What determines how detailed/resolving a midrange driver is?

[cotdt]

Hello all, I'm new to loudspeaker design and I'm on a quest to build a highly detailed and transparent speaker. I've been reading up on psychoacoustics to help me interpret the vast collection of measurement data on loudspeaker drivers. In theory for a stationary listener, measurement of frequency response and nonlinear distortion tell the entire story about a loudspeaker driver, and everything else is just a reflection of these. So I want to find out how to judge the resolution of a driver using these data. Assuming a flat frequency response with 1/6th octave smoothing...

Could linear distortion determine how detailed/transparent a driver sounds? While this may be a factor, so far it doesn't seem like it is the only factor. Reason is because certain polypropellene cone midranges score very well on energy storage tests, yet subjectively they don't sound very resolving. As in, test tones and piano notes seem to lack natural detail. This applies to testing only the raw driver and also in an actual speaker system.

Harmonic Distortion? The general consensus seems to be that under 3% harmonic distortion (down 30.4 dB) is inaudible with test tones, so they will be even less audible with real music. I did blind tests on myself, and while I can clearly distinguish 10% and 5% harmonic distortion for 2nd and 3rd order, I failed the 1% distortion test. Higher order harmonics are harder to notice than 2nd and 3rd order. 1% harmonic distortion can neither be distinguished with my high-end headphones or DIY low-distortion speakers. 1% is 40dB down, the best midranges can do -60dB or even -70dB at decent volumes, I wonder why people place such an important emphasis on this factor, when subjectively linear distortion is more serious concern with drivers. So I don't think ultra-low harmonic distortion is the answer to creating highly detailed speakers. If anything, they may even add some "detail" of their own, but I doubt this actually happens if the distortion is greater than 40 dB down. Also, harmonic distortion for loudspeaker drivers is only high in the low frequency region, and here we are interested in detail...

IM distortion? While maybe more easily audible than harmonic distortion, same things apply as above so I don't think this is it.

I notice that some drivers that perform mediocre in measurements are subjectively hailed by many, and can sound very detailed. Some drivers that measure similarly well can also sound very different, and have a wildly different soundstage (tested by ability to pinpoint the speaker location).

High efficiency drivers can also sound more dynamic, but I've never seen an explanation that is linked to measurement data.

Less dampened cones such as metal often have superior imaging, but poorer soundstage compared to well-dampened cones. This is a generalization but there are different drivers that have the same motor but different cones, and they sound radically different.

So my question is, how can you tell from the measurement data how resolving a loudspeaker driver is?

[cotdt]

Oh and I also don't think it's phase distortion. It's very subtle in very contrived tests, so it would be even more subtle in real music. The room reflections would totally make this a nonfactor I would think.

[cotdt]

I also notice that typically rigid cones sound subjectively more resolving than less rigid cones, even if they measure worse. But the distortion is low enough such that it shouldn't be audible, meaning I don't think it's adding it's own detail.

[soongsc]

Generally, the type if driver that produce better resolution is:

1. one with less dynamic mass.
2. harder material that transmits coil movement to the cone better.
3. Higher frequency capability.
4. smooth impedance curves.
5. short spectral decay time.

[Tenson]

You say phase is not it, but are you talking merely of a single driver? I think phase between, say, a three way system is extremely important to detail.

Low moving mass is also important so the driver can move fast and track the small 'micro' details of the waveform.

Low IMD is a must for me personally, I find myself very sensitive to it. Even to the level that different CD players produce.

Now in terms of room reflections, a directive driver is good as it lessens the effects of room reflections. Hence a di-pole or a driver with high beaming can help. HOWEVER, subjectively I find a wide but smooth off-axis dispersion preferable if you have a treated room with diffusers at the right points. This gives a more 'airy' sound and you still get the extra detail of having no reflections because they (ideally) are delayed t the point your mind can tell they are not the direct sound.

You say assume a flat response, but in the real world many tweeters and mid-range units vary by a few dB. This may sound a small amount but if you have a 1.5dB boost around 1KHz this can help a lot with presence and detail. Your ears are quite insensitive here so it helps. (see equal loudness curves)

Sensitive speakers can sound more dynamic I think, because they are usually very large. Large speakers have faster bass as they can move more air without going so far.

Just some thoughts…

Also just to add, distortion tests may not be noticeable in a a/b test but try listening to all your music with 2% added distortion of whatever kind for a few days, then take it away. Time will tell

[phidauex]

One thing that distortion does, that isn't necessarily conciously audible, is make music sound louder than it is. Very low distortion systems can play louder without sounding louder. However, a cheap boombox sounds too loud very quickly! Its fatiguing to listen to distortion.

So lower distortion can make music more pleasurable to listen to, by allowing you to get the volume to a decent level to hear all the little parts, without making it sound loud.

Just an idea.

peace,
sam

[Tenson]

Less compression helps as well I think.

[cotdt]

soongsc's list exact mirror my experience with drivers! Maybe in the same order of importance too, but I wonder how it's so hard to measure distortion and correlate that with subjective experience. So would smaller midranges sound more detailed than larger ones using the same motor? Based on that list they should have better resolution though they would also have higher harmonic distortion. I do need a lot of output for my open baffle speaker, so I still have my eyes set on a 7" or 8". Based on what I've learned so far, when I select my drivers, I should base it more on physical measurements rather than acoustical measurements of distortion, except for spectral decay which does seem to have a large impact. Currently I have my eyes set on the Vifa XG18's, which I will definately use in one speaker or another sometime down the road.

Maybe I am wrong in my analysis of phase distortion. Yes I am talking about a 3-way. I will try active phase correction and see if it makes a difference. Others have told me it didn't, but I guess I'll try it anyway since Tenson says it's extremely important to detail. Low moving mass and low IMD will also be looked out for. Peak at 1kHz? Done! A lot of mids already have a small peak there.

I also think that wide dispersion is very desirable based on my experience adjusting crossover points, so I'd rather focus on room treatments rather than limiting dispersion for reduction of room reflections. I like that airy sound too, and am very skeptical when others claim it is distortion I am hearing. I'm sure this issue will eventually be resolved.

I'm wondering if I should take a look at the pro drivers? They are usually 8" or 10", so I'm wondering if their cones would be too heavy to have great detail. They do have great sensitivity however, and output which I need for my open baffles. Also less compression as Tenson mentioned. How rigid are they? And how well do they do on decay tests? Which ones are good?

The distortion test I also tried with my normal music, 10% added harmonic distortion sounded horrible. But at 1% I failed the test. I thought I could tell the difference (I tried so hard I might have imagined it) but the blind test showed it was placebo. Generally I think piano music is best for doing nonlinear distortion tests.

phidauex - That's true, but I was mainly comparing drivers that already have low measurable distortion, so none of these drivers have that problem with fatiguing. Although some people do like that fatiguing sound and play metal cone drivers fullrange o_O.

What would some of you guys recommend for highly detailed midranges (150-3000Hz, crossed maybe 300-2500Hz)?
What properties should be looked for in increasing soundstage without significantly removing the detail?
And do you guys think that the unsmoothed frequency response is meaningful? After all, in theory frequency response and nonlinear distortion tell the entire story about a loudspeaker driver, and everything else (ie. physical properties) is just a reflection of these. I'm really bummed that I can't derive much meaningful subjective data from these measurements. So far the physical data seems to be the most useful.

[lndm]

An unsmoothed frequency response may help you notice cone breakup modes. Sometimes it is better to keep the worst of these out of the pass band.

Paper cones can give excellent results, and can be more straightforward when crossing over. Some very good ones out there.

I think that room reflections should be thought of separately to other problems. I think the notion of them masking other problems is greatly exaggerated.

[sqlkev]

Slowly, this is what I've gathered about speakers and resolution.
-Most newer drivers today have surpassed yesteryear's technology (morel, dyn etc..)
-Drivers integration and room acoustics play a huge role in the final outcome of the sound.
-Soundstage, imaging can all be taken care of by dsp
-and I prefer dipole > box sound