Dynamics in Loudspeakers

Status
Not open for further replies.
Jump to Latest
#122
WithTarragon said:

Gee, We seem to be going backwards here. On the first page of this thread it was suggested 1) to measure the "scaling" ie, "double" the level of the impulse to see if you get "double" the output, 2) to use impulses, 3) to use impulses (or MLS) on top of a carrier (so its response could be easily subtracted). Gee the level of the carrier and the onset time of the pulse (or duration of the carrier) ould be easily manipulated and measured.

A reasonable and interesting question has become side-tracked.
-Tom
Click to expand...

I started this track and the test that you described is not the one that I am talking about. I used white noise at a high level (TBD) and simply tracked the frequency response changes with time as it heated up. This is a trivially easy test.

We also track the frequency response changes as we step up the input by 6 dB to find the -3 dB compression point, but that is another more steady state test, which is not what I am talking about here.
 
#123
Tom Danley said:
I would bet this frame of reference would be useful for a “figure of merit” comparison if one based the comparisons on a specific sound level and frequency band. Tracking dynamic nonlinearity makes sense.
Click to expand...

Tom

IMHO, anything using THD, IMD, second order, third order, etc. is kind of pointless since these things don't correlate to perception. Myself and many others have tested this hypothesis and its true.

But, to my knowledge, there is not a lot of good data (or agreement) on what the thermal modulations effects will do from an audibility standpoint. These are not the same things as nonlinearities, they are more correctly called time-invariances.

What I am trying to do is come up with an objective test to use on a loudspeaker to scale these thermal time variances. Then I will model the effect at different levels and do a valid subjective test to scale the effect perceptually.

I am a little concerned about this test getting way off base from what I intended, and that it be simple enough to do without a great deal of effort.

DDR had some great input, but much of what he proposed would be very time consuming to impliment and I'm not sure that it would be all that much better than what I did or want to do.

What I have taken away from here is that I should not be using white noise and that I should be modulating the signal more in line with real music. I tend to think that a random amplitude modulation - with a bandwidth of say 1-2 seconds, i.e. somewhat slow - of pink noise would be a good input signal. Then I would track the frequency response changes in the loudspeaker over a 30 second test and plot out the worst case deviation. This plot would be the test result. Different forms of single number metrics could be derived from this plot.

Then the subjective test would tell me if this metric correalted with perception and what the "scale" of the regression of the data was. I could then tell you when a loudspeaker had a problem and when it didn't.

For Hi-Fi at low levels, I suspect that this is pointless, but for you and I (pro sound applications), I think that this might be a major factor that is not well quantified.
 
#124
gedlee said:


That is my claim. I use them in my 7m x 4 m room. I would suggest a smaller set of speakers (10's not 15's) for that tiny room of yours, but otherwise I don't see a problem. 3m x 3m is like listening to headphones.
Click to expand...
Now we see some differences being sort out. Different needs relate with how far one listens, and how much the distance range spans. Of course there are other factors, but for live large audience performance, the energy pattern must be spread such that the people in near field don't go deaf while allowing the people in the far back hear clearly. This design is certainly different from normal near field listening experienced at home. The required dynamics are also different since live performance of large scale is in a noisy environment.
 
#125
soongsc said:

Now we see some differences being sort out. Different needs relate with how far one listens, and how much the distance range spans. Of course there are other factors, but for live large audience performance, the energy pattern must be spread such that the people in near field don't go deaf while allowing the people in the far back hear clearly. This design is certainly different from normal near field listening experienced at home. The required dynamics are also different since live performance of large scale is in a noisy environment.
Click to expand...

I think that you missed the point. They are not different.
 
#127
gedlee said:



This is in stark contrast to our study, which found that levels much higher than this, approaching 20%, were inaudible.

I don't have that book, could you ellaborate more on the test.
Click to expand...


While I believe that's the case, it is quite easy, especially with the ubiquitous 2 way, to generate distortion levels in the 20-30% range. While I understand your comments in context, it's also apparent that distortion can become audible in many two way speakers if some thought isn't given in that regard. Below is a graph of a small 2 way, driven at a very modest level, 86dB at 1/2 meter. Already, 3rd order products are in the 10% range. On test signals, this can be very easy to distinguish, though I agree with program material the perception of distortion can be quite masked.

Also, there is the theoretical underpinning that those nonlinearities are not present in the original signal and, as much as possible, should not be there.

An externally hosted image should be here but it was not working when we last tested it.
 
#128
ucla88 said:
While I understand your comments in context, it's also apparent that distortion can become audible in many two way speakers if some thought isn't given in that regard.
Click to expand...

But you have not considered my points in their entirity. I said that there is no reason why distortion in a loudspeaker can't be made to be inaudible. I do it. I never said that all the loudspeakers out there are designed correctly - that couldn't be more untrue.
 
#130
gedlee said:


But you have not considered my points in their entirity. I said that there is no reason why distortion in a loudspeaker can't be made to be inaudible. I do it. I never said that all the loudspeakers out there are designed correctly - that couldn't be more untrue.
Click to expand...

Well, I agree completely with this statement. I really only bring it up as there have been a number of posts-not in this thread, but elsewhere-and posters have essentially used your comments loosely about nonlinear distortion to really justify some fairly poor design choices. So it's, not really directed at you, but at these other posters.
 
#133
"this is exactly why we only make closed boxes. This design limits the cone excursions to relatively small excursions owing to the use of relatively small closed boxes. Trade-off is LF extention."

Sorry for the OT, but isn't excursion significantly lower in a vented system with the same F3 and suitable HP filter?
 
#134
dynamics in loudspeakers - very interesting topic

I have two related questions:

- how much SPL is needed for realistic sound reproduction in home environment? 80-90 dB? 90-100 dB? 100-110dB?
I understand that it is subjective but I am interested in Your personal opinions and particularly in the results of measurements done at home

- how much low bass (two lowest octaves 16-32-64 Hz) is in the music? I understand that it depends on the kind of music but how? And how much is it in those different kinds?

I am a "classical music" lover. Therefore I am particularly interested in the question of realistic SPL and "bass quantity" in "classical music"
 
#136
Tom Danley said:
There are a couple recordings of trains from the back yard at the new web site if interested. Use headphones first.
http://www.danleysoundlabs.com/technical downloads.html
Click to expand...

Tom,

Thanks for these recordings... I have downloaded and auditioned the Harley and Full Coal train tracks - very interesting. BTW, the Fireworks and Train Starting tracks are malfunctioning. The Fireworks link is broken and the Train Starting link downloads a 'tiny' 48k file.

I am pleased to report that these tracks sound increadibly life-like on my system. I especially like the Harley track, since it captures the realism of an un-muffled engine - which I am quite familiar with. It really is quite spooky... My ears are telling me that somebody is starting a motorcycle in my listening room, and I am expecting to smell the exhaust any moment 🙂

The other cool aspect of these recordings is the ambient sounds that are in the background, and how they convince me that I am sitting outside at your place. This has to be due to the simplicity and purity of the recordings - no processing of any kind 😀

Best Regards,
Edward
 
#137
noah katz[/i] [b]Sorry for the OT said:
- how much SPL is needed for realistic sound reproduction in home environment? 80-90 dB? 90-100 dB? 100-110dB?
I understand that it is subjective but I am interested in Your personal opinions and particularly in the results of measurements done at home

- how much low bass (two lowest octaves 16-32-64 Hz) is in the music? I understand that it depends on the kind of music but how? And how much is it in those different kinds?

I am a "classical music" lover. Therefore I am particularly interested in the question of realistic SPL and "bass quantity" in "classical music"
Click to expand...

Graff,
For realistic classical reproduction, I would target 100 dB continuous - full range - at the listening position. BTW, I was just listening to Tom's recording of a Harley motorcycle the other night, and the peak SPL level was 110 dB (C-weighted) at the listening position. This is a high crest-factor circumstance, but it didn't seem too loud - it actually seemed real. For classical music, there is a big advantage to systems which have significant output capability because they can maintain the realism at higher SPL's due to low distortion.

Regarding low bass in classical music:
From an analytical perspective, there is very little musical content below 32 Hz. But the octave from 32 to 64 Hz contains the foundation for many classical compositions. With this in mind, I wouldn't even consider a system serious if it can't extend to an F3 of 35 Hz.

From a perceptual perspective, there is quite a lot of ambient content below 32 Hz. This includes the very-low-frequency envelop content of percussive events, to the low-frequency character of the performance venue. While a system may provide a satisfactory result with an F3 of 35 Hz, greater realism is found with extension to the sub-audible.

As a counter-point, I would not choose a system with a 20 Hz F3 over a system with a 40 Hz F3 if the quality of the bottom octave was poor. For me a sealed system with the higher F3 is preferable to vented system with the lower F3 - all else being equal.

Edward
 
#138
noah katz said:
Sorry for the OT, but isn't excursion significantly lower in a vented system with the same F3 and suitable HP filter?
Click to expand...

This is a complex point. There are frequency regions where the excursion is lowered for a ported box, but there are also regions where it is increased. Below cutoff the excursion of a ported driver goes sky high. If the speaker sees any signals in this region its cone will be flying. The complete unloading of the cone Below F3 is seldom discussed, prsumably because it is assumed that there are no signals in that region. Since our systems have relatively high cutoffs - because we design for subs, a ported design would have significantly higher excursion on the whole, unless a HP filter were used on the main speakers. But in a passive installation this is impractical.

For a given driver, if you look at the total cone excursion for a small closed box versus a proper tuned ported one you will see that it is reduced by a substantial amount over the ported design. But, of course F3 is substantially higher. For a given F3, the closed box would probably be the lower excursion, but its not as significant a difference.
 
#139
graaf said:
dynamics in loudspeakers - very interesting topic

I have two related questions:

- how much SPL is needed for realistic sound reproduction in home environment? 80-90 dB? 90-100 dB? 100-110dB?
I understand that it is subjective but I am interested in Your personal opinions and particularly in the results of measurements done at home
Click to expand...

The situation that I know well and what I have designed my system for is Home Theater. A commercial theater, and I have measured this, will often hit 110-120 dB SPL at the seats. So this was my design target This would not be unrealistic for playbacks of rock music at "realistic" levels either and Orchestras CAN hit these levels in the front of the audience.

- how much low bass (two lowest octaves 16-32-64 Hz) is in the music? I understand that it depends on the kind of music but how? And how much is it in those different kinds?

I am a "classical music" lover. Therefore I am particularly interested in the question of realistic SPL and "bass quantity" in "classical music"
Click to expand...

I would suspect that orchestral music would not have a lot of bass when compared to more popular music of movie tracks. To me its not so much how low or high in level the bass goes, but how natural it sounds. There is no question that a non-smooth ringing low frequency sound field does not sound natural. I use, as I say so very often, three subs on top of my 3 15" main speakers. Thats a lot of low end capability. The drivers are placed all around the room - front-back, high-low, etc. The shear number gives me a lot of LF headroom by default, but more importanly it yields a much smoother bass response. Nothing in my years of sound installations even comes close to the use of multiple subs for the low end. This also seems to be a universal experience - everyone does it.
 
#140
EdwardWest said:
For me an equalized closed box design with a large box and large driver absolutely destroys a vented system for music listening applications.
Click to expand...


Hmmm.... I suppose it comes down to "what we've heard." The very best bass systems I've heard have all been ported. But they are all big, or even gigantic boxes, with low internal pressures. All as clean as a whistle.
But the very worst I've heard have also been ported. 😉 A good sealed box is much easier to do.


Back to dynamic tests: Whatever happened to DDF's idea of a marker embedded in the signal for measuring dynamics. Seems like a cool idea. How could it be extracted and read?
 
Status
Not open for further replies.
Top Bottom