This is a tweeter with "standard" behavior (not T2030). We can see that spectral components are nicely decaying.
I am not saying that non-linear distortion is the most important speaker parameter. It is not. But, on the other hand, it should not be overlooked and called unimportant, or with negligible effect on sound. It is just a part of parameter set we use to evaluate speakers, and IMO one of the important parts.
I am not saying that non-linear distortion is the most important speaker parameter. It is not. But, on the other hand, it should not be overlooked and called unimportant, or with negligible effect on sound. It is just a part of parameter set we use to evaluate speakers, and IMO one of the important parts.
Attachments
Good point.
After all, how can we say that distortion (or any parameter/phenomena) is more or less important than any other when it's not measured with the same scale.. does it really make sense, is it impossible?
Everything must be judged and dealt with on its own... sort of, but also in combination with other phenomena (since they are often related and interacts). It's really complex and I often see simplifications that really does not move anyone or anything forward.
That said, nonlinear distortion IS hardest to get right given a budget and size constraints. That's why in a way nonlinear distortion IS the biggest factor and the biggest enemy to good sound reproduction .
How's that for a contradiction? 🙂
can the min/max phenomenon i've observed be attributable to the non-linearity of air.
as in until sufficient pressure is developed it behaves badly and gets better after that.
as in until sufficient pressure is developed it behaves badly and gets better after that.
so in going back over how this discussion developed can material choices and construction techniques be used to reduce perceptable/measurable distortion in horns?
an example of this from my past is encountering cast iron horns that had been fiberglassed or swaddled with fiberglass insulation and finished with a layer of duct tape all in an effort to minimize the contribution the material made to the sound of the horn (audible distortion in my books i'll leave it to other to decide on classification and relevance) i've also seen expandable foam used to this ends.
the other "mod" to horns/compression driver assemblies i've encountered involved mesh/screen in sometimes multiple layers between the drivers and horns. can't say that it achieved anything in terms of distortion reduction but can attest to what it did in terms of equalizing the frequency response of the horn/driver combination.
perhaps it functioned like the foam in the good DR's waveguides.
an example of this from my past is encountering cast iron horns that had been fiberglassed or swaddled with fiberglass insulation and finished with a layer of duct tape all in an effort to minimize the contribution the material made to the sound of the horn (audible distortion in my books i'll leave it to other to decide on classification and relevance) i've also seen expandable foam used to this ends.
the other "mod" to horns/compression driver assemblies i've encountered involved mesh/screen in sometimes multiple layers between the drivers and horns. can't say that it achieved anything in terms of distortion reduction but can attest to what it did in terms of equalizing the frequency response of the horn/driver combination.
perhaps it functioned like the foam in the good DR's waveguides.
I find speaker HD very easy to hear on sines and sweeps. Much harder to detect in a music or speech signal.
Also, the perception of distortion seems to change with the frequency register. Low, mid, high. Does anyone know of research on that subject?
Also, the perception of distortion seems to change with the frequency register. Low, mid, high. Does anyone know of research on that subject?
Linear distortion is "the" dominate issue. In my world I would put NLD as a 10% problem - worth looking at to make sure you have covered the well know things like flux modulation in the woofer, or oil-canning of the diaphragm, or over excursion in the VC for the desired SPL. These are audible things, yes, but easily resolved. The 90% that's left is linear distortion - highly audible in any and all tests that I have ever seen.
Toole would probably put the NLD at 1% based on how much emphasis he puts on it in his book. So in his world I am an NLD believer!
Your questions are an admirable attempt to understand a complex topic - critical understanding to move forward designing audio systems.
Yea, I would say that those Dutch guys might be wasting their time. We tried all that decades ago at B&C and stopped after we couldn't find any audible significance.
The names are Toole and Olive. Floyds position on NLD is clear from his book -virtually no mention at all, Sean Olive's opinion comes from a conversation that I had with him at a round table discussion on loudspeaker perceptions. His comment was "So you now agree with us that the dominate distortion in a loudspeaker is linear?", to which I responded: "Yes."
Yes, well, of course, from a marketing standpoint you have to say "lower distortion", but that doesn't mean much. It's apple pie and motherhood. How highly would they rate the importance? Alex Vioshvillo - head transducer designer at JBL Pro has talked with me several times on this topic. He did an internal JBL show describing how the perception of distortion does not track the metrics. Basically the same talk that Lidia and I gave.
Laurie Fincham once comment to me that my papers gave him cause for concern about how THX was rating sound systems. He commented that "There is no point in specifying something that doesn't matter just because we can."
Turk,
From all I have read, air non linearity is not much of a problem until around 150 dB SPL. Air non linearity distortion gets progressively worse as SPL increases.
Although air non-linearity distortion is not a problem at the under 1 watt levels usually encountered in a home situation, at the rated power level of most modern compression drivers distortion can easily rise to double digits, and 150 dB at the throat exceeded by more than an order of magnitude.
I tested a B&C DE82 (16 ohm 3" diaphragm) using pink noise at 110 dBA at 2 meters, at the throat the level was 143.7 dBA (144+ on impulse, flat, about 8.2 volts). The conical horn used was 13 x 13 degrees, so is about 6 dB more sensitive on axis than a 90 x 90 horn.
Throat SPL with a more typical horn at 120 dB at two meters would be 159.7 dB, with diaphragm pressures perhaps another 2-3 dB louder, definitely well into the air non linearity region at drive levels far less than the driver is capable of handling within its thermal limits.
Art
Last edited:
Of course there might be "bad" drivers, that's not the point. The point is that you have no "metric" on which to ***** if that driver is "bad" or not since the THD number is meaningless. How would you compare it to another driver? When would you "bet on" the NLD being audible? Those questions cannot be answered at this point.
I use the same software as you "spectra-plus", but near field might be misleading. Current nonlinearity, while it does exist, has been shown not to track the radiated NLD. It's a useful "check" for major problems, but not accurate enough for subjective perception stuff.
Probably most important is the fact that it is very low order.
HOWEVER, air nonlinearity can create distortion of the distortion and create higher orders even though the effect itself is lower order. This can occur only if the wave propagation remains at the very high SPL's, like it would in a plane wave tube, etc. But a rapidly flaring device will dissipate this very high SPL very quickly and so there isn't much of this harmonic growth over distance.
Art, I don't disagree with this is any way. Levels of distortion in a driver at lower levels is not audible - that's what our study found. Things start to sound awful at higher SPLs - we have all heard that, BUT, the question here is what amount of this "awful" sound is linear - with a nonlinear perception - and what is NLD. That has not been resolved - scientifically, that is. In my mind based on my own system I would say that it has been resolved because my waveguides do not sound bad until one is in the painful SPL region. I attribute this to extreme attention to linear distortions like diffraction and internal reflections.
I can't see that happening. I am not sure that I have ever experienced the effect of sound getting worse at lower levels. It changes, sure, but this could be a lot of things, like you nonlinear hearing.
If you mean linear distortion then yes - a lot can be done and there is a world of information here at DIY (See the Geddes on Waveguides thread), or my books and papers. This is what I have spent my life doing and a lot can be done.
In case you are not aware of it I fill my waveguides with foam. This has a major effect on sound quality. Does it reduce NLD - no, that's not possible, but it does reduce diffraction and internal reflections - a lot. My waveguides are dead, dead, dead - custom molded out of glass filled polyurethane and typically two inches think. Thin fiberglass shells are not a good idea.
Earl,
Good points. Unfortunately, most sound reinforcement has tended to use line arrays for the past few decades, and all large format line arrays use throats that are like multiple twisted tiny plane wave tubes for around the first 150mm.
The NLD then becomes even more apparent because of the LD diffraction effects inherent in the compromise of the horn throat designs required for the array to be phase coherent at high frequencies, where air loss over distance requires the most power.
Art
Hi Pano - of course sine tones are the most revealing, but not very realistic.
As far as the perception of NLD goes I know of no one besides Lidia and I who have done these studies. We would have loved to continue to do them, but people did not want to hear what we were saying.
According to hearing theory there would be a marked change in perception with frequency. Our ears have peak resolving power in the 700-7000 Hz range. Distortion in that range is going to be the most acute - linear and nonlinear.
- Status
- Not open for further replies.