Measurements

[Anthony]

Question for all the builders here:

When you measure a finished speaker FR, where do you put the mic?

I had a hard time getting a representative sweep. I tried quasi-near (1' from the tweeter), listening position, 3' from the tweeter, and 3" from all the individual drivers. the results were "similar", but all over the place, especially with the listening position measurement.

These are open baffle speakers, so a listening measurement is key, but I don't want to adjust the design based on room effects.

What do you all do for your measurement?

[phaseshift]

What are you using for the measurements? What is the stimulus that is fed to the system?

My experience for a DIY setup would be to try and get rid of the reflections by taking the measurement outside or to a huge room. This is particularly true of you have no way of gating the measurement. Even with a good test system, your low end response is going to be somewhat hit or miss due to standing waves and interference in the room.

There is a lot of good info about this in the Loudspeaker Cookbook. That may be the best place to start on a road to a solution that fits your situation.

[Anthony]

I'm using Room EQ Wizard to do the frequency sweep. I have a Behringer ECM8000 mic and an external soundcard that feeds the stereo through the Aux input (no tone control set, crossover disabled)

I have Speaker Workshop as well, but haven't played around with the advanced gating yet (you can gate measurement in it so that the reflections don't affect the results as much).

It's funny, my weird readings at the listening position are in the 1800 to 3k Hz region, not the bass. Close in, they are actually pretty flat. At listening position, there's a broad dropoff shelf that comes back up. It's weird.

[phaseshift]

I would recommend checking the calibration and contacting the guy that wrote the software. He can probably help.

FYI for near field measurements, you should be very close - like 6mm or so.

[Anthony]

Yeah, I know. Problem is that there was too much junk to get the tripod that close. I'll try to get those true near-fields tonight.

I'm pretty sure the calibration is fine, since I get good measurements for other speakers (like my sub), and get some decent results on this one. it's just that it is EXTREMELY sensitive to mic location.

I was just wondering how you all do it (the measurements).

Thanks for the help.

[JCD]

I did a quick search, and found the following at the following link:

Near Field Speaker Analyzing

When analyzing the frequency response of a sound system and/or speaker it is desirable to minimize the effects of the room on the measurement results. Sound reflections from surfaces near the speaker, low frequency room modes, and room absorption effects are examples of room influences that can alter the pink noise response test results. Short of testing the system outside above the ground free of reflections, the next best alternative is to perform a near-field test.

A near-field frequency response test positions the microphone in front and very close to the speaker. The speaker must be placed away from nearby walls or table surfaces to avoid reflections. Place the measurement microphone no more than 3 times the widest dimension of the speaker enclosure. If there are two speakers in the enclosure, position the microphone in the middle of the two speakers and back slightly so the sound has a chance to converge.

JCD

[Dennis H]

Hey Anthony,

With your dipole, you need to be some distance from the speaker for the backwave to integrate. Nearfield isn't going to tell you much. Since your problem area is with the mid drivers, I'd get SW going and measure at 3' or so. Gate the response out before the first reflection and you should still get a good picture of what's going on above 500 or so.

Edit: if you still have the speaker sitting close to the floor like in your pic, that first reflection will come pretty fast and limit your LF accuracy. Raising it would be good and/or your could throw a fiberglass batt on the floor between the speaker and the mic.

[Anthony]

JCD and Dennis, thanks for the responses.

I now feel the speaker is broken in enough (we've been using it as a center for about 2 weeks now), so I'm going to try to get some measurements this week.

I haven't played around with SW and gating, but that's definitely on the list of things to do.

I'll keep you all posted!

[Doug Plumb]

When I measure the axial response of a speaker I put the mic in random places within a 5 degree cone that emenates from the center point between the midrange and tweeter.

A variety of random positions within this imaginary cone gives me lots of axial measurements to average and therefore I can average out many room reflections when I average these measurements.

I use a gate time that includes the first few reflections because they can average out.

Repeat the experiment twice to make sure you get consistent results.

This usually shows speaker behaviour very clearly for direct radiator point source type speakers. The method doesn't work for horns.

Then the power of an EQ comes in because I can smooth the midrange and get more natural sounding vocals and dialogue.

I think an EQ is best for improving critical band response 100 Hz - 5000 Hz