Measuring harmonic distortion with the ECM8000

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Jump to Latest
owdi

owdi

Member
Joined 2003
I want measure harmonic distortion using my Behringer ECM8000, but I'm afraid the microphone itself introduces distortion, causing invalid results.

Does anyone know what maximum SPL the ECM8000 can measure before distortion starts quickly increasing? Has anyone measured the harmonic distortion introduced by the ECM8000 itself?

I remember reading that the ECM8000 had an upper limit of 92db, but I can't find that reference now.

Below is a nearfield measurement of the Vifa XG18 in an OB. I don't trust the numbers because the noise floor in my room is above the measured distortion.

Dan

An externally hosted image should be here but it was not working when we last tested it.
 
ShinOBIWAN

ShinOBIWAN

diyAudio Member
Joined 2004
Trust me when I say that even a budget mic such as the ECM8000 isn't going to meaningfully influence the distortion measurements taken from a loudspeaker driver.

Just as bit of name dropping to further reinforce that, I believe Zaph uses an ECM8000 for all his testing.

The Behringer will also happily measure SPL levels at 120dB, if really high SPL measurements are needed then take a look at Earthworks range or even those used for car SPL competitions.
 
owdi

owdi

Member
Joined 2003
The Behringer will also happily measure SPL levels at 120dB, if really high SPL measurements are needed then take a look at Earthworks range or even those used for car SPL competitions.
Click to expand...

Well I'm very happy to hear that. I live in an apartment where there is always some type of noise. If I want to measure harmonic distortion levels below -60db, then I need the fundamental to be at least 100db.

I have a generic calibration file for the ECM8000, which I use with Speaker Workshop. I'd like to use it with Arta and Steps, does anyone know how?

Dan
 
Svante

Svante

Member
Joined 2004
One way of testing this would be to measure the same loudspeaker at two different distances, using the same voltage to drive the loudspeaker. If the frequency response remains the same, room effects can be neglected. If the relative amplitude of the distortion components change, there must be distortion components of the microphone that cause the change.

The measurement should be done with a small loudspeaker driver, preferrably mounted in-wall, such that it approximates a point source and has no baffle effects. I predict no large distortion differences between low and highish (<4 kHz) frequencies, so it might be easiest to test this with a mid driver.

Maybe...

Edit: Yet another way: Put the microphone inside a loudspeaker box. For low frequencies, the level easily becomes very high (140 dB or so). This means that if the level is lowered, the signal is almost distortion free. The remaining measured distortion most certainly comes from the microphone, at least for a cheap microphone.
 
owdi

owdi

Member
Joined 2003
Svante - I'll try that, although it will be hard to discount room effects. A reflection that is -30db will not show up in the FR, but could affect the readings.

ente - thanks for the link.


I have another question, and don't want to start a new thread. I want to measure the harmonic distortion of my Seas 27TBFC/G tweeters. Can they survive a test at 2.83 volts, without a crossover?

Dan
 
Svante

Svante

Member
Joined 2004
owdi said:
Svante - I'll try that, although it will be hard to discount room effects. A reflection that is -30db will not show up in the FR, but could affect the readings.
Click to expand...

Hmm, I was under the impression that the software that you use gated the impulse response. In that case, room reflections should not be an issue, at least not for f>~500 Hz.
 
owdi

owdi

Member
Joined 2003
I'm using Steps from the Arta suite of software. I don't believe this software takes gated measurements, but instead, relies on nearfield measurements. Why do you think room reflections will not be an issue above 500hz?

If the mic is 5mm from the driver, and the nearest reflective surface is 1m, the reflection could be as much as -26db. Intuitively this seems like a problem, but steps must have a way of handling this. It measured the XG18's harmonic distortion down to -80 db consistently with other measurements I've seen on the net.

Assuming reflections are not an issue, my next concern is the mic. I am very interested in how distortion changes at various power levels, and would like to measure the Vifa XG18 at .1 watt, 1 watt and 10 watts. To get an accurate nearfield measurement at 10 watts, the ECM8000 will be subject to nearly 120db. ShinOBIWAN believes it is capable of doing this. I guess I'll just have to try it.

On a side note, I finished the cutouts for the second baffle today :) Happy day, it's almost time to take measurments and design the crossover. Here's a pic.

FrontAndBack.jpg


Dan
 
Svante

Svante

Member
Joined 2004
owdi said:
I'm using Steps from the Arta suite of software. I don't believe this software takes gated measurements, but instead, relies on nearfield measurements. Why do you think room reflections will not be an issue above 500hz?
Click to expand...

If a gated approach would be used, reflections coming from the walls would simply be gated away. That is, the software would calculate an impulse response, which is truncated before this first reflection. The time between the direct sound and the first reflection would set lower limiting frequency, which is typically in the 500 Hz range.

owdi said:
... the ECM8000 will be subject to nearly 120db. ShinOBIWAN believes it is capable of doing this. I guess I'll just have to try it.

Dan
Click to expand...

I don't think you could rely on a distortion measurement @ 120 dB with the ECM 8000.

The easy way to test that is to put the microphone inside a loudspeaker box, adjust the frequency to ~40 Hz and the level to level to 120 dB (which typically corresponds to some 80 dB outside the box at 1m) and see what amounts of distortion the signal contains. I bet the distortion levels will turn out a lot higher inside the box, and this is due to mic distortion.
 
ShinOBIWAN

ShinOBIWAN

diyAudio Member
Joined 2004
Svante said:


If a gated approach would be used, reflections coming from the walls would simply be gated away. That is, the software would calculate an impulse response, which is truncated before this first reflection. The time between the direct sound and the first reflection would set lower limiting frequency, which is typically in the 500 Hz range.
Click to expand...


Truncating distortion measurements would lead to inaccuracies across the whole range of measurements with a progressive effect as you moved down in frequency.

Its why you never see it in a package. If your taking distortion measurements then semi-anechoic is your friend here or nearfield + arithmetic to adjust the relative SPL for more realistic listening distances when doing in-room.

I don't think you could rely on a distortion measurement @ 120 dB with the ECM 8000.

The easy way to test that is to put the microphone inside a loudspeaker box, adjust the frequency to ~40 Hz and the level to level to 120 dB (which typically corresponds to some 80 dB outside the box at 1m) and see what amounts of distortion the signal contains. I bet the distortion levels will turn out a lot higher inside the box, and this is due to mic distortion.
Click to expand...

That testing methodoloy won't work because there will be a ton of distortion from resonances in the box to start with. I've used/hired the Earthwork M50 which is one of the best. Measuring port output revealed a good amount of distortion, was it the mic? No, same thing applies here.

I had the same concerns as Owdi a couple of years ago and because of that I measured the ATC midrange in the nearfield(2" away) at an SPL of 120dB and it scaled well with the ones at 90, 100 and 110dB indicating that mic distortion was not the issue, at least in the 200-7000hz range I used for the sweeps.
 
Svante

Svante

Member
Joined 2004
ShinOBIWAN said:


Truncating distortion measurements would lead to inaccuracies across the whole range of measurements with a progressive effect as you moved down in frequency.

Its why you never see it in a package. If your taking distortion measurements then semi-anechoic is your friend here or nearfield + arithmetic to adjust the relative SPL for more realistic listening distances when doing in-room.
Click to expand...


Well... Such a method is actually described in the literature. It uses log sweeps which are used to derive an impulse response. It turns out that distortion components appear as separate impulses in the impulse response, and these can be used to measure the distortion as a function of frequency. Sure, it has all the restrictions gated measurements have (resulting in a lower frequency limit around some 500 Hz, but I have seen it working).

Here is a link to a swedish forum that shows a matlab implementation of the method. In spite of the language, I think one can tell from the images that it works.

Trolldist

ShinOBIWAN said:


That testing methodoloy won't work because there will be a ton of distortion from resonances in the box to start with. I've used/hired the Earthwork M50 which is one of the best. Measuring port output revealed a good amount of distortion, was it the mic? No, same thing applies here.
Click to expand...


Resonances are linear phenomena that not inherently produce distortion. Apart from that, there are few resonances at 40 Hz inside a box. Possibly a poor vent can cause distortion, but my point is that it is very easy to acheive low-distortion high SPL inside the box. There is an acoustic gain of 40-50 dB, typically at low frequencies, compared to the level at 1 m.

The earthworks mic is a very good one, and it might actually have lower distortion than the loudspeaker produces, even inside the box. But we (or at least I) were talking about an ECM 8000 here, and until someone has compared measurement inside and outside a box with similar distortion, I vote for a significant distortion contribution from the ECM 8000 at 120 dB.

ShinOBIWAN said:

I had the same concerns as Owdi a couple of years ago and because of that I measured the ATC midrange in the nearfield(2" away) at an SPL of 120dB and it scaled well with the ones at 90, 100 and 110dB indicating that mic distortion was not the issue, at least in the 200-7000hz range I used for the sweeps.
Click to expand...

With an ECM 8000? How much distortion did the loudspeaker produce?
 
owdi

owdi

Member
Joined 2003
Gave this some more thought, and I think finding the limit of the ECM8000 will be easier than I thought. All I really need to do is start at 1m, take an ungated measurement, then move the mic closer and repeat. As I move the mic towards the driver, the fundamental will be louder relative to background noise and reflections, so measured distortion should fall. However, once I move the mic so close it's pushed beyond it's limits, distortion should start rising again.

Dan
 
J

Jmmlc

R.I.P.
Joined 2005
Hello,

I don't know if your ECM-8000 use an electret capsule.

Most often electret capsule as the MCE2000 possess an integrated fet inside.

THis fet is used at very small drain-source current in an operating zone called "quadratic", this means that it's transfer function is in power 2 creating a second order (H2) distortion.

You can find distorsion for different microphones in the document

http://www.fesb.hr/~mateljan/arta/AppNotes/AP5_MikroMeasChamber-Rev03Eng.pdf

give a look to figure A1 page 15 (last page).

Best regards from Paris

Jean-Michel Le Cléac'h
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Top Bottom