Could it be caused by the seal of microphone capsule of the SPL meter PCE 322-A, that the mic capsule is not properly sealed to the nozzle of the meter?
I have 2x UMIK-1, when new both mic capsule were not sealed properly and causing bass response to drop when using for IEM measurement with coupler. I have to open the tip of the UMIK-1, applied some glue to properly seal the mic capsule, only after that it can be used for IEM measurement with coupler.
I also spent some time to get a proper seal and location of the mics inside the couplers. At last, found that the best and easiest seal and right location was achieved by using tight silicone sleeves.
Tested it with my SPL calibrators as well: used same sleeve when inserting the SPL meter mic into the 12mm hole of the calibrators.
You can check this pics about adapting non-threaded 6mm mics and the mic of SPL meters into 12mm couplers.
When you mentioned 439mV measured at the right driver, is that means at this voltage across the right driver, and you plugged the IEM to the coupler, your SPL meter + IEC coupler will measure 106 dB SPL @ 500 Hz?
Did you measure the voltage when it was measured 94dB SPL @ 500 Hz?
Thanks!
Yes. Generated a 0dBFS (100%) 500Hz sine tone in audacity. Played it while measuring voltage across the right driver (+cable), which was attached to the SPL meter.
Adjusted output volume till I got a 106dB reading in the SPL meter. Then, watched the voltage reading. It varied from 423 to 441mV, depending of insertion depth. Resolution of the SPL meter is 0.1dB (but error/tolerance is another matter). I said 439mV, because that was the reading when the insertion depth was approximately the same than used in posterior measurements.
[I use very low resistance leads and adapters with the DMM, around 10mΩ total, despicable when compared to a 15Ω IEM]
Aimed to 106dBSPL, because the sweep level I use when measuring at REW is -12dBFS (and I mark "Full scale sine rms is 0 dBFS" checkbox at Preferences-View, to avoid further misunderstanding). So the SPL when measuring at REW will be 94dB SPL at 500Hz.
You can calculate the voltage to get 94dB SPL @ 500Hz from the voltage measured at 106dB SPL.
V2= V1 * 10 ^ ((dB2 - dB1)/20)
V@94dB = 439mV * 10^((94dB-106dB)/20) = 110.3mV
In fact, it wouldn't be necessary at all, but I use to measure voltage at 90, 94, 96, 100, 102, 106, 108 (and 112 and 114 if the IEM allows it) when calculating sensitivity, to get an average value and to check that everything is ok.
@hakuzen I can't help for your variations. I cry every time I look at Head-fi's measurements of an ER4S and then look at my measurement of my pair on my cheap rig. And I'm confident you will cry too once this IEM reaches Jude. Or maybe if we ruin the IEM with violent shocks and overly loud signals, he'll end up measuring worst THD? See there's hope already. ^_^ from your usual experience, wouldn't this tiny amount of THD variation be found when you measure an IEM a little quieter or louder? Perhaps your suspicion of issue on output calibration could indeed explain both your problems. But that's only one guess, there are just too many things that could affect THD measurements.
LOL, it's a real pleasure to read your posts. Your continuously ascending humor sense reflects your evolution degree. Thanks!! Keep on it, please.
I'm not worried at all by the disto variation, but you know I try to be perfectionist when doing sci stuff and possible.
Yes, after checking source and soundcard preamps distortion levels, I guessed that the THD variation could be caused by different volume level when measuring. That's why I dedicated a big part of my post to discuss about the SPL calibration methods we use. I was going to measure THD at different levels to confirm it when the mic preamp got fried.
This is important when dealing about THD. If we don't have well calibrated SPL meters, but can measure the applied voltage, we could agree on measuring at X mVrms, instead of X dBSPL.
And to calculate first that Vrms value needed to get 94dB @ 500Hz, we can use the official sensitivity and dB response difference between 1kHz and 500Hz. It doesn't matter that this value is very exact. It can be approximated, but we'd use very similar volume level when measuring, avoiding SPL calibration issues.
The other possible reason, based on my ignorance about mics, could be the mic itself and/or mic preamp. I'd have to check THD measurements of our database further to get an idea. It's probable to find notable differences between cheap electret, cheap condenser, and precision condenser mics. Focused csglinux measurements, because they were made using a total reputable precision condenser mic from GRAS, and was intrigued by the difference with my Chinese precision mic.
Awesome work,
@hakuzen!
I don't see there's any rush for you to send the package on (unless somebody else out there is absolute dire need to receive it very quickly?). Right now, most of the world is either in lockdown or about to be, so as far as I'm concerned, I'd say spend your quarantine time doing something more interesting and less stressful than reading the news
I'm sorry to hear about your mic pre-amp. Hope you get up and running with a new one soon.
These results look very good. Most of us are seeing fairly broad agreement with these generic 711 couplers, which is really encouraging. On each individual measurement issue...
Calibration: I know Etymotic used to be famed for their high precision, but I think the idea of the ER2 line was to outsource the manufacturing, make them more cheaply, and one of the costs cut was the (previously, i.e., ER4 series) careful calibration and pairing. We may have been lucky that this particular set was quite well balanced. We both saw ~0.1dB difference in L and R channels 94 dB @ 1 kHz. (Left channel being ever so slightly louder than the right). That's a really, really good tolerance. I recently auditioned a $3000 IEM whose channels were off by more than 2dB under the same conditions. Even my ER4XR's official graph/certificate specifies a 0.4 dB difference between L and R at 1 kHz @ 200 mV. I didn't actually try to measure the ER2SE's SPL at 100 mV, but I've no reason to assume your measurements are off. I'd bet there's some leeway in Etymotic's officially-quoted product specs.
That's a good thought. I guess it's possible. A significant bass roll-off would be easy to see when you're measuring the whole spectrum, but not so obvious if you were only looking at one number. Still, I suspect Etymotic's tolerances aren't all that close. Maybe the next person in the tour can double check this? But regardless, I think within one decibel is actually pretty good and almost certainly enough for our purposes.
Impedance: Looks identical to my measurements
Frequency response: Looks very close to those from a GRAS coupler, even around the 1 kHz region. Nice job
As for the eartip hitting the coupler mesh/grill, I've experienced the same problem. It's probably easy to do this and not even realize it's happened. I went for a slightly shallower insertion (dropping the primary canal-resonance peak to around 10.5 kHz) to avoid this, but I was still able to get a seal. Other brands of coupler might, perhaps, have shorter canal extenders? The main differences in FR appear to be the secondary peaks around 15-16 kHz. In my experience, these are the most sensitive to the eartip type and size. I believe certain eartips (small frost triple flange being one) won't necessarily seal in the coupler canal properly at the first flange, but only at the second (or third?). This might result in a small Helmoltz cavity in the gap(s). It's likely this happens in certain people's ears too. I've seen huge differences in the ER4XR measurements in the 13 kHz+ region just by switching from small to large flange tips. There's an audible difference there also - unfortunately, none of these triple flange tips are much use to me on a long-term basis, because I just find them too uncomfortable. One way or another, I think we have to keep the eartip from touching the coupler grill.
Impulse response: Looks identical to that from a GRAS coupler. Nice
Distortion: There may be differences in quoted THD specs for various input/output soundcards or ADCs/DACs, but these are all still likely to be an order of magnitude better than the distortion in most IEMs. I'm almost sure the differences here are in the couplers' microphones. I've recently spent more time and money than any sane person would looking into this. While condenser mics generally seem to have way lower distortion than the cheaper 3.5 mm dynamic/electret mics, there's still a bit of spread. The two mics I've found with the lowest distortion are the GRAS 40AO (plus 26CA pre-amp) and the Earthworks M23. But these are both pricey. Even the M23 is ~$500 and still needs to be carefully fixed to a coupler. (In my experience, although they have very low distortion, Earthworks' mics aren't quite as flat as their specs claim.) The coupler response will also have a small effect, because if certain resonance peaks are more pronounced in one coupler than another, that will also give rise to higher THD at the lower frequencies that would look like the fundamental to some higher-order harmonic. For example, the GRAS hi-res couplers (RA0401 and RA0402) will report artificially low values of THD across the board, because they intentionally miss the ear canal resonances. I recently made some measurements of the Beyerdynamic Xelentos and the Final Audio A8000. Both seem to have the same total distortion characteristics at 80 dB. Is that likely? I doubt it. I suspect I've just hit a floor which is the THD of the coupler mics:
Although it's still fun to look at, does THD really matter? Really? (A lot of folks are going to plug their headphones into a tube amp anyway.) I have a feeling FR is way more important than THD. Like 100% more important
Stay safe everybody. Social distancing. You have a great excuse right now to stay at home and listen to some music
Thank you so much!! =))
I wanted to provide a full study like yours after so long, but couldn't due to the mic preamp issue. The new one will be shipped from China tomorrow, using a fast shipping carrier. If no problems with customs, it should arrive soon.
I'm working only 7 hours daily since this week (compared to 10 hours of last months) plus 3 commuting hours. So I'll have some more time from now on to dedicate to it.
Agree 100% with your observations. They are spot on!
Noticed the differences between old and new GRAS couplers, thanks a bunch for providing measurements from them!
Yes, the response (FR and THD) of resonances have varied a lot between them. And the mic could be a possible reason of the THD overall level discrepancy. Together with SPL used when measuring, as discussed above.
"A lot of folks are going to plug their headphones into a tube amp anyway" LOLOL. Yea, all this is not so important, but we can know the weaknesses of our rigs thanks to this tour, and get help to better calibrate them; so we can be a bit finicky. The work doneby all the participants of this tour is amazing. Thanks to all, very well done. Yuriv study is ultra complete. But personally have to remark your measurements made with GRAS top quality stuff; I couldn't dream of being able to compare to my rig by measuring the same exact IEM. It's a reference and can help to improve my rig calibration.
@csglinux thanks for the write up above!
May I know what's the reason for measuring THD at 80 dB SPL? To be honest I don't know much about standard, I only know about IEC 60268-7 mentioned test level @ 94 dB SPL @ 500 Hz. What is the agreed SPL level for IEM THD measurement?
Btw, i agree with you, the mic plays a big part in THD measurement, so with the generic 711 couplers and mics, it is better not compare the THD measurements.
Btw, going back to coupler mic SPL calibration for IEM measurement, with the generic 711 couplers and mics, for non highly accurate measurement, like mentioned above, is it a good idea to use Etymotic IEM sensitivity specs to calibrate the SPL level of the generic 711 couplers? Or using SPL meter + 711 coupler like what
@hakuzen did is more accurate? Or is there other budget friendly way to do it?
I don't understand of the definition of IEM sensitivity specs, is the SPL value is reading at a certain distance from the IEM nozzle, or is it SPL reading on industry standard coupler + calibrated mic? Appreciate anyone who can share their knowledge here.
Btw, so far I calibrated my generic 60318-4 compliant coupler (+ mic) using Etymotic ER2XR:
Sensitivity (@1 kHz) SPL at 0.1v : 96 dB
https://www.etymotic.com/er2-new.html
I use REW:
1. Set playback tone: 1kHz.
2. Split the headphone output to ER2XR and True RMS voltmeter.
3. Set REW playback volume to ER2XR at exactly 100 mV / 0.1V measured at voltmeter.
4. Calibrate the SPL reading on REW SPL Meter to 96 dB SPL.
For audio interface I use RME Babyface Pro + Rode VXLR+. Settings on the RME:
RME Babyface Pro 3.5mm output: 0 dBFS
RME Babyface Pro In 1 XLR + Rode VXLR: Gain 20 dB (PAD Off)
We could use a voltage reference instead of SPL reference, as discussed above. To determine that voltage reference, we could use official sensitivity, or SPL measured by a well calibrated meter (not much difference probably).
I think that the right way to calibrate REW SPL meter without having a calibrated SPL meter could be:
1. Set the voltage needed to achieve 106dBSPL in the IEM (if your sweep level when measuring will be -12dBFS, so you'll measure at 94dBSPL).
2. Adjust input preamps and level to get -18dBFS in REW (Preferences - Soundcard - Check levels), if you are going to use that input level when measuring, as suggested by REW.
3. Run REW SPL meter calibrator. Before introducing the value, check the level (dBFS) of the calibration signal (showed into calibration window). If it is -18dBFS, for example, you'd have to enter 106dBSPL-18dBFS = 88 dBSPL, because 0dBFS is 106dBSPL. If anyone find that this is wrong, please correct me.
You can use that SPL calibration while you don't touch input levels (preamps, etc.). If you change them, you'll have to calibrate REW SPL meter again.