ECM8000 microphone measuring techniques and usage discussion

[Anthony]

But still, you figure the ECM would show a better top octave than the RS meter and yet they still both show that steep drop above 15kHz.

Hmm, I think my next project may be to build one of those condenser mics from scratch and see how it performs. If for no other reason, to learn more about what goes into all this. I've had enough tweeter crossover problems that I want to make sure I have something I trust up there (and I don't have $500 for a EarthWorks )

[brucek]

So I believe we've ruled out the mic, and via nearfield measurements ruled out the speaker. It's not the calibration or loop, but it seems to be the soundcard.

I don't see how it could be the soundcard if your loopback measures flat to 20KHz. That completely eliminates the soundcard. It's the only thing you can be sure of...

I don't think you can be 100% sure of the mic until you have it calibrated.

How have you ruled out the speaker?


brucek

[hifisponge]

I just find it to be too much of a coincedence that both anthony and I are showing a steep roll-off in the HF above 15KHz. I show it when using both an ECM Mic and DBX mic.

Here are a series of nearfield plots I took using the DBX mic. The different plots represent on-axis, slightly above, slightly below, and to the left and right. (I was trying to find some axis, any axis, that I would get flatish treble response above 15KHz).



I'll be getting my calibration files for my DBX mic this week, but Cross Spectrum has already sent me a preview of the plot, and there is no roll-off in the HF, so I don't expect the cal file to fix this problem. In fact the DBX mic's response looks very similar to the ECM, just with less of a hump at 10KHz.

[hifisponge]

Anthony -

I'm not sure that the Maggies are a good reference for this test. I've read in the past that it is difficult to get an accurate FR reading of panel speakers because of the way they project sound.

For reference, I looked for FR measurements done by AV mags on the maggies and found these:

Magnepan MMG W (measured by Ultimate AV Mag)


Magnepan MG 3.6 (measured by Ultimate AV Mag)


As you can see, they are showing a fairly steep roll-off in the upper HF. Do you have any other more conventional speakers around to test?

Here's what my speakers are supposed to be doing in the HF.

Wilson-Benesch speakers (measured by Hi-Fi News UK)

[Anthony]

Heh, I wonder if Ultimate AV is using our mic

That's a good point, though, about testing other speakers.

I have some Adire HE10.1's (don't know why that number is in both of my speaker sets ) in the tube system upstairs. I can bring one downstairs and remeasure.

Panel speakers radiate in a "middle finger" lobe pattern (for lack of a better image). The on-axis is very strong but narrow then flanked by weaker lobes around the 15 to 30 degree mark, then another null and another set of even weaker nodes around the 45 degree mark. Overall it looks like the classic figure 8 pattern of a dipole, but with nulls inbetween the fingers. A neat trick to getting a great center image is to aim the speaker so the second lobe is aimed at the listening position (about a 30 degree toe-in). It's weird but it keeps a pretty stable center image across the stage without the dreaded "sweet spot".

Anyway, I was measuring about 8 inches from the panel, which should be enough for the ribbon and panel to integrate, but you are right, all it takes is for the mic to be in the wrong null and the FR goes all wacky.

I'll try the other speakers tomorrow night. I can also try a raw BG Neo3PDR and Neo8PDR tweeter/panels to see how they fare.

Thanks for the input guys, it'll be great to get to the bottom of this.

[Anthony]

Well, I just looked through some old measurements and this is not apparent in all of them -- just the ones of the Magnepans. Some random measurements don't seem to have it (like almost touching the panel nearfield), but almost all of the 1m to 3m measurements do. Hmm, I guess I started with a bad assumption, since the speaker was supposed to be flat to 24kHz. But dipoles are tough to measure, tweeters are tough to measure, and in-room is tough to measure. So dipole tweeters in room must be really tough to measure

My guess is it's some sort of cancellation effect from position or frame/panel. Sorta like a broad comb filter effect. It's too high to be dipole cancellation and the likelihood of me putting the mic in an exact null every time is slim. I'm just thinking out loud here to try and find a solution that fits your problem as well, Tim.

I'll test those horn tweeters tomorrow and check back. Seems we're at least a bit closer.

[hifisponge]

Hmmm ....

I ran some more measurements, and I don't know what I did differently but I'm able to get respectable HF extension when I use the mixed cal file for the ECM mic, BUT only when I point the mic at the speaker .

Yellow plot is horizontal mic position, blue is vertical.

[robbo266317]

I have been prototyping a new setup, which is why I found this forum. I have never had a tool like REW before! All I can say is WOW. Anything else I said would ensure that the moderators would delete it.
Anyway, I would not be in a position to comment without REW, So...
I couldn't get the ECM8000, However the local shop did have a JBL re-badged DBX TRA M for a similar price. So I bought it and a small mixer to supply power and set about measuring my new speaker. (singular)
I have been extremely happy with the JBL as my Shiva X measures (close field) to what was modelled in WinISD for a 120 litre cabinet. ie ~7 Db down at 20Hz.
As for the top end, this mic happily did a good job of following the Raven 3.2's response with no major anomolies.
If you cannot find the ECM then I can quite happily suggest using the JBL.

[brucek]

JBL re-badged DBX TRA M for a similar price.

What microphone calibration file are you using for it?

brucek

[Anthony]

Nice measurements there. Interesting that the JBL tweeter has that top octave falloff as well -- not exactly a cheapo tweeter in its own right.

I'll post mine tonight in this exact same format. You said 1 foot, 6ms gate (from zero, or from first impulse received?). The candidates will be the MG10.1 ribbon tweeter and the compression tweeter used in the Adire speaker. I do not remember the brand name on it, but it screwed into the back of the Eminence woofer.

[brucek]

from zero, or from first impulse received?

In the IR window popup (after the measurement is complete), I set 1msec in the Left Pre-Window and 5msec in the Right Post-Window, and then Apply to the measurement.

brucek

[robbo266317]

The one from the DBX RTA-M usable thread http://www.hometheatershack.com/foru...ion-curve.html

Here is a screenshot:

[Anthony]

Okay, my results:
Red = Maggie, Vertical Mic
Green = Maggie, Horizontal Mic
Orange = Adire, Vertical Mic
Blue = Adire, Horizontal Mic

Attachment 11936

Gates were 6 ms, "Split" ECM8k cal file, 12" from tweeter on axis. 1/2 octave smoothing. Shifted to line up like curves.

You can see the trademark split between horizontal and vertical orientation of the mic in both sets of curves.

However, most notably, the Adire HE10.1 does not show the fall-off that the Magnepans show. Very interesting. I stared with a bad assumption and look where it got me

So at least now I have renewed confidence in the mic and soundcard, but less in the Magnepans

[Anthony]

So, it looks like the New ECM cal file, horizontal orientation is the best for direct measurements.

Behringer insists that vertical is better for diffuse field measurements like pink noise and such, but since 90% of my measurements are drivers and speakers directly, I will stick with horizontal.

[brucek]

Yeah, I think we can assume the ECM and the cal file are fine.

The Maggies still go fairly high, and unless your hearing is really top notch, I doubt you'd be disappointed in the sound.

The Adire measures quite good for sure....

Maybe we should be listening to music instead of swept tones.....

brucek

[Anthony]

Bruce, to me that is music!

[hifisponge]

Bruce & Anthony -

A big thanks for helping me troubleshoot, and for all of the effort you two have put into this series of experiements. I'm still waiting for my DBX mic and cal files to come back from Cross-Spectrum and I will post my results when I get those. In the meantime, I think I'll do another run with the ECM and split cal file.

Bruce, can I ask a favor? Would you please retest at 3 feet (horizontal mic) to see if that induces some roll-off in your Proacs? And would you also measure on the midrange axis as well? You're Proac's use a tweeter that is similar to mine and I want to see if distance affects the HF response and also what sort effect being off-axis has.

Thanks,

- Tim

[hifisponge]

Robbo -

I have the DBX mic, and I just had it calibrated. I don't think that cal file listed in the link you provided is all that accurate. It only shows about a 1.5dB rise in the treble response of the mic, when it is really more like 5dB.

- Tim

[robbo266317]

Hi Tim,

The close mic scan of my raven 3.2 is fairly flat from 1k to 25k. (As per it's measured graph that came with it)
I can repost just the 1k to 25k graph if you like.

[hifisponge]

Looking a little deeper at your guys' graphs vs. mine, both of you are showing a difference of roughly 6dB at 20KHz in your horizontal vs. vertical mic positions. This difference coincides with the differences shown in the FR plot of the ECM mic. My measurements show a difference of 12dB at 20KHz. What's up with that?

[brucek]

hifisponge wrote:

..........can I ask a favor? Would you please retest at 3 feet (horizontal mic) to see if that induces some roll-off in your Proacs? And would you also measure on the midrange axis as well? You're Proac's use a tweeter that is similar to mine and I want to see if distance affects the HF response and also what sort effect being off-axis has.

Well, you got lucky, since I hadn't returned my ProAcs to their regular spot in my main system, I was able to do another test for you.

Anyway, I thought I would divide your test in two.

I had a couple of small graphs of my particular speaker model, that were done in an anechoic chamber at 50" (reasonable far-field measure). I thought I would try and duplicate the test conditions for those two graphs.

The first graph is the tweeter and mid-range individually tested on-axis at 50" and shown separately on the small graph.

The second is a measurement of the mic placed first between the tweeter and mid-range at 50" on-axis and then at 30 degrees off-axis in the horizontal plane. The plot is averaged across the 30 degrees horizontal window.

It seemed reasonable to carry out these two tests.

The graph below is the anechoic graph showing ~1KHz to 30KHz.
This is the on-axis, taken at 50" of the 1" SEAS tweeter and the on-axis of the 7" ScanSpeak mid-range driver. Note the rise that I experienced when I measured my tweeter at 12".



The graph below is the anechoic graph showing the average across a 30 degrees horizontal window, measured at a point vertically equal between the tweeter and mid driver at 50".
The comments about the graphs in the article says, The tweeter has a gently rising on-axis response all the way to 30kHz, but it is very directional above 15kHz. This results in a more or less flat response when averaged across a 30 degrees horizontal window.



I think my measurements below support what the anechoic responses show, in that the tweeter is certainly good past 20KHz when on-axis, but drops off at about 15KHz when off axis in the vertical or horizontal plane. This may be what you're experiencing somewhat, where you're suspecting your measurement equipment. But the way to be sure is to measure near-field on-axis, and that will prove it if the response doesn't drop prematurely.

Anyway, below is my first graph, measured at 50".
The mic is on-axis with the tweeter for the purple line, and then horizontally moved to 30 degrees for the green line (with the mic still pointing at the tweeters center).
I also shortened the gate to 3msec to remove some reflections from the stand or wires or whatever it was. I did check the gating envelope in each case on its impulse response. All these are plots with no smoothing.
You can see the drop off when it's off-axis.



Below is simply an average plot of the above two measures (an REW feature allows averaging of multiple plots).
It shows that there is a decent response to 20KHz within a 30 degree horizontal window of the tweeter.



Now I do the same experiment, except the mic is moved down to point at a spot equidistant between the mid-range driver and the tweeter.
Note that the vertical off axis from the tweeter creates a drop off at both on-axis (horizontally) and at 30 degrees (horizontally).



This is simply an average plot of the above two measures. It shows that there still is a fairly decent response to 20KHz off axis within a 30 degree vertical or horizontal window of the tweeter.

My measurements show a difference of 12dB at 20KHz. What's up with that?

If you examine my two plots, I measured a 12db and 10db difference.

brucek

[hifisponge]

Bruce -

You rock!

Thanks a million for taking the time to conduct all of those measurements. They explain a lot and put my mind at ease. I also did some digging around for graphs of the ScanSpeak tweeter used in my speakers and found that that they all showed about a 12-15dB drop above 15KHz when measured 30 degrees off axis.

The difference between the tweeter in my speakers vs the one in yours is that it does not have rising response on-axis. In fact, based on measurements I did last night, it is about 3db down at 20KHz on-axis, so the off-axis roll-off above 15K looks more pronounced.

My measurements still don’t quite match those published on the ScanSpeak tweeter, but they do appear to be within a reasonable deviation.

Thanks again.

- Tim

[hifisponge]

Robbo -

I was under the impression that you were using a JBL / DBX mic with a cal file that was published on this site. If that is the case, I was only suggesting that if you want accurate high-frequency readings from that mic that you should have a cal file made for your specific mic.

- Tim

[hifisponge]

brucek wrote:

That seems consistent with the graph of mic orientation versus response...



brucek

Bruce, Anthony -

I got my DBX mic back from Cross-Spectrum Labs today and the response graphs the provided explain at least part of the roll-off I was getting when I did my measurements with this mic. You see, I was doing all of my measurements with this mic in the vertical position, as I thought that was the way to get the flattest response from an omni mic. As it turns out, the DBX mic is 6dB down at 20KHz when used vertically, with a +2dB hump at 11KHz. In the horizontal position there is a 5dB peak at 11KHz and it flattens out by the time it reaches 20KHz. I would post the FR plot, but I was only given a paper graph.

For the record, the cal file I have now is significantly different that the one that was posted by another user in this forum, and neither is good to use when the mic is in the vertical position.

Cheers,

- Tim

[brucek]

and neither is good to use when the mic is in the vertical position.

It's my understanding that your mic was calibrated to be used in the vertical position, when placed in a room in a diffuse field (such as the listening position). That would mean it is accurate to rely on the results if you use the cal file in REW under those normal conditions.

The only time it would be inaccurate is when you take a close speaker measurement of the tweeter, where you need to point the tip horizontally at the speaker, and so the cal file is not accurate for that upper octave. Is that not a correct assessment?

Comments?

brucek