My program does as accurately as 1/48 octave smoothing. I've seen some commercial plots that use 1/3 octave smoothing which when I use it on my cheap tiny little computer speakers look positively spectacular. I would prefer at least 1/12 octave.

What do people use to represent?

Marlboro

Those are questions with a lot of right answers and a lot of wrong answers. There is no one right answer for each question, but there are a lot more wrong answers.

For me, there are only a few absolutes:

   1. If you don't have the right equipment or you don't have experience with test equipment or both, you're better off NOT measuring. The results you get may be wrong and may lead you in the wrong direction.
   2. Unless, of course, you're practicing. In that case, it's worth doing but don't necessarily trust your results.
   3. You can measure a speaker, or you can measure a room. That's two entirely different things.
   4. When measuring a speaker, bass should never be measured indoors.
   5. When measuring response in a room, bass must be measured at several locations within the room.
   6. Bass measurements should not be smoothed at all.
   7. Above 300Hz or so, some smoothing can be used to remove the "grass". But smoothing beyond 1/6 octave removes so much detail pretty much everything looks smooth.
   8. When comparing two speakers, the same level of smoothing should be used. Best to compare using the same type of test equipment and test setup.
   9. Outdoor ground plane measurements are easiest to get accurate measurements from. Bass response can be measured with the speaker sitting on the ground in a wide open space outdoors. Higher frequencies (wide band) should be done with all drivers within 1/4λ of the ground. That may require you to dig a pit and/or make a false baffle and point the speakers upward with the measurement mic suspended above it.
   10. Indoor pseudo-anechoic measurements can be made using some equipment that has gating, but it is only accurate above a few hundred Hertz, depending on the position of the speaker under test, the microphone and boundaries (floor, walls and ceiling).
   11. Real-world in-room measurements should be made with the speakers where they will be used. Several measurements should be made throughout the room. This will show room modes and other effects of boundaries like floor bounce notches, etc.
   12. In-room measurements should never be used to compare speakers, unless they are gated to remove reflections. If the room is measured, the speaker is not. Room effects swamp almost all other acoustic details that could have been seen specific to the loudspeakers at least at low frequencies in the modal range (below about 300Hz).

And the one most important - Listen to the experts. Measurements are not a trivial excercise. You can't just get an RTA and a microphone and expect anything reliable to come from it. I've seen a lot of people do that, and they'd be better off using a modeling program. It would give them more accurate results, particularly below 300Hz. There are a lot of measurement methods, and a lot of ways to get it wrong. For that reason, I'd ask this same question over on the Measurement forum. Maybe Keith Larson will chime in. He's done a lot of work in the field, and I think his input would be very helpful.

While there are a lot of wrong ways to do it, but there are a lot of right ways too. There are several good methods, a lot more than what I described above. Maybe some of those methods can be examined in detail in a series of threads.

So many people over on the PE forum are most likely doing bad measurement.

And then we have the situation with the commercial speaker manufacturers who are using 1/3 octave smoothing for their speaker systems, and the curve looks so good, for their $5000 speaker system. Using 1/3 octave with a bad mic makes my 1 inch computer speakers look great!

Thanks so much.

Marlboro

Those who put passive crossovers into their systems may be able to measure the speaker system, but when you are talking about a system that has two 12 inch woofers, 34 three inch midranges, and 60 3/4 dome tweeters, I can't see anything other than measuring the whole room.

Marlboro

If comb filter distortion starts at 9,000 you'll hear it as loss of sparkle. Problem is that the line array's sound stage and lack of distortion may mask the lack of highs unless you can hear another line array with much higher start up of comb filter distortion as a comparison.

Before I hooked up the tweeters, I ran the system with the 3.5 inch middies running full range. It sounded great, but the comb filter distortion from a 5 inch center to center caused a huge loss of upper frequencies. I didn't realize how much better is could sound until I connected the tweeters, and comb filter distortion didn't start until pretty near the limits of my 58+ year hearing(15,800 hz).

Marlboro

There are a couple of averaging methods to consider. One of the most common is smoothing the dB response using a line smoothing algorithm. This simply removes the squiggles, or as it was aptly put 'cuts the grass'. On the other hand if you take many responses true amplitude noise can be reduced producing the raw frequency response (with grass), that you can then 'line smooth'.

Hope this helps,
Keith Larson

That sounds like an equalizer in front of a standard passive crossover. An active crossover has to do the splitting, and by its nature requires multi amping since the crossover network goes Pre-amp -> active crossover -> 2 or 3 different amps -> 2 or 3 different speakers.

Ed