increased ca. 2.15 cm [@durwood himself also could not produce that plateau with his plastic coupler but demonstrated the dependency of >10 kHz peaks from coupler/insertion depth interaction].

Our bass is now unrealistically exaggerated and so are the treble peaks (we found an 8 kHz one by playing with insertion depth). And we also couldn't produce that upper midrange plateau. That it's still significantly different from IEC isn't surprising; even though internal volume is closer, width and length are different and there's no internal damping (acoustic impedance).

But I had said it before: even the IEC 711 measurements don't match well...compare the relative magnitudes between upper treble peaks and bass....Headflux (black diagram) have the perfect rig and the most elevated bass. If one had plotted these graphs on top of each other, we would be surprised...the upper treble match is also problematic. I will sit back and watch the IEC protagonists fighting it out who is right and who is wrong.

I also can't get used to the idea to force an 8 kHz peak as a reference. In summary, we may as well keep our first coupler as it does the job we want for us.

But more importantly, I discovered a new generation of Bluetooth speakers at IKEA and tortured customers with my music picks.

Spoiler: Bad Video Clip: Rammstein and Otto at IKEA

 

[CoiL]

The whole discussion started when somebody pulled the KPE graph out of our database and started bickering. This was followed by personal attacks paired with the usual lack of (scientific) rigour.

Geesh... calm down dude!
And when You throw graphs online in many threads - some disagreement and "arguing" is to be expected. Take it with calm mind and keep on doing what You do (as You think it is right).

About Your new coupler sitance etc. ....that red line look pretty good now, that "plateau" is there actually and is affected by insertion distance as someone posted graph about earlier.
The closer You move insertion depth to mic, the further away from 8kHz point that upper treble peak moves (tried myself around 2cm distance vs. 1.3cm distance). But since I have small ears and ear-canals, I prefer to measure about same distance/depth my ears actually have (averaged according to my own taken ear impressions). It`s just that I wish to measure thing I`m actually hearing and for me and for my needs it is "correct" way to measure.
Also different source gear (output impedance and dac/opamp sound signature) and method (sine sweep vs. pink noise) along with tip choice etc. can all alter results. Heck, even low-end distortion/vibration is serious issue with iMM-6 and different couplers (see hakuzen thoughts about this).
So, don`t worry so much if there are slight differences.
I have slight bumps @ 2.3 and 4.3kHz at my graphs but this is due to software limitation (calibration points "resolution" is too low). But general "average" curve is pretty similar to other graphs. Also my custom nozzle filters might be somewhat cause of those "bumps"..

 

[durwood]

yep, i'm one of these who started with iMM6+tube and ended with a 60318-4 rig, as told in my signature. the advantages are huge.
you can build your database using iMM6+tube and keep consistency between your own measurements. but if you plan to do many measurements, compare to others, and post them in a website, investing in a IEC 60318-4(711) rig is very recommendable

Just curious, how much did the IEC rig set you back and do you know where to purchase if in the US? I think you are located somewhere where you had easy access right? Since I am doing this for my own basic knowledge as a hobbyist, not sure I feel like investing in yet another measurement system.

Our bass is now unrealistically exaggerated and so are the treble peaks (we found an 8 kHz one by playing with insertion depth). And we also couldn't produce that upper midrange plateau. That it's still significantly different from IEC isn't surprising; even though internal volume is closer, width and length are different and there's no internal damping (acoustic impedance).[/SIZE]

Maybe the inner vent was blocked? I know how easy it is to accidentally block the vent on many IEM's when trying to get ones with large nozzles inserted in the 8mm diameter coupler. I have to make sure the tips does not shift and cover the vent. That boost in the bass looks consistent with the measurement floating around where someone taped the inner vent.

I also can't get used to the idea to force an 8 kHz peak as a reference. In summary, we may as well keep our first coupler as it does the job we want for us.

No one says 8khz is the right place, 7khz is probably more legit given the dimensions mentioned in the IEC infromation. As @crinacle had mentioned, it was a arbitrary decision and some followed just to keep consistency between their measurements and also against others perhaps. The peak is not necessarily in the IEM, but some may have a peak there too and will get masked.

I was thinking about how we like to show very detailed measurements too, i.e. 1/48, 1/24th octave etc but in reality our hearing is going to smooth it out to 1/3 octave. This would make the KPE look more similar to Harman, and we would not see or hear that dip @ ~6.5khz. I had been bothered by the claims it was Harman tuned (another thread), when detailed measurements show the dip. In hearing it, we don't perceive that dip, but then the MH755 is absent of the dip completely and follows Harman better but with bass boost.

 

[durwood]

FYI-

If you want to overlay these graphs, here is a nice little application I found on parts express techtalk looking for an older one called SPLtrace. It is pretty awesome at extracting the curve out of a graph where you can then use it to overlay elsewhere.

http://labs.fprawn.com/index.php/fpgraphtracer/
http://techtalk.parts-express.com/forum/tech-talk-forum/64837-new-automatic-graph-tracing-program

 

[Otto Motor]

It is not only the coupler that can make a big difference. It is also the software settings. For example, I was recommended to measure at 94 dB at 1 kHz, but this causes the clipping of peaks, I therefore stayed with 85 dB. I am presently finalizing my extensive "calibration" and "measurement" essay and will post it on my blog within the next couple of days...it will include screenshots of every relevant pane and window. This will serve as yet another discussion basis and may eventually lead to the improvement of my writeup.

Below an example, where the coupler discussion is futile as the job has been achieved with the plastic coupler.

1. The iBasso IT01 with and without screens...the treble difference was audibly strong so that I re-added the screens after having removed them.
2. Same pair later: the channel imbalance caused by ???? --> iems are presently on their way for warranty repair.

--> Purpose fully achieved.

Also, we (my blog partner and me as the only two) can reproduce the same results on two different rigs, which means we have a good precision. This shows that precision (a measure of the reproducibility of results) and accuracy are not related.

 

[hakuzen]

Just curious, how much did the IEC rig set you back and do you know where to purchase if in the US? I think you are located somewhere where you had easy access right? Since I am doing this for my own basic knowledge as a hobbyist, not sure I feel like investing in yet another measurement system.

i'm located in spain, and got my chi-rig from taobao by using an agent (you'd probably get even easier access to agents than me). couldn't find it at europe nor usa (except of the better but more expensive GRAS or B&K rigs).
search for "iec711" and for "60318-4" at taobao, and you'll find several models and prices (in yuans and usd).
i got 4 different couplers. best one was AWA, but the cheaper ones are good enough to get a nice improvement in your measurements, and they are very affordable. for hobbyist purposes, you could pass on iec60318-4 mic, preamp, and stand; just consider the coupler.

 

[Otto Motor]

. for hobbyist purposes, you could pass on iec60318-4 mic, preamp, and stand; just consider the coupler.

Our built-in own couplers (=our ears) produce likely much different results from any of the couplers mentioned here...which is important for correlating perceived sound and measurements.

The question is, are our ears iec60318-4 compatible?

 

[hakuzen]

Our built-in own couplers (=our ears) produce likely much different results from any of the couplers mentioned here...which is important for correlating perceived sound and measurements.

The question is, are our ears iec60318-4 compatible?

no, years of studying, researching, and developing only served to create an arbitrary ear which has nothing to do with average human ear channels..
the vinyl tube or your diy coupler is surely much more accurate and closer to what we hear