Correlation is a statistical measure.
For THD that includes the series of 2nd HD, 3rd HD, 4th HD, 5th HD, 6th HD and on up the series zero correlation is difficult to achieve.
With zero correlation that would also mean minus 2nd and 3rd HD’s that 4th HD, 5th HD, 6th HD and on up the series have zero correlation to sound quality.
No one except you is making that claim. Sounds like confabulation to me.
Perhaps a better Harmonic Distortion measure would be to subtract 2nd and 3rd Harmonic Distortions and evaluate the remaining non-zero value of the sum of 4th HD, 5th HD, 6th HD and on up the series.
Thanks DT
My preference is a10 inch cone midrange crossed over to a constant directivity waveguide near 2kHz. This higher crossover is above the blatant acoustic impedance disturbances of a horn/waveguide at a lower crossover frequency.
A crossover near 2kHZ matches closely the -6dB coverage angle of both the 10 inch midrange driver and the constant directivity waveguide.
underlined for emphasis
..and the Audioholics synopsis (though in relation to bass freq.s is largely applicable to higher freq.s as well):
https://www.audioholics.com/loudspe...tortion-at-bass/total-harmonic-distortion-thd
"..THD is not a good indicator of how audible harmonic distortion will be unless it is of an extremely high or extremely low value."
DT is qualifying his statement of THD being a good indicator.
Earl if you strip-out that qualifying statement (its context) then it is substantively misrepresented, and you are arguing against something that was never intended. He mentions previously that you "are assigning words that were not said"; well this (above) would also be a prime example.
Com pletely dependent on the room conditions, but yes, the importance does vary with distance as well as the reverberation levels at those locations.
Yes, when they are on your head, but then the definition doesn't make much sense in this context.
"Constancy" of what? I don't follow the question.
Directivity only has meaning beyond the near-field, it's basically a far field measure (as it should be,) which is where we listen. What we hear, in a real room is a combination of direct sound and all the other arriving sounds. the later of which is strongly dependent on the directivity.
Last edited:
Hi Scott
Then I am lost because I thought the claim was that THD was a good measure of perception. If that is not the claim then I agree. Otherwise I am confused.
Constancy of directivity...
Okay thank you for clarifying... so an electrostatic driver that performs better in all respects at near field listening distances and levels than a conventional driver would be a better option in that application?
I share your sentiments with regards to preferring a live room to an overly damp and anechoic one. I have the 15 inch woofers of my speakers in open baffle enclosures which yields a better sound than closed baffles or ported after applying some makeup gain from an active crossover. A major benefit is lack of cabinet resonances which yields a much more effortless and clean sound. I believe it is true that subtractive equalization and filtering is almost never observed as obnoxious where as additive resonances and equalization above the linear median is pretty much always perceived as an annoyance or focal point.
The added sense of room liveliness achieved through an open baffle design it is preference thing derived from an assumption that on average the reverb applied to a recording is insufficient. I find this to be true, especially with contemporary pop recordings which I do not even bother listening to. I prefer a very natural miced recorded quality in a lively room to the addition of digital reverb. I think a well recorded live performance does not benefit from the additional reflections introduced by an open baffle speaker. If the recording is to my liking from the get go, then the need for additional room interactions increasing a sense of spaciousness should not be necessary. Reiterating though, I prefer the sound of a dipole configuration regardless due to reduced enclosure resonances, whether a woofer or compression driver, or electrostat. With regards to dipole designs an electrostat is superior given the equal forward and rear projection of sound inherent by virtue of a flat panel geometry and equal forces applied to positive and negative vectors with basically zero enclosure interaction in either direction.
I enjoy the lively sound of conventional dipoles woofers, compression drivers and waveguides filling up a room with sound at cranking volumes, otherwise electrostats would be my go to.
The added sense of room liveliness achieved through an open baffle design it is preference thing derived from an assumption that on average the reverb applied to a recording is insufficient. I find this to be true, especially with contemporary pop recordings which I do not even bother listening to. I prefer a very natural miced recorded quality in a lively room to the addition of digital reverb. I think a well recorded live performance does not benefit from the additional reflections introduced by an open baffle speaker. If the recording is to my liking from the get go, then the need for additional room interactions increasing a sense of spaciousness should not be necessary. Reiterating though, I prefer the sound of a dipole configuration regardless due to reduced enclosure resonances, whether a woofer or compression driver, or electrostat. With regards to dipole designs an electrostat is superior given the equal forward and rear projection of sound inherent by virtue of a flat panel geometry and equal forces applied to positive and negative vectors with basically zero enclosure interaction in either direction.
I enjoy the lively sound of conventional dipoles woofers, compression drivers and waveguides filling up a room with sound at cranking volumes, otherwise electrostats would be my go to.
THD is a likely indicator when (very) high, not a measure of perception.
Really all he's saying is that THD can be useful as a point of exploration for more likely higher level higher-order products.
In other words THD is still useful (not "so pointless" as you've stated and DT objected to), not in and of itself, but quite likely to indicate other problems (higher level higher order non-linear distortion) that are more often matched with perception.
Of course Fluid makes a good argument that it might not be true (especially with compression drivers and ring radiators not operated near their fundamental resonance). Moreover I'll add that operational bandwidth is also "key": if the bandwidth is very limited THD could be high while higher-order products could be very low. Still, as DT mentions in post 377, these are what most would call exceptions and not the "rule".
What application? Near field? My near field definition may differ from yours. Mine is the objective scientific one. Yours I presume is more loosely based in "being close"? These definitions are often confused stemming from the "nearfield monitor" concept.
In the near field of a source, the sound does not radiate outward, but along a kind of tube. It does not fall as 1/r as sound does in a free field and it varies with frequency. It is a complex concept. In the near field, directivity has no meaning.
Your "electrostatic driver" sounds more hypothetical than a real device.
Last edited:
As far as I understand what you are saying (and it's not at all clear to me) I simply have to disagree. There is no reason to suspect that high THD, 2nd or third means that the higher orders will be significant. They simply have no relationship to each other, they are independent, uncorrelated, ... Personally, I give THD no credence what-so-ever. But believe what you like. I'll stick with the data.
I will agree that there can be cases where THD and higher orders track, it's possible, maybe even common (but maybe not.) The point is that you never know when this is true. But yes, if I saw high THD i'd look to make sure that its low order and not high order, but as I said before, in loudspeakers there is not likely much correlation between low orders and high orders.
This all started with the suggestion to look at the higher order harmonics (hopefully rather than THD or 2nd or 3rd,) which I agree with.
Have a nice holiday.
Last edited:
This statement caught my attention.
First bandwidth has no bearing on order. I can have a low bandwidth speaker with a high order nonlinearity and the harmonics of this nonlinearity will still be present beyond the bandwidth of the source. (The level of these harmonics will differ depending on how the limited bandwidth is achieved.)
Well having looked at the driver performance of so many graphs with 2nd through 5th, accepting of course that THD is a composite of 2nd and 3rd even though it's not expressed on the graph as THD, and looking at 2nd and 3rd in relationship to 4th and 5th, I will strongly disagree.
Ex. there isn't a single compression driver listed here and nor is THD actually plotted, still the concept is (I think) pretty well indicated on this grouping of drivers alone:
https://hificompass.com/en/speakers/measurements
Yeah, happy holiday as well (and hopefully a good holiday meal)! 🙂
Ex. there isn't a single compression driver listed here and nor is THD actually plotted, still the concept is (I think) pretty well indicated on this grouping of drivers alone:
https://hificompass.com/en/speakers/measurements
Yeah, happy holiday as well (and hopefully a good holiday meal)! 🙂
"Order" as-in harmonic? (as in 2nd order, 3rd order; "Higher" being 4th order, 5th order, etc..)
(..as I look back I can see where stating "higher-order products" could be construed for something other than I intended.) 😱
I do not know what the exact range of near field distance is though lets say 1.5m or closer. I own a pair of Genelec 8331 monitors which are true near field monitors and I sit about 2 feet from them. My active two way home brewed system with the Beymas, waveguides and 15 inch woofers is far more impressive and resolving compared with the Genelecs at any distance, as it should be. I have owned several top of the line Stax headphones, Quad 57s, Acoustat Xs with direct drive amplifiers, and have auditioned many electrostatic speakers including Soundlabs, Martin Logan, etc.
I currently have a pair of ER Audio Mini Panels which I have not set up yet and plan to experiment with as an alternative to the Beyma drivers, situated atop open baffle 15 inch woofers. They do not play loud though they are able to be crossed much lower than the Beymas, and are true dipoles by design with very low distortion characteristics. I plan to build full size panels of some sort as an alternative to my current two way system to see if they can exceed the resolution and coherence of my current system.
Last edited:
Here is an example of a live recording that sounds good with a nice sense of space, good levels, and reverberation:
Happy Holidays!
Happy Holidays!
It's not clear to me what you think isn't likely?
Surely not and I'm not trying to score points by finding an obscure exception. In all my perusing of other people's distortion measurements I see very little correlation of how distortion orders appear in overall terms. Driver types and motor designs from certain manufacturers do tend to separate themselves out and follow simple patterns and within them then THD might well be a good quick check metric.
Most compression drivers behave similarly to those in the test I linked, the high 2nd order is as you say always going to be there from the compression action.
Dome tweeters in the more expensive range from Scanspeak and SB Acoustics seem to do something similar, high second order with much lower everything else.
Cone drivers can be all over the map but more generally follow the trend of monotonically reducing level with increasing order. Some buck the trend and have dominant third order distortion.
I don't see enough consistency for there to be a rule or an exception. Taking out the second and third orders and using the rest could be interesting to see what comes out.
THD+N should be followed by a spec of wich order where included. THD+N, just by itself, isn't 2nd+3d only per def., is it? Its just sloppy tech writing really 🙂
//
//
The T stands for total so is meant to be all orders that were measurable. In many cases of speakers drivers the second and third are so dominant as to dwarf all the others even though it could well be the little bit left that is doing the damage.
"Order" and "harmonic" are different, although related.
In the analysis of a nonlinear system, the Volterra approach is to expand the nonlinearity into a Taylor series as a0* x + a1* x^2 + a2*x^3 + .... The "order" is basically the power of each term. Now each order generates a unique set of harmonics IF the input signal is a sine wave (insert sin(x) it the Taylor series and you will see how this happens.) Each order generates a new unique harmonic, but it also generates lower harmonics (but never higher ones). Odd orders generate only odd harmonics and even orders only even harmonics.
For example a 5th order nonlinearity will generate a fifth harmonic, but also a third and a first (i.e. it changes the level of the input frequency.)
So while a limited bandwidth will decrease the levels of any harmonics in the input signal on passage, it will not necessarily decrease the level of a harmonic generated by a nonlinearity (as I said, it depends on the how a bandwidth is limited.)
Again, an example: Consider a bandlimited midrange driver with a LP electrical filter. A nonlinearity order of say 5 will generate a 5th harmonic and this harmonic will still appear in the radiated sound even if it is outside of the drivers limited bandwidth. Say the bandwidth is 200-500 Hz. and the input signal is 300 Hz. The 1500 Hz harmonic will still be present in the radiated sound even though it is outside of the bandwidth of the driver. BUT, if the LP function is acoustic or part of the drivers limitations then this 1500 Hz will be attenuated just as if it were in the original signal.
Nonlinear systems are a very complex animal, orders of magnitude more complex that linear systems theory.