crinacle's IEM FR measurement database

shotgunshane said:

Most iems do roll off in sub bass. However Andromeda remains pretty strong down deep, so I'll test with it soon.

Click to expand...

 
I think there's also the caveat that what we're doing with the coupler is essentially providing a perfect seal, which is usually not possible in real world circumstances.
 
However because of this, measurements should not have (or have minimal) sub-bass rolloff.

crinacle said:

I think there's also the caveat that what we're doing with the coupler is essentially providing a perfect seal, which is usually not possible in real world circumstances.

However because of this, measurements should not have (or have minimal) sub-bass rolloff.

Click to expand...

Sorry - but that is rubbish. A lot of IEMs have sub-bass roll-off because that is the way they are tuned. There are a few (Andromeda being one) which have great extension with practically no roll-off. I mean for Petes sake - are you listening to some of these IEMs with sub -bass heavy music? The roll-off is pretty audible.

shotgunshane: Can you please overlay the L/R Channel measurements for 1More Triple with normalization at 1kHz turned on? It looks to me like something went awry with your L/R measurements. I've seen that odd stepping occasionally (very rarely), and it usually coincides with a measurement misfire...

or could it be that Niks unit is actually measuring more accurately perhaps?


Most of @nmatheis
' AudioTools graphs also have that rolloff. Seems to me like a limitation/glitch rather than an accurate representation.
[/quote]

could it be some weighting setting that fails to be included sometimes? sorry if that's nonsense, I never tried audiotools.

brooko said:

or could it be that Niks unit is actually measuring more accurately perhaps?


Most of @nmatheis
' AudioTools graphs also have that rolloff. Seems to me like a limitation/glitch rather than an accurate representation.

Click to expand...

The two tools (AudioTools and Spectrum Analyzer) certainly give different FR Curves. That's for certain!

I'm all for using both for completeness, but I much prefer AudioTools. It generates less jittery curves much faster and has the ability to overlay curves, average curves, normalizes at 1kHz, and catalogs curves so you can recall them later for comparative purposes.

My philosophy is that even though my FR curves aren't completely accurate, they can at least be used as comparative tools. I'd love to get a pair of Ety ER4 in to measure and include as a neutral reference. If anyone is willing to loan me a pair, I'd appreciate it!

brooko said:

crinacle said:

I think there's also the caveat that what we're doing with the coupler is essentially providing a perfect seal, which is usually not possible in real world circumstances.

However because of this, measurements should not have (or have minimal) sub-bass rolloff.

Click to expand...

Sorry - but that is rubbish. A lot of IEMs have sub-bass roll-off because that is the way they are tuned. There are a few (Andromeda being one) which have great extension with practically no roll-off. I mean for Petes sake - are you listening to some of these IEMs with sub -bass heavy music? The roll-off is pretty audible.

Click to expand...

There would be probably some roll-off but not to the degree that the AudioTool program is showing.

I've compared mine and Nick's curves to various other sources (speakerphone, Tyll, headflux, 84, yours even just to name a few) and they match up rather nicely 1kHz down. Throw in AudioTools measurements into the mix and it's clear that something's very off.

I'll concede the point that there may be some roll-off that I'm missing (or other IEC couplers for that matter). However, it's definitely not a 10dB drop. That is way too drastic.

Also if you compare your Veritas measurements with Ken Ball's on the Andromeda (I can't do proper quoting now, I'm on mobile), you'll see that yours doesn't have the same linearity as his from 100Hz down, instead having a slight roll-off.

Of course, it's not accurate to compare different rigs at all, but my first point still stands; there's a weird compensation going on in AudioTools that makes it inaccurate in the sub-bass, specifically 100Hz down. If it truly is accurate, it'll stand to reason that every other IEC standard coupler isn't, and that includes Mr. Ball's 15k setup.

Slightly off topic, but I live in Portland so I went over to visit ALO for kicks one day and pick up an Rx. Met Ken and got to try early versions of the first CFA IEM. We even threw Orion on his measurement rig and generated a couple FR curves. He told me Vitaly from (then) 1964 EARS helped him set the rig up and taught him to use it. Nice seeing two IEM rock stars working together more that!

If you guys are not on the go and have a PC with a low Zout amp just use ARTA...this should be your benchmark.

Just measured the ER4XR again with AudioTools to determine its accuracy due to the XR's known low-end boost:
 

For reference, measurements from other sources:
 

 
If this isn't an indication of wonkiness on AudioTools' part then I don't know what is. I did save some of my graphs there though and I do like the ability to save the raw data for future use so I really hope that my wishlist for the app (decade scaling, bass rolloff) can be achieved, then I can say goodbye to Spectrum Analyzer forever. For now though, maybe if we compensated for AudioTool's roll-off on top of the normal mic calibration...?

And now for the meat of the meeting, a new small update! (Raw on the left, compensated on the right)
 
Eternal Melody EM-2

 
 
Eternal Melody EM-5

 
Eternal Melody EM-5H

 
 
Audio Technica ATH-IM02

I submitted a bug report to AudioTools and gave them the screen shots. They've already acknowledged receipt. I'll keep you guys posted.

shotgunshane said:

I submitted a bug report to AudioTools and gave them the screen shots. They've already acknowledged receipt. I'll keep you guys posted.

Click to expand...

Right on!

Very interesting thread everybody. I'm always impressed by the ingenuity of the folks in this community
 
The major problem seems to be the (in)ability to compare measurements made on different rigs, using different drivers, software, etc. I know there are a zillion uncertainties, including how the IEMs are inserted in the coupler (something none of us can probably every exactly replicate). I want to start with a more basic question though. Do you folks know what is on your y axis? I see most FR graphs with the y-axis labeled as "SPL" - and I've committed the same sin, but this is horribly ambiguous. I asked this question on another forum, but so far haven't had any feedback on this issue. 
 
If you don't apply any type of octave-band smoothing, I presume (I would hope) that the Y axis values are actually power spectral density, i.e., dB/Hz. Because if not, the amplitude from your FFT is going to depend on your sample rate and/or sample length. In other words, you could make the exact same measurement again with a different sampling rate and get a different answer - from the exact same IEM/coupler, without even moving anything. I saw mention earlier in this thread of a "decale" smoothing. Can anybody explain what that is? I've never heard of that before. Even if one wants to apply a 'standard' smoothing, like 1/nth octave, technically one still needs to agree on the pass-band edge frequencies first (perhaps less critical if n is large, i.e., 24 or 48). BTW, I've been using REW and have been unable to confirm what its non-smoothed data output corresponds to. If it's described in the REW user manual, I've not been able to find it. However, the ARTA manual does discuss this, starting on page 35, including how it does its octave bin resampling:
 

 
I think what's needed  - at least as a starting point - is to figure out if we're actually putting the same thing on all our respective y-axes. Do you folks know exactly what output all these various iOS apps are creating?