How much tweeter distortion is audible?

[Pano]

Modo - try cutting it to 10 seconds and putting it in a zip file. That should work. Should bring it down to about a meg.

And BTW, what kind of harmonic is 1.3K? I'd call that "inharmonic". At least for a 1K fundamental.

Here is a 1KHz + 1.3KHz file in FLAC and MP3 formats. 1K is -3dBFS and 1.3Khz is mixed in at -33dBFS (30db below fundamental)

Let me know if this works.

 

[mondogenerator]

These are the valid file extensions:

bmp doc gif jpe jpeg jpg pdf png psd txt zip

Says so when you hit the "Manage Attachments" button

yup lol i see it now, pity i have winrar and not winzip really.

@ panomaniac: im aware that 1k and 1.3k is not harmonically correct, it was just an illustration as i SHOULD have done 1k and 2k or 3k depending if i wanted to show a single harmonic 2nd or 3rd blah blah. i just entered 1 and 1.3k as i know it to have a harmonious nature(i believe its a third, or perhaps a fifth, in musical instrument terms) and i also know 2nd to be similar as it is octave based. also being a harmonious musical tone it should have less issues with IMD confusing things, or at least that was the thought behind it! hehe

after doing this i can see how an octave chorded overtone, whether 2nd OR 3rd (one octave or two?) would be less distinguishable than my example, although from experience with guitar etc im not really sure there would be any real difference between 2nd and 3rd generated harmonics at all. maybe if a speaker produced lots of 4th or 5th then it would probably cut your head off!

saying that, a -30dB attenuation STILL being audible in either of these cases, causes my to doubt the theory/tests still. -30dB is way to high a threshold for me at least, and I would hope that Toole and Ged are right, (even though that would mean i have to agree with them), since if they are not then a -30dB is NOT good enough, and the old theories are right, meaning people would still cliong to the supposed 0.00001% THD ideal. That would be very sad indeed, since even i know many amps with THD approaching 1% sound miles better than some MOSFET ones with ultra low THD, mainly due to over use of NFB in the latter.

An interesting test would be to use gated pink noise with added single harmonic at a similar attenuation. I wonder if THAT would be audible at -30dB? if so that would pretty much prove the transient thresholds of audibility....

 

[Patrick Bateman]

Let me interpret since I did the work. The "weak correlation" of THD and IMD to perception was NEGATIVE - you forgot to mention that. This means that, as a metric of perception, one should increase the THD to make it sound better. This is of course absurd, but it's the absurdity of THD and IMD measures that is the culprit not the results of the test.

I swapped out your loudspeakers (my Gedlee Summas) for a couple days, to evaluate some JBLs that I bought at Costco which seemed to have well-behaved directivity. *Immediately* I noticed that the sound was a lot worse, particularly in the treble. This got me to wondering why the JBLs measure relatively well* but sound so inferior.

First, I'm apt to agree that harmonic distortion isn't offensive per se, but I think you can be trained to hear it. For instance, I have some subs which measure very clean, and I notice that they sound different than other subs which do not measure as clean. Admittedly, the "clean" subs do sound a bit less "exciting" and more clinical. I could see how many would prefer the sound of distortion, and may even perceive a low-distortion speaker as being inferior to one with high distortion.

The THD and IMD numbers are indeed shown in all the graphs. They are the X-axis.

Its NOT the level of the harmonics that matter but where the nonlinearity occurs - at low levels or high levels and the order, 2nd, sixth, etc. Low level nonlinearity is by far the most insidious especially if high order - like crossover distortion in an amp. This is why an amp with extremely low levels of THD can still sound terrible. But loudspeakers, on the other hand, tend to have nonlinearities that increase with level and are likely very low level like second or third. This makes them fairly benign. In fact, for the most part, nonlinearity in a loudspeaker (as long as its not broke) is a non-issue. In a test of compression drivers we had twenty five people evaluate distortion levels up to 25% and statistically noone could detect it at those levels.

Its hard to make blanket statements about the audibility of nonlinearities in specific cases, but for the most part nonlinearity is a major concern in electronics, but so much so in loudspeakers.
There are, of course, going to be exceptions to this.

You've said that "low level nonlinearity is the most insidious."

Would the diffraction off the cabinet edge of a typical bookshelf speaker create the kind of low level nonlinearity which you refer to? Since the diffraction will occur at all volume levels? (IE, harmonic distortion tends to rise with level, but diffraction exists at *all* volume levels.)

I'm curious about this, because I've noticed that some speakers have audible harmonic distortion which I can pick out with ease, but they don't sound *bad* to me.

An externally hosted image should be here but it was not working when we last tested it.

The best example of this are the speakers from Anthony Gallo, which are spherical. I can hear major 2nd harmonic, but they sound very pleasant. Perhaps the Gallo stands have something to do with it too, since they're abous at low diffraction as you're going to get. OTOH, most bookshelf speakers make me want to run out of a room, no matter how clean they sound.

And I've often argued that people spend too much time focusing on the profile of your waveguides, and not enough time studying the cabinet. (It's my belief that a great deal of the speakers imaging and laidback sound is due to the cabinet.)

* I posted some measurements of the JBL at my forum. The response is tipped up in the treble, but I fixed that before I compared it to the Summas.

 

[Patrick Bateman]

Josh

I understand. Lidia said this exact same thing when she went to RMAF. She said after looking around that people can't buy good loudspeakers so they buy what they can and hope that it helps. They don't ask for vailidity of the purchase because none is required - it's the "hope and faith" that counts.

I think there's a novelty factor too. New gear always sounds better than old gear, not necessarily because it's better, but because it sounds *different.* I think this partly explains the popularity of tube amps, because most solid state amps sound quite similar. The sound of tube amps is all over the map. (IE, if you buy three solid state amps in three years it won't be satisfying because they'll all sound the same. Spend that money on three tube amps and there will be features of each which will be memorable. And many audiophiles will then prefer the tube amps, because they'll feel as if they got something for their investment.)

 

[gedlee]

That "different" in audio is almost always assume to be "better" is one of its biggest problems. All "perfect" amps would sound the same, right? There wouldn't be anything left to talk about. The hobby would become quite boring. All that would be left is to actually just listen to music.

 

[gedlee]

Would the diffraction off the cabinet edge of a typical bookshelf speaker create the kind of low level nonlinearity which you refer to? Since the diffraction will occur at all volume levels? (IE, harmonic distortion tends to rise with level, but diffraction exists at *all* volume levels.)

I'm curious about this, because I've noticed that some speakers have audible harmonic distortion which I can pick out with ease, but they don't sound *bad* to me.

Lidia and I did a study of this which was presented at AES. Turns out that diffraction is mostly a temporal effect and behaves quite differently from spectral effects like nonlinearities. Nonlinearities - spectral effects - become masked at higher SPLs, but temporal effects, quite surprisingly, become unmasked at higher SPLs. This was the major conclusion of our study. This means that low level nonlinearities are the biggest problem (as typically found in electronics) but at high SPLs it is diffraction that limits the ability of a system to play loud and still "sound right". Can your Summas play loud without sounding bad? That's not nonlinear distortion, that's low diffraction.

 

[33Polkhigh]

Distortion levels in drivers.

This an old thread but I've come across it a number of times and want to weigh in, if anyone is still listening.

Let's just agree that distortion is a bad thing and what separates high fidelity from low fidelity. The problem is that distortion itself is a complex phenomenon and measuring isolating and defining it isn't that easy. Many studying it, such as Klippel have made similar statements. Distortion itself may in some ways be related to our perception of sound itself and so very difficult to study.

Harmonic distortion is only one metric of distortion. Just like you wouldn't rely only on ldl cholesterol to assess health, so thd numbers have to be understood in context and taken for what they are. When people post thd numbers as the ranking of driver quality its fair to say that it can be misleading to a lot of people. Harmonic distortion isn't the boogeyman some think it is.

That said, blanket statements about nonlinear distortion, like "you can't hear it" at some ridiculous level like 10% distortion ought to be met with skepticism, even if some so called experiment produced that result. You can definitely hear nonlinear distortion is a typical driver.

Take two different drivers, one metal and one paper, and equalize them flat on axis. I can guarantee that most people here who know how to listen will hear the subtle "woof" of the paper or the crispness of the metal. Much more so at high volumes.

That "flavor" of the driver will at least in part show up in non linear distortion tests. It also happens to be fairly benign to sound quality, but not something to ignore either. Spending an extra 30$ on a driver to have less total distortion seems like a good idea to most builders.

I appreciate Dr. Geddes contribution to the field, but if people really can't hear 20% distortion in a horn (i forget the exact number in his experiment)
then why did he choose a 100$ B&C driver for his summa speaker? Why not the cheapo 20$ horn? I'll tell you why, because the cheap PA horn sounds like junk and we all know it, even if some experiment might contradict. Maybe equalized flat it would sound better than we might think, but still not as good.

Companies like Harmon want to take this experimental "objectivist" approach to building speakers and I appreciate their contribution, but I think its not the right approach. Certainly flat response and smooth off axis roll off are very important, even in my own experience, but theres other stuff going on. If they want to put cheap drivers in junk cabinets that meet only those criteria I doubt they'll be in business long.

You can do experiments all you want and show that the mcdonalds hamburger is preferred to Kobe beef, or that all violins sound the same, even broken ones, or all mattresses are the same. Honestly most people can't tell a good speaker to a bad one, at least not in the short run.

I don't even bother asking "non educated" opinions of speakers anymore because all I get is a dumb look and "sounds fine to me". They don't know what to listen for or a reference for what its supposed to sound like. A lot of people think boomy awful distorted base sounds good. That doesn't mean that there isn't such a thing as good base.

So in summary, speaker designing is still somewhat an art. Linkwitz calls a speaker and it's environment a "gestalt". There is such a thing as high fidelity just like good acting, but its not something that is easily quantified or deduced by experiment. As others have said, many small incremental changes and decisions can have big results on the final result. If a tweeter sounds tizzy or cheap, regardless of numbers, use a better one.