Agreed. But we have people that keep asking for explanations. Trying to keep it intuitive as much as possible.
The general theory of nonlinear systems can handle differences in frequency response as well as differences in level, but it cannot handle time variant systems. Thermal problems can fall into this category, but otherwise most situations can be handled.
Agreed that these details muddy up the landscape. The fundamentals are the variations of the signal with level and the fact that a single nonlinear system will react very differently to different signals (IMD, HD, ...), which is why we always used music in our tests. Sine waves seem to me to be useless for auditioning nonlinear distortion because they are not even close to real music.
The real problems come when the nonlinearity is large enough that we have to consider the distortion of the distortion. Then hold on to your hats as things get multi-dimensional!
Agreed that these details muddy up the landscape. The fundamentals are the variations of the signal with level and the fact that a single nonlinear system will react very differently to different signals (IMD, HD, ...), which is why we always used music in our tests. Sine waves seem to me to be useless for auditioning nonlinear distortion because they are not even close to real music.
The real problems come when the nonlinearity is large enough that we have to consider the distortion of the distortion. Then hold on to your hats as things get multi-dimensional!
Last edited:
That's in line with what I've measured in my various listening rooms and the rooms of others. Low 80s average SPL is where I find a lot of loud-ish listening levels. Bigger, better treated rooms go louder.
I've also measured concerts, they are LOUD. Average C weighted levels are in the mid to upper 90s, often going above 100 dB. But a lot of that is massive bass because live sound is often far more bass heavy than home playback. A-weighting drops the levels quite a bit. The classic P.A. rider specs a system that can deliver 112dB clean at the mix position. That may have changed recently. I work mostly corporate gigs, where we ask that it be pulled back a bit - it's still loud.
Different results from different signals? Harmonic distortions occur at different frequencies than intermodulation, so they coexist, this makes them distinct, I would think.
According to that reasoning, it sounds like harmonic distortion of 900Hz would be a different form of distortion from harmonic distortion of 1100Hz, because the "distortions occur at different frequencies."
So, I'm not clear on why distortions occurring at different frequencies would be thought to be of particular significance.
Last edited:
Maybe a way to explain harmonic distortion in relation to intermodulation distortion is by looking at the diagram attached below of a very basic type of RF mixer circuit. Suppose we put one signal in at the input labeled RF (with some DC bias on it so the diode is always forward biased maybe at 0.5 VDC, and we keep the AC part of RF signal not too big like maybe 0.1V peak to peak. With just that one signal present as the input, the output would show a lot of harmonic distortion.
If we then put in another AC signal at the input labeled LO, and we don't let it get too big, maybe 0.1V peak to peak, what we would see at the output would be a lot of IMD.
Nothing has changed with the diode, the thing responsible for the distortion, in other words, the nonlinearity is exactly the same.
Maybe the above description can help illustrate how THD and IMD can be two different ways of looking at the same nonlinearity, since the diode has stayed exactly the same. I think that's what Earl was trying to get at in what he was saying about THD and IMD.
If we then put in another AC signal at the input labeled LO, and we don't let it get too big, maybe 0.1V peak to peak, what we would see at the output would be a lot of IMD.
Nothing has changed with the diode, the thing responsible for the distortion, in other words, the nonlinearity is exactly the same.
Maybe the above description can help illustrate how THD and IMD can be two different ways of looking at the same nonlinearity, since the diode has stayed exactly the same. I think that's what Earl was trying to get at in what he was saying about THD and IMD.
Attachments
Last edited:
True, disco pub were the days when I had no means of measurement. Today, one can use a good phone to get a good idea even though they may not be so accurate.
I can't say I have the background to understand the circuit. I'll take another look at it though.
With respect to IMD vs harmonic. Harmonic distortion(s) occur at multiples of a frequency, 2 or 3 or 4 times etc. IMD is the result of an interaction between two different frequencies and doesn't occur at multiples (though I suppose it could by chance).
It seems almost common sense that IMD would be a worse offender since its less in harmony with the sound.
I believe the typical "small speaker struggling in large room" garbled sound is more the result of IMD than harmonic distortion. There are also other forms as well.
I've haven't been impressed with much of the research done into the perceptibility of distortion. There's too many variables, the largest of which is people. Conflicting results abound.
Seems like nowadays "Science says" has become a kind of Jedi mind trick to get people to stop thinking altogether.
Maybe a better way to look at it is that IMD is the result of both frequencies interacting with the nonlinearity at the same time, rather than them interacting between themselves.
In the case of the diode circuit, distortion is caused because of the diode's resistance change according to the voltage across it. If only one frequency is present then only that frequency is causing the resistance change. If two frequencies are present, then each frequency sees resistance changes from two different causes at the same time. That's what makes the IMD.
Of course, in reality everything is changing at all at once together, not as independent sub-processes. But, I'm trying to see if I can help get you thinking in a different way, hopefully developing more of an intuitive sense of how THD and IMD can be the same thing in a certain basic sense, the same nonlinearity effect, just different outputs only due to different inputs, not with any difference in the underlying mechanism that is causing distortion.
Last edited:
It all depends on the crest factor (CF) of the music being listened to.
With most commercial recordings, CF < 10dB, and the quoted ~94dB peak are probably adequate.
With audiophile jazz and classical recordings, CF may be as high as 20 - 25 dB, and then, for the same average SPL, much more headroom is required, like 105 - 110 dB peak. And, contrary to popular belief, provided the average level is kept constant, these latter recordings do NOT sound any louder - just more "alive".
And alas, this is where most hi-fi systems fall flat on their face.
Marco
Yes, and I like recordings of that kind - even though they sometimes "eat power for breakfast" when you turn things up a little.
With some jazz-rock and funkjazz albums I can drive the 2x700 Watts amp (that is drivin the double 15" woofer sections of one of my systems) just below clipping during peaks without sounding painfully loud. When I do the same with some rock albums it almost parts my hair.
Regards
Charles
I write for those who want to learn, not those who believe they know it all already. If my writing is over someones head then that should tell them something. If I had to lower my discussions to the lowest common denominator, then I wouldn't post here very often either. If a few get the point, then I am willing to leave the other 98% behind.
Surprisingly, it has been proven that sound travels farther upwind than it does downwind. So much for intuition.
Dr. Geddes, thank you for your benevolent condescension, but may I please point out three errors you made while flying over our heads:
1) The theory that narrow dispersion is required to make better speakers because they minimize room effects is faulty. From my own experiences it appears to be rather the opposite.
2) Distortion in speakers is audible. There is no simple one-on-one relationship between THD and audibility, it obviously is more complicated than that. But, as I mentioned, listening tests have on the whole indicated that panels prefer lower over higher distortion, ceteris paribus. Thus, driving down driver distortion remains a worthwhile goal.
3) Although they represent the same reality, FR and CSD provide different perspectives. Not every wiggle in a FR is caused by resonant effects.
Dr. Geddes point out that THD in loudspeakers is not an issue, at least in any good/competent loudspeaker. While I agree that one-number THD percentage is not a good representation of loudspeaker sound quality (separate second, third, fourth and fifth harmonics percentages are way better indicators), "good/competent loudspeaker" is a rather vague description, as this example will show:
There is one certain big German loudspeaker manufacturer, which models consistently gets excellent reviews in German hi-fi magazines. German magazine Stereoplay routinely measures frequency response and THD at various levels from 80 to 95 dB/1m and many models from that manufacturer exhibits high THD in the vicinity of 2 kHz. Nothing to worry, isn't it?
One day, one audiophile contact me to measure his brand new loudspeaker (the usual suspect) because "it distorts badly" (his words) from the first day. I measured it and, to no surprise, there was about 2% third harmonic at 1W level centered at 2 kHz - the same as Stereoplay measurements. Yes, audibly it distorts badly, in spite of "THD inaudibility of competent loudspeakers". Is it to blame the definition of "competent loudspeaker", or "THD inaudibility"? Choose one. Or two.
From my experience, even 0.6 % of third harmonic is audible - venerable Eminence PSD 2001 is an example of that.
There is one certain big German loudspeaker manufacturer, which models consistently gets excellent reviews in German hi-fi magazines. German magazine Stereoplay routinely measures frequency response and THD at various levels from 80 to 95 dB/1m and many models from that manufacturer exhibits high THD in the vicinity of 2 kHz. Nothing to worry, isn't it?
One day, one audiophile contact me to measure his brand new loudspeaker (the usual suspect) because "it distorts badly" (his words) from the first day. I measured it and, to no surprise, there was about 2% third harmonic at 1W level centered at 2 kHz - the same as Stereoplay measurements. Yes, audibly it distorts badly, in spite of "THD inaudibility of competent loudspeakers". Is it to blame the definition of "competent loudspeaker", or "THD inaudibility"? Choose one. Or two.
From my experience, even 0.6 % of third harmonic is audible - venerable Eminence PSD 2001 is an example of that.
Last edited:
i'd like to thank DPH for his post it has helped.(i do have to ask though, you referenced a link that seems to be missing?)
if i could be allowed to share more of my thinking i feel that it would go a long way to resolving my problem, would that be ok?
i don't want to create the impression i'm trolling. i can see how some of my questions and comments would appear that way ,not what i'm trying to do at all i'm just trying to get more info.
if i could be allowed to share more of my thinking i feel that it would go a long way to resolving my problem, would that be ok?
i don't want to create the impression i'm trolling. i can see how some of my questions and comments would appear that way ,not what i'm trying to do at all i'm just trying to get more info.
A sudden change in distortion centered at one frequency would certainly lead me to at least consider a pathologically flawed driver, i.e. not a "competent loudspeaker".
It sounds like what you're saying is that IMD is the distortion resulting from multiple frequencies playing at the same time, whereas harmonic dis is the result of a single frequency being played. So they don't coexist. Is this correct?
That's pretty tight. It may be accurate for highly compressed recent releases, but I've only seen that on metal bands. The loudness wars are squeezing all dynamics, that's for sure.
A typical mastering level in the early days of CD was an RMS value 18dB below full scale. A lot of rock, pop and jazz was done at the level. Peaks are 18dB above average which is not too hard to do. On less dynamic music, like simple singer recordings, 16dB below full scale is common - sometimes 14dB. Classical is more dynamic, often 22-25dB RMS below peak. You need way more headroom for those recordings.
Regarding IMD and harmonic, this article suggests they are different, but at least in amplifiers they can be indicative of eachother, IOW if one is bad so is the other. Not sure if this applies to speakers.
THD and IMD Distortion—Sidebar | Audioholics
THD and IMD Distortion—Sidebar | Audioholics
Unfortunately they do ! :-( If we take an example with two frequencies as input signal, both will create their own harmonic distortion spectrum - accompanied by the intermodulation spectrum caused by both frequencies intermodulatimg with each other due to the same nonlinearity that causes the harmonic distortion.
Regards
Charles