[dongster]

what can be the cause of the 4khz on the second graph doing the unsure decay game? I drop like mad, or maybe not, let me think a little, oh ok now I want to drop again, etc. ^_^ (I know I'm really good at doing technical reports). between how I could get about the same behavior on different IEMs, and how there seems to be a clear periodicity to it, I wondered if it was a setting artifact or something real.

i suspect, has to do with relative position to resonant frequency, which btw isnt necessarily singular, there could be multiple with varying concentration of energy.

 

[goodvibes]

Data is data, and I've never shied away from that - rather, I like to educate about physics in general and audio in particular. Did quite a bit of that back in the 2000s when I was running Adire Audio, was one of the first to really "mainstream" the concept of linearity in speakers via linear BL, Cms, and Le over stroke. This is even pre-Klippel days, back when we had either a DUMAX system or a home-brew unit. We built our own BL/Cms/Le jig, it consisted of a forcemeter and a stepper motor. Attach a clamp to the diaphragm, and run a set of small-signal T/S measurements as you physically displace the transducer in or out. Worked well, but was slow and destructive to the moving parts (needed to perforate the dustcap for the displacement, and cut away the surround and spider to complete full Cms modeling).

Anyway, I'm torn about which graph is easier for consumers to understand; the standard CSD is, well, standard. You see it everywhere, including the pages of Stereophile. But I think the WCD is more "intuitive" to grasp what is happening but is limited in terms of use (I think it's kind of an Audiomatica exclusive thing) and can be misleading in that it can make "mountains of molehills"...

I'd use the traditional one. The cycles one will be misinterpreted as a waterfall graph and have folks thinking it rings like a bell at higher frequencies, LOL.

 

[piotrus-g]

Oh and @piotrus-g, hope this comparison would make the difference between IMM-6 and IEC711/318 easier to see.

Thanks I understand now, but it's pretty much impossible for me to apply this curve to every measurement lol I'm just too used to 711. I see it dozens time daily and my brain doesn't work with anything else

 

[DanWiggins]

Already promised the other party to not engage further in his thread, but I'll do so in my own thread. Should be a good educational post anyways.



Measurements, just like all other types of data, are very susceptible to cherry picking. The post above is prime example of that in practice, ignoring nearly all of the "rules" of measurements.

• Data from IEM manufacturers are not to be seen as fully reliable.

Professional measuring equipment or not, they are subject to the largest bias simply because they control the output. Changes in insert depth, volume and tips and SMOOTHING (as evident in the provided Galaxy FR curve) are just some of the few ways that can affect FR in significant ways.

• Do NOT compare between different equipment. Only reference internally.

Obviously this rule has been broken when the party decided to compare the IMM-6 measurements with others. However, this goes even further in that in the six graphs he provided, they come 4 different sources (3 of which are from my own rig). Differences in equipment and technique make this all a completely baseless argument to stand on.

• There is a difference between a DIY coupler and an IEC60711/318-4 standard.

Most obvious traits are in the impedance characteristics at 2,000Hz up. It can be shown below:

If you really, REALLY want to compare my measurements with other people's, you better ensure that you understand this point the most.

• There needs to be an equipment list or reference before analysis of any measurement can be made.

I'm openly show my measurement procedures, my coupler and any small details that may affect my curves. In this party's case, simply citing "professional measuring equipment" is not enough. (And laughable.)

And finally, the reason why this concept of "cherry picking" is so important in this context:

Rhapsodio Solar

(Yes, the dip actually exists.)

Rhapsodio Galaxy (V1)

Third party measurements are taken from Speakerphone. Equipment specifics here

(Peaks shift with insert depth as I've said multiple times before, but I guess I should expect less reading comprehension from now on)

Also to show the consistency between @malvinviriya's rig and mine:

Reference IEM: The same 64 Audio A3 Demo at Music Sanctuary, Singapore

---

TL;DR: DON'T CROSS THE STREAMS

I welcome all genuine criticism. But baseless accusations using cherry-picked data to push an agenda is not.

Personally, I use a CLIO with an IEC 60318-4 compliant coupler for most of my R&D, and I use a CRYSOUND 6151 with a pair of IEC 60318-4 compliant couplers for production. There are differences between them, but each is consistent within itself. Note that we also do NOT smooth our responses, we leave them raw - but then, I have a quiet environment so you can see it's really quite free of noise.

Too many people believe that all measurements must be identical - they are not! Measurement gear, acoustical couplers (even between different brands), background noise, even operator (depending upon how deep an IEM is seated into the coupler) will affect the measured responses. I caution people NOT to take any measurement as gospel, but look at groups of data by a given source to identify trends between products. If you like product X, but feel it is a bit bass-shy, then look for another product that measures as if it is the same as X but with a bit more bass, and those measurements are from the same person. Different rigs will produce different measurements!

And now with GRAS coming out with an "improved" wideband 711-type coupler, things are about to get even more complex...

TL;DR: if you don't understand how to spec, setup, and use a test system, use the data provided by others as interesting points but NOT means to judge the quality of a system.

 

[DanWiggins]

what can be the cause of the 4khz on the second graph doing the unsure decay game? I drop like mad, or maybe not, let me think a little, oh ok now I want to drop again, etc. ^_^ (I know I'm really good at doing technical reports). between how I could get about the same behavior on different IEMs, and how there seems to be a clear periodicity to it, I wondered if it was a setting artifact or something real.

That would be a visual indication of IMD. Lower frequencies modulating upper frequencies.

 

[piotrus-g]

And now with GRAS coming out with an "improved" wideband 711-type coupler, things are about to get even more complex...

Wait haven't they released it already? I was at G.R.A.S. seminar like a year ago where they presented the improved 711. It's 4 times more expensive than normal 711, but accuracy is extreme.

 

[castleofargh]

Personally, I use a CLIO with an IEC 60318-4 compliant coupler for most of my R&D, and I use a CRYSOUND 6151 with a pair of IEC 60318-4 compliant couplers for production. There are differences between them, but each is consistent within itself. Note that we also do NOT smooth our responses, we leave them raw - but then, I have a quiet environment so you can see it's really quite free of noise.

Too many people believe that all measurements must be identical - they are not! Measurement gear, acoustical couplers (even between different brands), background noise, even operator (depending upon how deep an IEM is seated into the coupler) will affect the measured responses. I caution people NOT to take any measurement as gospel, but look at groups of data by a given source to identify trends between products. If you like product X, but feel it is a bit bass-shy, then look for another product that measures as if it is the same as X but with a bit more bass, and those measurements are from the same person. Different rigs will produce different measurements!

And now with GRAS coming out with an "improved" wideband 711-type coupler, things are about to get even more complex...

TL;DR: if you don't understand how to spec, setup, and use a test system, use the data provided by others as interesting points but NOT means to judge the quality of a system.

it is the irony of measurements, they are the most reliable(because fairly repeatable) tools usually available, but somehow people will assume consistency everywhere it doesn't actually exist. just getting the all "no the flat line doesn't always mean neutral to your ear", is a Herculean task.
telling people not to look at measurements from different sources over the years feels like pissing against the wind. and those are just the obvious ones.

personally, just because I understood a few more things than the average audiophile, I thought I understood measurements pretty well(like almost every audiophile in this thread?^_^). then I started trying to do some myself, and since it's been a constant rain of "you know nothing Jon Snow". I start something to answer one question and by the time I have my answer(if I ever get there), I also have 15 new questions.
as for the setup and method I strictly follow this:

That would be a visual indication of IMD. Lower frequencies modulating upper frequencies.

TBH, I have no idea if that's the right answer, but it blew my mind just to consider it.

 

[goodvibes]

I always look at measurements as tools to try and address things we hear and not something to dictate what we should hear. Obviously, significant aberrations will be clear enough on audition.

 

[james444]

Already promised the other party to not engage further in his thread, but I'll do so in my own thread.

Hmmm, what's his problem? Tbh, I've seen worse graphs than your Prelude one, even from expensive IEMs. Personally, I don't think it looks all that bad.

I'm having a hard time digesting temporal masking, when dynamics of similar FR, have noticeably slower decay/bass than armature bass.

Our hearing is much more sensitive between 800 Hz and 3 kHz, so that's a region of concern. Not just in terms of amplitude, but time (ITDs and ILDs both affect that range).

Here's another study I've come across, that indicates our hearing is much more tolerant to differences in lower frequency decay than higher frequency decay: https://www.genelec.com/documents/publications/IOARP21.pdf

So, while we may indeed not be able to distinguish 'faster' and 'slower' bass at fundamental frequencies, it doesn't seem impossible that harmonic frequencies higher up make bass notes appear faster or slower to us.

 

[datranz]

From my perspective, crin's graph may not be the most accurate, but he is consistent, so I can take one iem and compare it to another pretty confidently. As much I enjoy flin's review, but at the end of the day, it's all subjective with no data for reference. He's very skill at conveying sound characters into words, but sometime a graph is worth a thousand words because there are minut details differences that our ears can not hear.

 

[DanWiggins]

TBH, I have no idea if that's the right answer, but it blew my mind just to consider it.

Choose the red pill...

Wait haven't they released it already? I was at G.R.A.S. seminar like a year ago where they presented the improved 711. It's 4 times more expensive than normal 711, but accuracy is extreme.

It's out, but I haven't seen one in the wild yet. I think in the next few years we'll see a ton of new measurements with it, and then thousands of posts about "company XYZ changed their product because measurement different!!!!" nonsense...

 

[DanWiggins]

Hmmm, what's his problem? Tbh, I've seen worse graphs than your Prelude one, even from expensive IEMs. Personally, I don't think it looks all that bad.





Here's another study I've come across, that indicates our hearing is much more tolerant to differences in lower frequency decay than higher frequency decay: https://www.genelec.com/documents/publications/IOARP21.pdf

So, while we may indeed not be able to distinguish 'faster' and 'slower' bass at fundamental frequencies, it doesn't seem impossible that harmonic frequencies higher up make bass notes appear faster or slower to us.

AWESOME FIND! Also interesting to see the conclusion I've seen elsewhere as well as in psychoacoustic research - hearing acuity for a lot of these effects is related to SPL. What you can hear at low SPL is not the same that you can hear at high SPL. And that even goes for "gross errors" like THD and such. Now add in the fact we "experience" total SPL not just as an absolute instantaneous pressure level but a sliding integral over time (that increases with SPL) and suddenly what we hear is hardly a consistent thing!

But what we perceive - what happens to what we hear after it is processed by our wetware - is startlingly consistent. Perception is NOT hearing - and measurements are just hearing.

 

[hakuzen]

hi everybody. subbed, thanks to the op and other contributors for this amazing thread. nice collection of measurements, very good job, congrats!

i started to measure FR of IEMs last year summer, using iMM-6 + tube and mobile AudioTools. since then, was trying to improve my rig till february; i had to stop to concentrate in other tasks, not related to audio.
now i've read this thread, would like to share my rig evolution, together with some comparing graphs, because it has many coincidences with crinacle's one. hope somebody finds it interesting.

1- iMM-6 + silicone tube. Mobile (input & output) + AudioTools (software).

Spoiler: iMM-6 + tube, mobile

my mobile input had a terrible bass and treble roll-off, and the output had about 17Ω (http://www.head-fi.org/t/800208/#post_13041235), so had to change the input to my pc soundcard (http://www.head-fi.org/t/800208/#post_13019304), output to a neutral DAC with output impedance lower than 1Ω (like iBasso D14 http://www.head-fi.org/t/800208/#post_12995932, or JDS C5D).

new software was needed. chose REW.

the nozzle of many IEMs wasn't wide enough to be sealed by the tube, so i had to find a good sealing tip to use with every iem. found that one of those used with stetoclips was a good solution.

2- iMM-6 + silicone tube + tip. PC soundcard (input) + iBassoD14 (output DAC) + REW.

Spoiler: iMM-6 + tube

compared to pro rigs' measurements, like Tyll's one, found many differences: lows and highs roll-off, different amplitude and location (frequency) of mid and highs peaks, etc.
tried to use the calibration file which came with iMM-6 (it was a joke), and played with the distance from the IEM's nozzle to the mic. that distance resulted to be very important; but even the best distance try showed too many differences to consider a huge effort in calibration of the whole rig.

suspected that the diy "coupler" was responsible of the main differences, so tried to make an stainless steel coupler instead, using some vape gear i had around.

3- iMM-6 + DIY SS "coupler".

Spoiler: iMM-6 + DIY SS 'coupler'

a bit better, but still many differences. the mic fixation to the coupler by using bluetac was not stable enough, and the distance issue was there too.

so decided to invest into a chinese iec711 coupler, made by aluminum and other metals, which came with a mic, without any calibration. and also purchased another chinese iec711 coupler after, made by stainless steel and other metals, to check the consistency of these chinese couplers.

4- "cheap" mic + 711 couplers.

Spoiler: 'cheap' mic + 711 couplers

couplers:






mic which came with aluminum coupler (6mm-1/4"- mic, like imm-06, fixed to a 12mm-1/2"- threaded cilindrical shank, to be screwed into the coupler):

Spoiler: Graph: cheap mic + 711 couplers" vs iMM-6 + DIY SS 'coupler'

found that lows and mids (till 5 or 6KHz) were more accurate this way. but from 5-6KHz, the peaks were better located (frequency) in my old rig (iMM-6 + DIY SS "coupler"), despite of their bigger highs roll-off after 10KHz.

this graph show the difference between them (the similar curves correspond to new rig, rolling tips and varying depth insertion). used VSonic GR07 classic ed.



to get better results in highs, i tried to fit the iMM-6 into the 711 couplers, because suspected that the cheap mic was responsible of the frequency offset's mess.
and compared mics and 711 couplers.

5- iMM-6 + 711 couplers.

Spoiler: iMM-6 + 711 couplers

to fit the iMM-6 (6mm mic) into the couplers (12mm threaded), used a thick silicone o-ring and bluetac.

Spoiler: Graphs: iMM-6 vs cheap mic, 711 coupler 1 vs coupler 2

and here are the compare graphs of GR07 classic ed, using both 711 couplers, iMM-6 and the cheap mic.

C1 is the aluminum coupler; C2 is the stainless steel one.
M1 is iMM-6; M2 is the cheap mic which came with the alu coupler.

C1 + iMM-6 (different tips and insertion depth):


C2 + cheapMic (different tips and insertion depth):


C1 vs C2, + cheapMic (to compare couplers):


slight amplitude differences at the 8-9KHz peak and at the 11- 12KHz dip. maybe the o-ring I added into C2 to get extra isolation, did vary some distances (my bad, i should repeat the measurement without that o-ring); or that is the slight difference between couplers, I don't know. anyway, minimal difference.

C1, iMM-6 vs cheapMic (to compare mics):


note the dip of the cheap mic in lows (40Hz centered), and its frequencies offset (from 9.5KHz to 8.5KHz); highs roll-off in iMM-6 (and some lows roll-off, but under 20Hz).


now, measuring Piston 3.

C1 + iMM-6 (different insertion depth):


C1 + iMM-6 vs C2 + cheapMic:


again similar differences, due to the mics, mainly. 9.5Khz -> 8.5KHz offset in the cheap one, and highs roll-off in iMM-6.


conclusions:
agreeing crinacle's comments, i think that an iec711 coupler helps to get closer to the real thing, and so does the mic.
having a iec711 coupler, i'd prefer using iMM-6 (easier to calibrate the highs roll-off) rather than the cheapMic (very difficult to fix that frequency offset).
but the insertion of iMM-6 into the coupler (using o-ring+bluetac) is not very stable, i guess. i'd probably cut the cilindrical threaded case of the cheap mic (about half size), insert iMM-6 into that case, adding some bluetack, and screw the result into the coupler.
i guess this rig, after some calibration (not so heavy), would satisfy my needs and expectations so far, and would be portable (laptop + dac + rig).

6- Chinese "pro" rig (iec711 coupler + precision mic + preamplifier + stand) + low noise and distortion input (E-MU 0404).

Spoiler: chinese 'pro' rig

my PC soundcard inputs has excessive noise and distortion to be used to measure other parameters different to frequency response. so i purchased a low noise and distortion one: E-MU 0404.

after spending in two iec711 couplers and this interface, thought i should try a "pro" calibrated mic. so purchased a chinese one, AWA14424D (calibration provided in a piece of paper, like GRAS ones), very very fragile.. which came mounted into a new iec711 coupler (AWA6162) to minimize the risk of damaging the mic while being mounted, and together with an appropriated preamplifier (AWA14604C) and an stand/base. Ouch, ouch, ouch, this did hurt my wallet badly.
now i have two spare iec711 couplers (if anyone is interested, pm me).

i've finished my other tasks, six months later, so i'll restart with audio again.

here are some pics of the new rig:

(another parameter, wrote in chinese -can't translate it-, says 76% at the end of the paper -wasn't caught by the cam-)

 

[ThomasHK]

Wait haven't they released it already? I was at G.R.A.S. seminar like a year ago where they presented the improved 711. It's 4 times more expensive than normal 711, but accuracy is extreme.

Yeah, they're out. The changed impedance in high frequencies is interesting, but the significant drop in usable dynamic range is a problem. You basically need to old and new 711 (with low noise mic) to cover all possible measurements.

 

[goodvibes]

AWESOME FIND! Also interesting to see the conclusion I've seen elsewhere as well as in psychoacoustic research - hearing acuity for a lot of these effects is related to SPL. What you can hear at low SPL is not the same that you can hear at high SPL. And that even goes for "gross errors" like THD and such. Now add in the fact we "experience" total SPL not just as an absolute instantaneous pressure level but a sliding integral over time (that increases with SPL) and suddenly what we hear is hardly a consistent thing!

But what we perceive - what happens to what we hear after it is processed by our wetware - is startlingly consistent. Perception is NOT hearing - and measurements are just hearing.

Wetware collates like nothing else. Procedure and associated kit can muck that up. Many work tirelessly to not hear differences and succeed.