[csglinux]

That makes sense that I'd want to see the response at the SPL level of my normal listening volume. Perhaps this is where microphone quality comes into play. Maybe some are just more accurate than others? Or can compensation take care of this, and quality differences wouldn't be a factor due to the compensation? You already told me about expensive ones with low noise.

I wouldn't worry at this stage about mic noise or sensitivity. That's only important if you're measuring THD and THD isn't anywhere near as significant as FR.

to improve the accuracy of option 1, you can use your coupler to cobble together your iem and spl meter. just place the spl meter mic where you'll place your measurement mic.

That's a very interesting and thought-provoking suggestion - for a couple of reasons! It could be a good option for those that regularly take the mics out of their couplers (@McMadface's UMIk-1 mic seems to fit perfectly into these 711 couplers just via a push-fit). On the other hand...

I was recently told by Bob O'Neil (GRAS) that people can destroy their GRAS couplers by removing the mics. Liliang Hong (one of the Taobao sellers) told me their coupler/mic combos are "debugged" (I presume that is a loose translation of "calibrated") to a close tolerance before being shipped out, and hence don't need compensation files. I may be wrong, but I think the only way you could "debug" hardware like this is via the adhesive and how far you tighten the mic down the thread. I can vouch for the fact that with those generic Taobao couplers - unscrewing and then re-seating the mic does lead to different measurements. The seal and amount of volume determines the Q and resonance of the coupler. I know for sure you have to get that right or the coupler/mic measurements can be off by unexpectedly large amounts.

If you have a 711 coupler that came with its mic sealed in it, I'd be very cautious about removing the mic. @SilverEars and @Slater in particular - please don't try to remove your coupler mics - that would not have a happy ending.

 

[hakuzen]

I wouldn't worry at this stage about mic noise or sensitivity. That's only important if you're measuring THD and THD isn't anywhere near as significant as FR.



That's a very interesting and thought-provoking suggestion - for a couple of reasons! It could be a good option for those that regularly take the mics out of their couplers (@McMadface's UMIk-1 mic seems to fit perfectly into these 711 couplers just via a push-fit). On the other hand...

I was recently told by Bob O'Neil (GRAS) that people can destroy their GRAS couplers by removing the mics. Liliang Hong (one of the Taobao sellers) told me their coupler/mic combos are "debugged" (I presume that is a loose translation of "calibrated") to a close tolerance before being shipped out, and hence don't need compensation files. I may be wrong, but I think the only way you could "debug" hardware like this is via the adhesive and how far you tighten the mic down the thread. I can vouch for the fact that with those generic Taobao couplers - unscrewing and then re-seating the mic does lead to different measurements. The seal and amount of volume determines the Q and resonance of the coupler. I know for sure you have to get that right or the coupler/mic measurements can be off by unexpectedly large amounts.

If you have a 711 coupler that came with its mic sealed in it, I'd be very cautious about removing the mic. @SilverEars and @Slater in particular - please don't try to remove your coupler mics - that would not have a happing ending.

yea, you are right.
my AWA precision mic came mounted into the coupler.
the component which is very susceptible of being broken is that kind of precision mics.
i've never unscrewed it, and hope i'll never have to do it. if you use that kind of mic+coupler combo, don't touch it, and take special care of it.
for measuring SPL, i use another iec711 coupler (cheaper than AWA). it is permanently attached to the spl meter.

 

[SilverEars]

If you have a 711 coupler that came with its mic sealed in it, I'd be very cautious about removing the mic. @SilverEars and @Slater in particular - please don't try to remove your coupler mics - that would not have a happing ending.

Yes, the coupler is sealed with some resin material to the mic, so it's not a simple disassembly. I do understand @hakuzen 's point regarding the calibration should be in relation to the coupler being used.

Lately, I've been interested in difference of iem response due to difference sources, and also due to different loudness output from the sources. I've always been interested in this because I do hear differences when I switch sources.

I've done some Andro tip rolling measurements, and it's here. What I was surprised is the result from bore size differences. Also, result with foam tip was interesting in how the peaks were damped.

https://www.head-fi.org/threads/cam...romeda-and-nova.805107/page-482#post-15045366

 

[hakuzen]

I do understand @hakuzen 's point regarding the calibration should be in relation to the coupler being used.

the purpose of using a similar coupler to measure the level is getting more SPL level accuracy, at 1kHz generally; because that will be the "true" volume your mic will receive.
there is not need to use the same exact coupler to measure SPL level. there won't be higher SPL difference between them than the error from your SPL meter.

another thing is rig calibration.

first, you calibrate your DAC+amp and ADC combo. for this, you won't use the coupler nor mic. i recommend using similar output voltage (preamp gain will also be similar) than the voltage that is going to be used with your iem (this is variable, depending of the iem sensitivity, that's why i save some calibrations for some different output voltages). that voltage will be usually into the 10-100mV range. the lower the voltage, the bigger roll-off (when using my rig, at least; considerable roll-off when going below 15mV).

coupler calibration.
you'll do nothing about the coupler. in theory, you wouldn't need any correction when using a iec60318-4(711) coupler. if you liked a really fine tuning, you'd need references from same exact phones measured with a fine calibrated rig (which is what you need if you use a non-iec711 coupler).

mic calibration.
you need calibration data for calibrating your mic. the manufacturer should provide this data, but reality says that only expensive mics from reputed brands come with trusty data. even G.R.A.S., they provide the individualized calibration data.. but in a piece of paper!
the alternatives are:
a) buying the calibration service. there are companies who offers these services (you'll send the mic to them, an you'll receive the mic together with the calibration).
b) renting a perfect calibrated mic, to use it as reference for calibrating yours. of course, you'll have to be very careful with the delicate rented mic.
c) getting a reference from same exact phones measured with a fine calibrated rig.

 

[CoiL]

Lately, I've been interested in difference of iem response due to difference sources, and also due to different loudness output from the sources. I've always been interested in this because I do hear differences when I switch sources.

This is also interesting to me, as volume-matched sources show clear difference to my ears (using DIY source-swap-switch with all/both source cables connected, just flip switch).
More and more I`m reaching to conclusion that writing in-depth reviews is really hard to compose in objective way as there are so many different chain factors that affect reviewer views and result heard.

I've done some Andro tip rolling measurements, and it's here. What I was surprised is the result from bore size differences. Also, result with foam tip was interesting in how the peaks were damped.
https://www.head-fi.org/threads/cam...romeda-and-nova.805107/page-482#post-15045366

Tip-rolling is another important thing ppl often forget or don`t know about. I like my KPE with stock tips (same You used for measurements).
Sometimes it is really hard to hit "jackpot" with tip-rolling as You might not find "right" FR/sound alternating tips for Your own hearing response.

Someone should write in-depth article about different factors affecting end result of sound and make it a sticky. I haven`t seen such article that involves many different factors and "enlightens" ppl how "wrong" they might be reading things out of reviews. Source gear "signature", output impedance, amping, software processing (hidden kernel/system EQ settings/processing etc.), tip-rolling, nozzle size and filtrering, cable impedance etc.

 

[castleofargh]

Someone should write in-depth article about different factors affecting end result of sound and make it a sticky. I haven`t seen such article that involves many different factors and "enlightens" ppl how "wrong" they might be reading things out of reviews. Source gear "signature", output impedance, amping, software processing (hidden kernel/system EQ settings/processing etc.), tip-rolling, nozzle size and filtrering, cable impedance etc.

I planned to do this for a long time, and even started https://www.head-fi.org/threads/feedback-about-gears-stop-doing-it-wrong-impedance.866714/ even though this one focuses more on discussing the source than the IEM, it's exactly the same thing as the resulting sound is always the reaction between both.

this was clearly intended to be just one part of a series, as the weird title might let you guess, but I concluded that I suck at this. being an educator requires some skills. my talent is to turn a simple sentence into 10 lines of convoluted angrish + unrelated rants. I can't even use my native language as an excuse as I do the same in French.

tips and insertion depths are 2 things that are very easy to demonstrate with our little rigs. I assume we all here have tried it. my fear is that if someone sees +3dB at 10kHz with tip X, then he will "infer" that:
- he will get +3dB in his own ear
- if he uses that tip on that other IEM, it will help boost the 10kHz area the same way.
I'm not sure how to present something the will not have people just fly over the post, see the graphs and dismiss, or maybe not even read the page long list of warnings about the graphs and what they mean or do not mean.

 

[SilverEars]

I just planted the Moondrop comparison measurements consisting of Kanas Pro, Crescent, and the new Spaceship here. https://www.head-fi.org/threads/moo...ressions-thread.894139/page-235#post-15055568

I have a stock Galaxy S8 iem that's pretty wore out, the housing layers coming off, and order 3 new sets of the same iems. I measured the old one that's wearing out and new one, and look at the difference. I wonder why there's such a significant difference? I measured a couple of MH755 and the measurements were pretty much identical.

 

[SilverEars]

I'm trying to measure custom iems. All advices appreciated. I have a 711 coupler.

 

[castleofargh]

I'm trying to measure custom iems. All advices appreciated. I have a 711 coupler.

blu tack all around to act as if the ear canal properly seals.

 

[SilverEars]

blu tack all around to act as if the ear canal properly seals.

I'm concerned about the distance inside the coupler.

Anybody know if there are issues with that? I'm trying for most accurate placement of peaks I can.

 

[castleofargh]

I'm concerned about the distance inside the coupler.

Anybody know if there are issues with that? I'm trying for most accurate placement of peaks I can.

I don't see a difference between IEM and CIEM. if it fits I try to get where I think it would go in the ear(usually mine). if it doesn't go there, then I measure what I can.

 

[guicnovaes]

Comparing some Graphs, on th Crinacle's graph comparison tool, it is not correct to normalize at a certain frequency response point (normalize at 1khz, for an example), to compare two different graphs. The correct thing is to normalize at some point on the db's scale (normalize at 60db, for an example).

For example, Comparing the curve of Hifi Boy OS V3 with the one of the UERR, these are the result when I normalize at 1khz:

Now, this is the result when i normalize at 2khz (note that the difference between, the two graphs at 5khz, for example, changed):

Now, This is the result when I normalize in 60db:

Now, when i normalize at 80db (note that there was no change in the difference between the two graphs, when i changed the db scale):

Yoy can try yourself with various IEMs on the comparison tool.

 

[castleofargh]

Comparing some Graphs, on th Crinacle's graph comparison tool, it is not correct to normalize at a certain frequency response point (normalize at 1khz, for an example), to compare two different graphs. The correct thing is to normalize at some point on the db's scale (normalize at 60db, for an example).

For example, Comparing the curve of Hifi Boy OS V3 with the one of the UERR, these are the result when I normalize at 1khz:



Now, this is the result when i normalize at 2khz (note that the difference between, the two graphs at 5khz, for example, changed):


Now, This is the result when I normalize in 60db:


Now, when i normalize at 80db (note that there was no change in the difference between the two graphs, when i changed the db scale):



Yoy can try yourself with various IEMs on the comparison tool.

measuring for a given loudness at 1kHz is the convention on FR graphs. so of course because of that when we put several measurements together on the same graph, they will usually be published matched at 1kHz and we'll usually just leave things that way or even align them that way to follow what people are used to seeing. but there is no law saying that we can't align graphs anyway we like to highlight certain differences or similarities between the graphs.

not sure normalize at 60dB means what you think it means, or maybe I'm misunderstanding you. isn't it just that you select a dB value and a frequency and the graph will show that dB amplitude at that selected frequency? and @crinacle you're probably aware already, but our friend just highlighted a bug in your app apparently, as I don't suppose there should be a change in the relative position between both graphs when he picks a different dB value. I could replicate that pretty easily and I'm guessing it may have something to do with the math when the lowest value of one graph reaches a certain lower limit? (or maybe not, IDK).

 

[guicnovaes]

measuring for a given loudness at 1kHz is the convention on FR graphs. so of course because of that when we put several measurements together on the same graph, they will usually be published matched at 1kHz and we'll usually just leave things that way or even align them that way to follow what people are used to seeing. but there is no law saying that we can't align graphs anyway we like to highlight certain differences or similarities between the graphs.

not sure normalize at 60dB means what you think it means, or maybe I'm misunderstanding you. isn't it just that you select a dB value and a frequency and the graph will show that dB amplitude at that selected frequency? and @crinacle you're probably aware already, but our friend just highlighted a bug in your app apparently, as I don't suppose there should be a change in the relative position between both graphs when he picks a different dB value. I could replicate that pretty easily and I'm guessing it may have something to do with the math when the lowest value of one graph reaches a certain lower limit? (or maybe not, IDK).

I think maybe you didn't understand what I meant.

"I don't suppose there should be a change in the relative position between both graphs when he picks a different dB value." In fact, there isn't a change in relative position between bother graphs when you pick a different DB VALUE; But it happens when you pick a different Frequency value, like I showed when you pick 1kz, and when you pick 2khz on the images.

You can try the Crinacle's graph comparison tool, change the db values and the frequency values, and normalize at theses values and you will see what happens.

I don't think it is a bug on his tool, I think it is a misunderstanding about reference point. The value of db selected relative to the two Curves. When you chose a frequency value (1khz, for example), the 1khz for the IEM A is at a different DB value than the 1khz from the IEM B. So, when you settle 1khz to compare the two, the result (different between the two Curves) will not be faithful, unless the frequency value of 1khz of the two IEM is on the same DB value.

 

[dhruvmeena96]

Ok

I saw something like chiFI chainsaw spike at 3kHz

Its more dependent on boost than frequency

If 3kHz is boosted 10dB in respect to 2kHz...only then it classify as chi fi burst peak

If it is compared to 1kHz
The 3kHz should be 10dB to 8dB boosted and the boost should be gradual(breakup tuning can and should be done by pros)

I design IEM, so I know

So every iem having 3kHz peak is not piercing stuff