:: The Problem With Hi Fidelity ::

[gedlee]

Sure I'm very interested (honestly) - please post links/sources etc.

Well, I didn't mean to imply that you had, but it seems to be the focus of much of what you've said here.

Go to http://www.gedlee.com/distortion_perception.htm There is one further paper from 2006 that is available from the AES, but maybe I'll post it.

The problem here can often be that making a particular point can seem like its your ONLY point. The thermal stuff is important, and I am more and more considering it significantly important, but it hasn't actually been a main focus of my work. Its only a factor at "high" SPLs, "high" being very dependent on the design particulars and the design goals. I definately tend towards higher SPL designs than most, and that comes, a lot, from home theater, where effects can often be excedingly loud and to get a "real world" impression they have to be that loud at home.

 

[Magnetar]

Go to http://www.gedlee.com/distortion_perception.htm There is one further paper from 2006 that is available from the AES, but maybe I'll post it.

The problem here can often be that making a particular point can seem like its your ONLY point. The thermal stuff is important, and I am more and more considering it significantly important, but it hasn't actually been a main focus of my work. Its only a factor at "high" SPLs, "high" being very dependent on the design particulars and the design goals. I definately tend towards higher SPL designs than most, and that comes, a lot, from home theater, where effects can often be excedingly loud and to get a "real world" impression they have to be that loud at home.

A 'Snip' of a pair of scissors has been measured to put out over an acoustic watt - try that with a pair of domes and a typical high end 200 watt amplifier-

 

[Patrick Bateman]

I for one have no problem with what's said above; I and others simply use different sets of priorities when approaching speaker design.

Maximum SPL and thermal capability for me do not (necessarily) equal high fidelity - my priority is distortion, regardless of SPL.

So I don't see real conflict here, just the age-old differences in goals and approach. The perennial multiple-ways-to-skin-a-cat thing.

I'd be willing to bet that many listeners would *prefer* a system with high distortion. I'm not arguing whether this is right or wrong. Simply observing that I've met many casual listeners who find speakers high in harmonic distortion pleasing and "exciting."

 

[soongsc]

I'd be willing to bet that many listeners would *prefer* a system with high distortion. I'm not arguing whether this is right or wrong. Simply observing that I've met many casual listeners who find speakers high in harmonic distortion pleasing and "exciting."

I'd bet the same as well.

 

[kstrain]

Yes, not "the" problem

I'd be willing to bet that many listeners would *prefer* a system with high distortion.

I also doubt that harmonic distortion is "the" problem with hi-fidelity.

There is, however, a need to distinguish between distortion that can or cannot normally be heard (by an individual or normal listener, and with music).

Except for low frequency it seems to me to be quite easy to reach the point where distortion is a minor issue compared to several others that have been mentioned in this thread. (I've done informal experiments in adding HD and ABXing.)

At low frequency the poor sensitivity of the ear makes the harmonics relatively easier to hear (at least potentially). However I observe that few speaker designs appear to make low distortion at low frequency a priority over other aspects of design.

Personally I prefer lower than normal distortion in the lowest couple of octaves (which does give a sound that could perhaps be described as "duller"), and that needs relatively large speakers designed with distortion in mind. Otherwise HD does not seem to be a significant design issue when half-decent modern drivers are used sensibly.

Ken

 

[soongsc]

Most people will associate frequency response with tone ballance. While this it is normally good practice to have as flat a response as possible, the released energy stored in the system, especially drivers, have a significant impact. Some people may relate this with distortion during listening tests. However, this normally does not seem to make so much difference in distortion measurements.

 

[danielwritesbac]

Under the testing condition used by Zaph and concentrating on the area of distortion, yes there are better HF drivers and for less money.

Hey, I just like your cat. 😀 It must be difficult to moderate a hifi thread.

I say that because I believe that the number of ears /2 should approximately equal the number of opinions, and most designers of hifi equipment are too proud for adjustment controls, thus the majority of designs may be compromised (wrong tools for the job error) even before the unit is built.

I'm just saying that the stuff is bound to please somebody, but has absolutely no chance of pleasing more than a few people if it isn't adjustable to accomodate rooms and ears.

I do have some examples of equipment from the days, long, long ago, when prosound and hifi were one and the same. In my opinion, there are still important lessons to be learned from comparisons of prosound and hifi.

And, I think that prosound techniques IN high-fidelity is about as concert-realistic as its going to get. 😉

 

[MaVo]

@ gedlee - I read your dipole related posts and have to agree from my subjective point of view. I tried dipole woofers and midrange and switched back to a closed box, as i didnt hear this massive improvements, which alot of poeple speak about. In the opposite, i think they had more drawbacks than advantages.

Its a relief to have this impression comfirmed by a trustworthy source.

 

[Brett]

@ gedlee - I read your dipole related posts and have to agree from my subjective point of view. I tried dipole woofers and midrange and switched back to a closed box, as i didnt hear this massive improvements, which alot of poeple speak about. In the opposite, i think they had more drawbacks than advantages.

You are not the only one.

Its a relief to have this impression comfirmed by a trustworthy source.

Ditto.

 

[kstrain]

OB trap

Its a relief to have this impression comfirmed by a trustworthy source.

Indeed! I fell into the OB trap, as without careful thought it does seem to have attractions. As far as I can tell they are mostly not real, and better solutions exist.

Apart from the directivity and room modes issues, discussed above, the "box colouration" issue is, as far as I can hear and measure, also false.

Structural modes: an open baffle is mechanically weaker than a box, so at least as much work is needed to make it non-resonant (and the surface area is only a factor of ~2 smaller than that of a box).

Internal modes: at first thought these are absent in an OB, but that is often not true due to the cavity behind the drivers (worst when there are "wings" but present to some exent in almost all cases). Then again careful box design, internal damping, and cone choice makes this far from the hardest problem to solve.

OB seems to be a good way to minimise the frequency range that can be covered by a given drive unit.

Ken

 

[gedlee]

There is a lot of data to support these last sets of remarks, particularly the ones on distortion.

High levels of low order nonlinear distortion are virtually always viewed as more pleasing. Its the high order stuff that sounds so bad (crossover distortion etc.). If the idea of high and low order nonlinearity is new to you then this is what you should look into and not THD etc. because understanding nonlinearity order will open up a whole new level of understanding of sound quality.

As to there being an "absolute" in audio design, perhaps, from a purely subjective standpoint there isn't one. But if there is one then it could only be completely nuetral, totally accurate reproduction. Then if you want, you can screw it up any way you like, but at least we all know where we started from.

 

[auplater]

OB

I have achieved remarkable success using BG 75 planars, ATC planars, Panasonic leaf tweeters, combined with low (critically damped) Q ported midbass (peerless 6.5"drivers) and multiple sonotube subwoofers ported to different Fb.. with active room equalization in a quasi line source dipole above ~300 Hz.... for both music and HT.

This setup measures >107dB C weighted @ 15 feet listening distance in a ~15,000 ft^2 two story room, with bass below 15 Hz,. All who hear it state they've never heard better, including audiophiles, musicians, students, wife, naive listeners, etc.

Nasty room modes (especially @ ~32 Hz) are supressed by both active room correction and judicious positioning of the sonotubes (at the nodes on the sidewalls)

I used to build horn designs, but my latest dipoles eclipse any I've heard. Never heard the Sumas, though.

So, dipoles can work fabulously, as long as one respects their limitations. Not sure what all the dipole bashing is about. I agree it's a stupid solution for frequencies from maybve 300 Hz down, though for multiple reasons.

John L.

 

[jzagaja]

Dear Dr. Geddes,

Can you explain small contradiction?

You said:

1) "< 2ms. region we found that the VER would actually sound like nonlinear distortion"

2) "Very early (< 2 ms.) reflection-like signals can be quite audible due to non constant group delay"

3) "We found that in about 30 blind listeners no one could detect nonlinear "distortions" in a compression driver at any level up to the thermal limit of the devices. We tested across three manufacturers. In essence nonlinear distortions were not a factor in audiblity, however there were statistically significant audible frequency response differences"

 

[gedlee]

Dear Dr. Geddes,

Can you explain small contradiction?

You said:

1) "< 2ms. region we found that the VER would actually sound like nonlinear distortion"

2) "Very early (< 2 ms.) reflection-like signals can be quite audible due to non constant group delay"

3) "We found that in about 30 blind listeners no one could detect nonlinear "distortions" in a compression driver at any level up to the thermal limit of the devices. We tested across three manufacturers. In essence nonlinear distortions were not a factor in audiblity, however there were statistically significant audible frequency response differences"

I'm sorry, I don't see any contradiction. Those three statements are all quite true and completely consistant with each other. Whats the problem?

 

[gedlee]

Re: OB

I have achieved remarkable success using BG 75 planars, ATC planars, Panasonic leaf tweeters, combined with low (critically damped) Q ported midbass (peerless 6.5"drivers) and multiple sonotube subwoofers ported to different Fb.. with active room equalization in a quasi line source dipole above ~300 Hz.... for both music and HT.

This setup measures >107dB C weighted @ 15 feet listening distance in a ~15,000 ft^2 two story room, with bass below 15 Hz,. All who hear it state they've never heard better, including audiophiles, musicians, students, wife, naive listeners, etc.

Nasty room modes (especially @ ~32 Hz) are supressed by both active room correction and judicious positioning of the sonotubes (at the nodes on the sidewalls)

I used to build horn designs, but my latest dipoles eclipse any I've heard. Never heard the Sumas, though.

So, dipoles can work fabulously, as long as one respects their limitations. Not sure what all the dipole bashing is about. I agree it's a stupid solution for frequencies from maybve 300 Hz down, though for multiple reasons.

John L.

I am not and have not bashed OB. I just don't see, on technical grounds, what the advantages are. Each time I hear someone describe one - like fewer room modes - it simply isn't true. And everyone who seems to love there OBs have almost without exception built them themselves - one really has to take these kinds of superlatives with a great deal of scepticism.

Put unreliable subjective claims aside and the OB design just doesn't stand up to scrutiny. Show me some real data - like I have posted all through this discussion and is on my web site etc.

You state that "it's a stupid solution for frequencies from maybve 300 Hz down", then what are the advantages from 300 Hz up? I don't see it. To me there is possibly only one advantage and that is a non-baffled driver has a higher directivity than a monopole and in a frequency region where the efficiency loss isn't killing you this could be useful.

Direct radiators just can't compete with a waveguide in that frequency region where the waveguide can be large enough to work effectively. This is about 1 kHz and up. Below this the waveguide needs to be too big to be practical. So from 300 Hz - 1 kHz there is some chance that an OB driver could be made to be an advantage. Although I can't seem to figure out how to effectively incorporate it, without adding unjustified cost and complexity to the system.

There is nothing subjective about the situation, its simple engineering and cost-benefit analysis.

 

[kstrain]

Re: OB

Not sure what all the dipole bashing is about. I agree it's a stupid solution for frequencies from maybve 300 Hz down, though for multiple reasons.
John L.

Hope I was not "bashing": my point was that OB is not an easier way to achieve certain goals. That does not of itself exclude OB designs from being excellent. (Particularly noting your exclusion of bass.)

My earlier mistake had been to regard OB as an easy solution to certain problems - that was not right.

About half of the speakers I have at the moment are dipoles (including electrostats). I won't be adding to that number, as I don't hear or measure any advantages.

Ken

edit - last sentence added

 

[graaf]

I'm sorry, I don't see any contradiction. Those three statements are all quite true and completely consistant with each other.

I agree
nevertheless jzagaja raised an interesting question IMHO

can we thus understand that "VER-related distortions":
1) are due to "non constant group delay" - that is their nature i.e. they are kind of "non constant group delay" distortion caused by very early reflection/diffraction?
2) they "sound like" nonlinear distortion?
3) they are (or can be) very audible unlike nonlinear distortion? (or are we to understand that nonlinear distortion are more audible in cases other than compression drivers?)

Can You explain what do You mean by saying that VER "actually sound like nonlinear distortion"?
Do they sound like even, odd, low order or high order harmonic distortion?
How can it be perceived subjectively? Does the sound become hazy, sweet, soft, bright, grainy, rough etc.?

best,
graaf

 

[auplater]

designs vs. results

I am not and have not bashed OB. I just don't see, on technical grounds, what the advantages are. Each time I hear someone describe one - like fewer room modes - it simply isn't true. And everyone who seems to love there OBs have almost without exception built them themselves - one really has to take these kinds of superlatives with a great deal of scepticism.

Put unreliable subjective claims aside and the OB design just doesn't stand up to scrutiny. Show me some real data - like I have posted all through this discussion and is on my web site etc.

Hmmnn.. last I checked this was a Do-It-Yourself forum...😉 Or am I missing something..

I never said anything about room modes... in fact , I avoid that problem wrt OB's by crossing to non-OB below ~300 - 400 hz.

Seeing as I use a single 6' driver from ~600Hz - ~12Khz, with a 25" planar for mid fill and a supertweeter from ~7Khz up, I avoid most of the lobing problems of many of the vertical array OB's. All this was built for ~ $850 / speaker; is it perfect? No, there are some horizontal interferences between the planars (directed well out of the listening area) and I can't hit 110+ dB continuous summed (though I don't want to!) So, in the spirit of DIY bang for the buck, I've built a SYSTEM for less than $2K over the years that sounds better subjectively than most I've heard up to maybe $15K or so. I've done some near and far field measurements, but, seeing as I have an engineering background myself and hold several patents in the electrochemistry field, my rigorous methodology and measurement equipment is reserved for those pursuits. This is just a hobby for me, so I use my measurements as non published intellectual property (as I'm sure you'll understand).

You state that "it's a stupid solution for frequencies from maybve 300 Hz down", then what are the advantages from 300 Hz up? I don't see it. To me there is possibly only one advantage and that is a non-baffled driver has a higher directivity than a monopole and in a frequency region where the efficiency loss isn't killing you this could be useful.

~300 Hz up avoids the excessive excursion inefficiencies, which is why I chose it. Quite simply, I choose OB for these drivers because the baffle is easier to build (I don't care for boxes if I can avoid them) and frankly, they sound better than commercial and/or other homebrew systems I've listened to. After all, whether correct or not, it's my ears that have to be happy.

Direct radiators just can't compete with a waveguide in that frequency region where the waveguide can be large enough to work effectively. This is about 1 kHz and up. Below this the waveguide needs to be too big to be practical. So from 300 Hz - 1 kHz there is some chance that an OB driver could be made to be an advantage. Although I can't seem to figure out how to effectively incorporate it, without adding unjustified cost and complexity to the system.

There is nothing subjective about the situation, its simple engineering and cost-benefit analysis.

No argument here. You're the expert on CD horn driven box systems it seems. I'm the expert on electrodeposited amorphous transition metal/metalloid alloys and systems 😀 😀
I can show you that work...

http://www.patentstorm.us/patents/5032464-description.html

John L.

 

[gedlee]

I agree
nevertheless jzagaja raised an interesting question IMHO

can we thus understand that "VER-related distortions":
1) are due to "non constant group delay" - that is their nature i.e. they are kind of "non constant group delay" distortion caused by very early reflection/diffraction?
2) they "sound like" nonlinear distortion?
3) they are (or can be) very audible unlike nonlinear distortion? (or are we to understand that nonlinear distortion are more audible in cases other than compression drivers?)

Can You explain what do You mean by saying that VER "actually sound like nonlinear distortion"?
Do they sound like even, odd, low order or high order harmonic distortion?
How can it be perceived subjectively? Does the sound become hazy, sweet, soft, bright, grainy, rough etc.?

best,
graaf

These are superb questions and reflect a real interest in understanding the situation regarding "distortion".

I would not say, or did not intend to say, that the VER are non-constant group delay - they may or may not be "constant", but most importnatly they are a form of group delay error or aberation or "distortion" if you like.

Then let me also say that I have not found the following statement of yours to be true:

"sound like even, odd, low order or high order harmonic distortion"

In other words I have not found a harmonic order to correlate with anything. The nonlinear order does, but there is no characteristic "sound" associated with them that has been characterized. So I'd like to avoid this approach.

The VER have a nonlinear effect in that they are more audible at higher SPLs. Hence at low SPLs they are not audible, but as the SPL goes up they reach the threshold and become audible and increase in audibility at higher levels. Isn't this exactly how we would describe classic nonlinear distortion? But remeber that the VERs are actually quite linear - they don't change with level. BUT OUR PERCEPTION OF THEM DOES!

Nonlinear distortion on the other hand becomes masked at higher and higher SPLs. But many nonlinear artifacts also increase with level so it cannot be said that nonlinear distortion artifacts are or are not perceptable at any level. It depends on how these artifacts change with level when compared to how the masking changes with level.

But, once again, the VER are becoming unmasked with level so there is no doubt that they WILL become audible at some point.

The only data that I know of to conclude if nonlinear distortion rises faster than masking or not is our compression driver study which showed that no one could detect THD level changes from about 5% to about 25%. That certainly implies that in a compression driver system what we hear as "distortion" is not likely to be nonlinearity in the driver, but nonlinearity in our perception of the diffraction. This is a radically new idea.

Hope this is clearer now.

 

[gedlee]

Re: designs vs. results

Seeing as I use a single 6' driver from ~600Hz - ~12Khz, with a 25" planar for mid fill and a supertweeter from ~7Khz up, I avoid most of the lobing problems of many of the vertical array OB's. All this was built for ~ $850 / speaker; is it perfect?

~300 Hz up avoids the excessive excursion inefficiencies, which is why I chose it. Quite simply, I choose OB for thiese drivers because the baffle is easier to build (I don't care for boxes if I can avoid them) and frankly, they sound better than commercial and/or other homebrew systems I've listened to. After all, whether correct or not, it's my ears that have to be happy.

John L.

A 6 foot driver would beam a lot at 12 kHz 🙂

I'm sure that you meant 6", but that would beam quite a bit too, but your description is not that clear so maybe you don't go up this high with that driver.

$850 / speaker for parts is extremely high, almost double what the Summa "parts" would cost. I designed my speakers to be as cost effective as possible and the line I am doing now will degrade performance by a little, but cost the cost almost in half. I have never understood the "cost is no object" design approach and there is certainly no correlation between the cost of audio products and thier sound quality.