Etymotic Research EVO impressions and discussion thread
Jun 14, 2021 at 12:44 PM Post #601 of 1,201
OK, this is just a single pair, not a high n overall average, but it's fairly representative. I was hoping to post this last week after my conversation with Zachi, but I didn't get to it.

For fun, I took data from four different ear simulators. They include:

GRAS RA402 - 1/2" mic, high-resolution coupler. 2-branch coupler rated to 20kHz. Suppressed peak at 13.5kHz

GRAS RA404 - 1/4" mic, high-resolution coupler. 2-branch coupler rated to 20kHz. Suppressed peak at 13.5kHz

Zwislocki DB100 - 1/2" mic, 4-branch coupler rated to 10kHz. This particular coupler has a higher than average spike at about 14kHz

GRAS RA0045 - 1/2" mic, 2-branch coupler rated to 10kHz. This is the industry standard for production testing, although the new high-resolution variants may eventually supplant it.

Historically, we've always used the Zwislocki as it's a bit better for recreating the impedance of the human ear (up 10kHz at least), but the increased bandwidth of the new GRAS couplers is compelling.
I wonder how it compares to ER3/4? Looks to roll-off fairly early in the highs.

Maybe it's the scale, but upper mids is pretty emphasized.


In regards to precog's measurments. His insertion depth was too shallow. The insertion depth moves the peak with the clone couplers.
 
Last edited:
Jun 14, 2021 at 12:49 PM Post #602 of 1,201
I wonder how it compares to ER3/4? Looks to roll-off fairly early in the highs.

Maybe it's the scale, but upper mids is pretty emphasized.


In regards to precog's measurments. Hus measurement was too shallow. The insertion depth moves the peak with the clone couplers.
Does anybody know if the clone couplers attempt to implement the internal side branching of a 60318-4 or is it just a metal cylinder? I've never been able to find an answer to that but it should have them (and might) if it's actually build to the standard.
 
Jun 14, 2021 at 12:51 PM Post #603 of 1,201
I wonder how it compares to ER3/4? Looks to roll-off fairly early in the highs.

Maybe it's the scale, but upper mids is pretty emphasized.


In regards to precog's measurments. His insertion depth was too shallow. The insertion depth moves the peak with the clone couplers.

EVO has more at 8kHz, less at 10kHz and more at higher frequencies. It's definitely a different response. The primary peak is roughly the same.
 
Jun 14, 2021 at 12:56 PM Post #604 of 1,201
Can you tell us a bit more about the testing setup? I'm curious which ear simulator this is.

Hi Dave, it‘s the same one most reviewers are using, an IEC-711 clone. I'm running it using the Apple USB-C dongle through a TRS splitter. It will have a resonance peak in that region depending on the insertion depth of the IEM in question. Most reviewers are also shooting for an ~8kHz resonance peak for consistency. Something that I might do differently than other reviewers, though, is sine sweep by ear to confirm what I'm seeing. In this case, I’d say the 8kHz peak is real, although the actual amplitude, of course, will not be accurately reflected by the graph due to the resonance peak.

graph-5.png


Above is a graph I plotted of the EVO against the ER2XR with a 10kHz resonance peak for comparability. You can see that the EVO maintains that bump at 8kHz which lends more credibility to it existing. By comparison, the ER2XR does not have a peak at 8kHz. I have verified this by ear in addition to what the graph shows and, at the very least, I think it's safe to say the EVO has more energy at 8kHz compared to the ER2XR.

Of course, everything after ~10kHz cannot be considered accurate. You can see that the EVO graph with the 10kHz resonance peak truncates everything after that point. I think the original EVO graph with the 8kHz resonance peak more accurately depicts what I hear. It is not as dark as what's shown in the 10kHz resonance peak measurement (or at least, I'd hope not!).
 
Jun 14, 2021 at 1:06 PM Post #605 of 1,201
Hi Dave, it‘s the same one most reviewers are using, an IEC-711 clone. I'm running it using the Apple USB-C dongle through a TRS splitter. It will have a resonance peak in that region depending on the insertion depth of the IEM in question. Most reviewers are also shooting for an ~8kHz resonance peak for consistency. Something that I might do differently than other reviewers, though, is sine sweep by ear to confirm what I'm seeing. In this case, I’d say the 8kHz peak is real, although the actual amplitude, of course, will not be accurately reflected by the graph due to the resonance peak.

graph-5.png

Above is a graph I plotted of the EVO against the ER2XR with a 10kHz resonance peak for comparability. You can see that the EVO maintains that bump at 8kHz which lends more credibility to it existing. By comparison, the ER2XR does not have a peak at 8kHz. I have verified this by ear in addition to what the graph shows and, at the very least, I think it's safe to say the EVO has more energy at 8kHz compared to the ER2XR.

Of course, everything after ~10kHz cannot be considered accurate. You can see that the EVO graph with the 10kHz resonance peak truncates everything after that point. I think the original EVO graph with the 8kHz resonance peak more accurately depicts what I hear. It is not as dark as what's shown in the 10kHz resonance peak measurement (or at least, I'd hope not!).
Thanks, that's very interesting and helpful. Just curious, what do you mean that "most reviewers are shooting for an 8kHz peak for consistency"? I am not sure I quite understand what that means.

Yes, there is definitely an 8kHz peak there. I just went through a bunch of data and the average delta between 8kHz and 2.5kHz is roughly 10dB (with the primary peak being higher). If 8kHz is higher than the primary peak, something is likely going on somewhere in the test setup.
 
Jun 14, 2021 at 1:06 PM Post #606 of 1,201
OK, this is just a single pair, not a high n overall average, but it's fairly representative. I was hoping to post this last week after my conversation with Zachi, but I didn't get to it.

For fun, I took data from four different ear simulators. They include:

GRAS RA402 - 1/2" mic, high-resolution coupler. 2-branch coupler rated to 20kHz. Suppressed peak at 13.5kHz

GRAS RA404 - 1/4" mic, high-resolution coupler. 2-branch coupler rated to 20kHz. Suppressed peak at 13.5kHz

Zwislocki DB100 - 1/2" mic, 4-branch coupler rated to 10kHz. This particular coupler has a higher than average spike at about 14kHz

GRAS RA0045 - 1/2" mic, 2-branch coupler rated to 10kHz. This is the industry standard for production testing, although the new high-resolution variants may eventually supplant it.

Historically, we've always used the Zwislocki as it's a bit better for recreating the impedance of the human ear (up 10kHz at least), but the increased bandwidth of the new GRAS couplers is compelling.
I'm more curious when compared to Harman IE
 
Jun 14, 2021 at 1:16 PM Post #607 of 1,201
Just curious, what do you mean that "most reviewers are shooting for an 8kHz peak for consistency"?

The way I understand it, as I've read @crinacle describe it, is that he's going for consistency over accuracy. Consistency = aligning the coupler resonance at 8k across all IEM measurements.

Didn't crinacle upgrade to a GRAS? I'm pretty sure he did.

AFAIK, he uses the new GRAS setup for his headphone measurements and continues to use the old rig for IEMs so that it's compatible with his existing database.
 
Last edited:
Jun 14, 2021 at 1:23 PM Post #608 of 1,201
Thanks, that's very interesting and helpful. Just curious, what do you mean that "most reviewers are shooting for an 8kHz peak for consistency"? I am not sure I quite understand what that means.

I'll steal a quotation from @crinacle and his excellent write-up on the matter:

Now since our ear canals (and in turn, the equipment that simulate it) are essentially hollow tubes, that means that sound waves going through it will result in half-wave resonances. Again, the exact mechanics of this requires a whole physics lesson by itself, so here’s the simplified TL;DR: when measuring IEMs, there will always be a consistent, repeatable “spike” in the higher frequencies. This “spike” is known as coupler resonance and is typically a constant that’s independent from the IEM being measured (assuming consistent methodology).

This resonance can be controlled with a consistent measurement methodology in which the insertion depth of the IEM into the canal is made constant. For my measurements, I have this resonance normalised at 8kHz (whenever possible).

While there are variances between even the same coupler, most of us are going for that 8kHz peak to match up with Crin's database (easily the largest) for the most comparability.

In regards to precog's measurments. His insertion depth was too shallow. The insertion depth moves the peak with the clone couplers.

I'd say the 8kHz resonance peak measurement most accurately depicts what I hear. The EVO uses double-flanges which won't seat as deeply, as something like the triple flanges, in one's ears. Furthermore, you can see that an insertion for a 10kHz resonance peak completely truncates anything after that point, which would make the EVO incredibly dark. And it's not that dark, at least not to my ears.
 
Jun 14, 2021 at 1:35 PM Post #610 of 1,201
Thanks, that's very interesting and helpful. Just curious, what do you mean that "most reviewers are shooting for an 8kHz peak for consistency"? I am not sure I quite understand what that means.

Yes, there is definitely an 8kHz peak there. I just went through a bunch of data and the average delta between 8kHz and 2.5kHz is roughly 10dB (with the primary peak being higher). If 8kHz is higher than the primary peak, something is likely going on somewhere in the test setup.
It's to match up with Crinicle's setup where he adjusts insertion depth to create an 8k resonance. I know others tend to use other things.

I wonder how it compares to ER3/4? Looks to roll-off fairly early in the highs.

Maybe it's the scale, but upper mids is pretty emphasized.


In regards to precog's measurments. His insertion depth was too shallow. The insertion depth moves the peak with the clone couplers.
The upper midrange doesn't seem anymore emphasized than other Etymotic IEMs I've seen. Keep in mind that none of these plots have any sort of compensation on them, so a spike in the 2-3k region is normal. Spikes around the 8-10k range can sometimes be the resonances of the actual coupler itself.
 
Last edited:
Jun 14, 2021 at 1:36 PM Post #611 of 1,201
I'd say the 8kHz resonance peak measurement most accurately depicts what I hear. The EVO uses double-flanges which won't seat as deeply, as something like the triple flanges, in one's ears. Furthermore, you can see that an insertion for a 10kHz resonance peak completely truncates anything after that point, which would make the EVO incredibly dark. And it's not that dark, at least not to my ears.
Well, it is well known that peaks are shifted from differences in depth of the coupler. I have no idea how you've inserted in your ears. Since there is a tour, we'll see how people's general impressions are.

The highs with those 711s are not reliable. The new high res couplers addresses such issues.

Sometimes, reviewers are influenced by measurements, but measurement may not in reality reflect what they are actually hearing. I saw this a lot with Youtube reviewers.
 
Last edited:
Jun 14, 2021 at 1:41 PM Post #613 of 1,201
I'll steal a quotation from @crinacle and his excellent write-up on the matter:



While there are variances between even the same coupler, most of us are going for that 8kHz peak to match up with Crin's database (easily the largest) for the most comparability.



I'd say the 8kHz resonance peak measurement most accurately depicts what I hear. The EVO uses double-flanges which won't seat as deeply, as something like the triple flanges, in one's ears. Furthermore, you can see that an insertion for a 10kHz resonance peak completely truncates anything after that point, which would make the EVO incredibly dark. And it's not that dark, at least not to my ears.

Based on that text, it sounds like crinacle is merely normalizing the graphs at 8kHz, which is perfectly fine. That has less to do with taking data and more to do with presenting data. For example, on the measurements I presented of the four couplers, I normalized the data at 1kHz to make up for sensitivity differences in the individual microphone used.

Yes, couplers do have resonances, but IEC60318-4 has definitive limits up to 10kHz. If this coupler has it's resonance at 8khz, that severely limits its usefulness and likely takes it out of conformance with the IEC standard. The original GRAS and Zwislocki couplers have coupler resonances well past 10kHz. If I order some of the cheaper couplers, we'll run a transfer impedance measurement on the couplers and see how they do.

And, just to be clear, I'm not not in any way discouraging folks from making measurements, inexpensive coupler or otherwise. I think that it's great that folks are so invested in the hobby and it's always great to see folks taking a scientific approach. I can't help but comment if something looks out of place to me; I hope that it is taken as helpful, which is my intent.
 
Last edited:
Jun 14, 2021 at 1:43 PM Post #614 of 1,201
Did you mean the review unit didn't come with the standard triple flange tips or that they are not usable with EVO for you?

I find my measurements match the ones from Etymotic only when using the triple flange tips and inserting till the eartip is at the reference plane.

The EVO ships with 2-flange and 3-flange eartips.
 
Jun 14, 2021 at 1:45 PM Post #615 of 1,201

Users who are viewing this thread

Back
Top