That is fantastic info as usual. I had the Mackies you mentioned, I can't remember why I sold them, but it was a live application. Oddly enough the Sp1's are Imaging right on top of them to the point of being sort of creepy, and the RTA shows the same. EG said the room probably had a lot to do with it. Cedar ceiling, carpet, live sheet rock walls, and really good dimensions too. FYI I dot even think I could master on these, but I sure as hell could mix on them and am currently mixing a rock band with these as partial reference. If it were not for the obsession with the DIYaudio university I would get more done and buy more speakers. 😀
Zounds! Do we need a tutorial in how to use the quote function? I've fixed a few here. Please try to get it right. If you don't know how it works, ask. ThanksThis crappy phone the cursor moves and puts random crap in on occasion has almost made me quit posting unless at a real keyboard, sorry and thanks I do *not* want to misquote Earl , especially I he's wrong Bwahahahaha.
Well at least the cop got a proper requote. I am stuck with this phone I have a grandfathered plan Cspire would love to get me off of.
Hello peteleoni
I don't see DSP as the 2 way horn savior. All you need is a decent driver and horn and a competent passive crossover. As I see it nothings really changed. Bad horn is going to sound bad no matter what you do to it.
As far as difraction slots or pinched throats or what have you. Well they don't all sound bad and some measure quite well. I wouldn't be so fast dismisss them all. Kind of like saying all smooth throated horns sound good.
Rob🙂
I don't see DSP as the 2 way horn savior. All you need is a decent driver and horn and a competent passive crossover. As I see it nothings really changed. Bad horn is going to sound bad no matter what you do to it.
As far as difraction slots or pinched throats or what have you. Well they don't all sound bad and some measure quite well. I wouldn't be so fast dismisss them all. Kind of like saying all smooth throated horns sound good.
Rob🙂
I need it explained to me again how diffraction is a linear effect but its perception is a function of level? Also why it can't be effectively treated with DSP or other precise EQ?
Odd is worse than even, as in second is worse than third? Don't tell the single ended triode brigade.
David
And this is exactly what I am needing to put together. Why does this horn sound so good after EQ? Is this old Peavey horn sort of unusually well designed, to me it sounds very much like the horn in the Sentry IIIs after I EQed it ( the Peavey) using but 4 bands of parametric on the DCX2496. After time alignment it was over. I'm switching quick back and forth using Audiogate and 2496 Krall 801s, 804s (damn nice BTW) Peavey Sp1 DSP edition (-: level matched in same room. At this poibnt I'm beginning to think with the tools we have now you'd be a butt not to get this right. Take this with a geaun if salt if you like but this simply was not that hard to do with this speaker, and I get into minutia not on obsessed over curves but in overall flatness and phaseyness caused by peaks etc. This old horn is OK. Another one I'd like to try is the old sp3. I think that one already did sound very nice, it was an 800 Hz CH2 on a semi horn loaded direct radiator, JBL ish.
Last edited:
I guess a badhorn will always sound bad. And here DSP will sort out the good from the bad, but then you can passively EQ it and see if that is better. Remember the more you mess with the sound the more you lose.
DCX etc is no different. I ran a Beghringer Ultramatch into my amp taking the digital from my sound card. Sound OK. took the analogue from the card ....better.
I'm hoping that I don't have to explain why diffraction is linear, that should be obvious, but why its perception is not is a valid question.
In a controlled subjective test that Lidia and I performed and gave an AES paper on (about 2006), we tested the audibility of a diffraction like signal for about 25 test subjects. The test results showed that the audibility was greater for a greater delay, a greater amplitude of the diffracted signal and a greater playback level. The first two parameters being expected, but the third being the most interesting. The audibility increase with playback level was the greatest of the three parameters.
Now why this is true is not clear. Brian Moore makes the claim in his AES paper on the audibility of non-minimum phase that it becomes more audible at higher SPL levels, which is consistent with our results. But he does actually say "why" either. Clearly the diffracted signal becomes unmasked at higher SPLs, the ear being very clearly not linear with SPL. I think that Greisingers model of hearing might be useful in explaining the "why", but clearly it happens.
The effects of diffraction could only be nulled out (complex EQ with the right delay) at a single point in space and would always be made worse at other points.
Last edited:
Perhaps "non-linear" might benefit from a better definition here?
I'm not certain but I think you have some basic misconceptions about distortion.
One can say that *any* distortion is a deviation from the input signal at the output. Or, one could say that any amplitude deviation from the input signal (in proportion, in the case of an amplifier for example) at the output *is* distortion.
3rd harmonic distortion may or may not be particularly audible, a lot depends on what you are listening to. Many instruments naturally are rich in 3rd harmonic, so that makes added 3rd difficult to detect by ear.
Distortion that follows a simple function is not generally (afaik) considered to be "non-linear distortion". Non-linear distortion is just that, not linear or does not follow a simple relationship. As in does not follow distortion vs. frequency or amplitude.
Cone or diaphragm break up modes might be a good example of non-linear distortion.
Bear
Actually, your definition is not correct either. "distortion" can be anything where the output differs from the input, which means that any frequency response change is distortion. But most frequency response effects are independent of the level of the output, they stay the same as the output is raised or lowered. Some effects change character as the signal level changes - these are non-linear, i.e. not linear with the level. These nonlinearities can be level dependent (displacement in a speaker, voltage in an amp) or they can be rate-of-change, (i.e. velocity dependent in a speaker, current in an amp). They could also be acceleration dependent, but this would be rare.
At any rate all of these nonlinear effects can be analyzed if desired, but this can get exceedingly complex. That is because there is no simple form of nonlinearity, it can take an infinite variety of forms and each form has a different type and level of audibility. This is why no one value, say THD, can reflect the true audibility of a nonlinearity.
Cone breakup is dominantly linear unless the material of the cone is stretched to its elastic limit, which would be rare. I have heard people use a "not linear" frequency response to mean non-linear, but that simply not the case. The term has a very precise definition and can't just be redefined at will.
Actually, your definition is not correct either. "distortion" can be anything where the output differs from the input, which means that any frequency response change is distortion. But most frequency response effects are independent of the level of the output, they stay the same as the output is raised or lowered. Some effects change character as the signal level changes - these are non-linear, i.e. not linear with the level. These nonlinearities can be level dependent (displacement in a speaker, voltage in an amp) or they can be rate-of-change, (i.e. velocity dependent in a speaker, current in an amp). They could also be acceleration dependent, but this would be rare.
At any rate all of these nonlinear effects can be analyzed if desired, but this can get exceedingly complex. That is because there is no simple form of nonlinearity, it can take an infinite variety of forms and each form has a different type and level of audibility. This is why no one value, say THD, can reflect the true audibility of a nonlinearity.
Cone breakup is dominantly linear unless the material of the cone is stretched to its elastic limit, which would be rare. I have heard people use a "not linear" frequency response to mean non-linear, but that simply not the case. The term has a very precise definition and can't just be redefined at will.
Last edited:
Earl,
Is there any way I could get a copy of your paper about the audibility of diffraction (varying with intensity)?
Is there any way I could get a copy of your paper about the audibility of diffraction (varying with intensity)?
Pete,
The Mackies you had and sold were probably the SA1521, not the HD1521.
Similar look, size and format, but the HD1521 has more power, better DSP, flatter frequency and phase response (Gunness "magic"), and better sounding limiters.
The older series were decent sounding, and a good value.
The HD series is a fairly big step up as far as subtle differences go.
I encourage any one to audition them, they sound accurate over a very wide dynamic and frequency range, they actually do a real 125 dB at one meter before the limiters kick in.
Art
Dr. Geddes,
Perhaps my choice of words didn't convey what I intended. No problem with what you have said.
I am curious, is there *any* distortion that can not be seen as a change in amplitude WRT the input?
The cone breakup. Not saying that you could not analyze that (I can not), rather that in the case where level increases, and distortion increases (more or less) *linearly* the onset and advent substantial cone breakup *looks* like a non-linearity. Does that work?
Can you give an example of some effect that changes character with level?
I am not clear on what ur indicating on this.
_-_-
Perhaps my choice of words didn't convey what I intended. No problem with what you have said.
I am curious, is there *any* distortion that can not be seen as a change in amplitude WRT the input?
The cone breakup. Not saying that you could not analyze that (I can not), rather that in the case where level increases, and distortion increases (more or less) *linearly* the onset and advent substantial cone breakup *looks* like a non-linearity. Does that work?
Can you give an example of some effect that changes character with level?
I am not clear on what ur indicating on this.
_-_-
Thanks for that.
We have discussed this a couple of times and I have always come away a bit unsatisfied as if I haven't fully understood the concepts. So here are a few more questions sincerely offered.
By linear effect I think we are both agreeing that it is a small-signal effect that doesn't change with input level, at least over a broad "normal" range.
There seems to be a distinction made between diffraction (actually I believe we are speaking of reflections related to diffraction) as opposed to other response issues. Your horns strive to have minimal diffraction reflections and the JBL Biradials are sometimes cited as being offenders in terms of diffraction reflections.
I can agree that the JBL horns can have significant response errors when the diffraction gap is quite narrow as might be designed into a horn giving wide dispersion to a high frequency. The units I have measured have periodic LF ripple that is related to short length and inadequate area at the gap, leading to classic "mouth" reflections (but from the midpoint). In this regard they are no different than any horn, exponential or otherwise, with a low flare rate but inadequate mouth area. The only distinction is that straight horns have one major diffraction point, the mouth, whereas the Biradials have two, the diffraction slot and the mouth.
The Biradials can be designed to have a larger and softer diffraction gap. In this manner a trade off between polar uniformity and LF response is always available, and my sense is that the current JBL engineers have let the pendulum swing back a bit. You can also design a horn with the Keele CD contours and no gap, i.e. with both expansions beginning from the driver exit. In the end this becomes just a conical horn with end flaring.
Now when you speak of horns or waveguides you make a point of singling out diffraction as a unique and undesirable attribute that can be left in or designed out of a horn. You also seem to give it unique perceptual attributes, as in it is subjectively worse than other horn effects. This is where I get lost. It seems to me that diffraction effects are part of every horn at multiple places. Angle transitions occur at the compression driver (and within the compression driver), they occur at the unterminated mouth, gentle diffraction occurs around the curving horn sidewalls (and without it we wouldn't hear a sound out of the horn if we couldn't directly see the compression driver exit). All these physical jogs and kinks cause their reflections in the time domain and contribute to the frequency response of the unit.
If they are linear effects then they are part of the frequency response and they can be equalized out with sophisticated EQ (is that not the case?). If they are reflection related, I would think they are under the same rules of perception that any other reflections would be under. That is, they are audible based on reflection strength, delay, arrival direction, and frequency. I don't think their audibility would be any different than similar strength and delay reflections from, say sharp cabinet corners, adjacent nearby walls, perforated movie screens, etc. (is that not the case?).
I do know that the original Biradials had an audible coarseness to pink noise that I believe was part and parcel to the LF ripple from the inadequate mouth area. I never considered that a unique cause and effect. The response had ripple and you could hear it. I assume if it was carefully EQed away it would no longer be audible (is that not the case?). I would note that the response ripple was uniform with axis and could be EQed out for a wide seating area.
I think your premise is that the diffraction effects (or sometimes you refer to higher order modes) create an audible effect out of proportion to their measurable impact. As such they should be avoided at all cost and other aspects should be compromised, if needed, to get rid of them. That's where I get lost because it seems that, no matter what the cause, the effect is delayed reflections no better or worse than other delayed reflections from other causes.
Can you comment on that, please?
If they are a linear effect, even if they are "punching beyond their weight", they must be measurable. Can you give a measurement example with either time or frequency evidence of the particular diffraction effects? Especially can you show measurements that separate them from the traditionally known horn defects of simple mouth reflections?
You claim they are a linear effect but that their audibility is nonlinear, in that they become rapidly more audible as the test level increases. I assume you are claiming that that isn't the case with conventional reflection audibility? Do you know why that would be the case? Is it unusual that perceptibility of aberrations are related to test level (not level relative to stimulus)?
I know, lots of questions, but I think the group would benefit from some discussion of all the points.
Regards,
David S.
Dave
Yes, too many question, I can't address them all, but there are several misconceptions that seem to be present, that if corrected might make the whole idea somewhat more palatable.
Take your diffraction slot question as a good point to start. When the wave front reaches this slot two things happen, but you only acknowledge one. First there is a reflection of some of the wave (as you state), but there is also transmission and part of this transmission are waves diffracted from this aperture. The diffracted waves do not travel along the axis of the device, but travel at angles to the axis, bouncing off of the walls of the device. These diffracted waves arrive at the listening, any listener anywhere, delayed in time from the direct wave at delays that depend on location. It is this delayed diffraction signal that is audible beyond its measure.
The reflected wave back down the device creates standing waves which themselves are audible, but correctable in that they are one dimensional. But the diffracted wave that is transmitted is not correctable except at a single point, but then only by creating a worse effect - two diffraction signals - at every other point.
The signal diffracted off of the discontinuity - any discontinuity - can only travel down the device to the mouth if they are allowed by the solution - these are called higher order modes (HOM). The diffraction at the discontinuity goes in all direction, but only that direction that aligns with an HOM will propagate to the mouth. All the other directions will decrease exponentially with distance as what are called evanescent wave. These waves CAN reach the mouth if the device is short enough and the frequency low enough, but the exponential decrease in amplitude is pretty steep. At some frequencies the "gain" of an HOM can be greater than the "gain" of the main mode - which you might recognize as a form of 'coincidence'.
Now, what of this has been tested? Not as much as I'd like, but then I am working alone with only my office as a lab and very limited resources.
HOMs have been shown by Makarsky (SP?) at Aachen in his Phd thesis. He concluded that they have little effect on the polar response and I have no issue with this conclusion, but he did not study the perception of them. He only confirmed that they exist.
At B&C we showed that compression driver nonlinearity was not a major issue (although there might be some evidence that at very high SPLs they may be audible, but clearly at normal home room levels they will not be audible).
Yet, horns all seemed to have a distortion that made them sound worse as the level went up. I found this most puzzling at first. Much as many readers are likely to now.
When Lidia and I did our study, we showed that diffraction like effects do increase in audibility as the level went up despite the fact that the effect does not change with level. Moore also noted a similar effect in another context.
Put this all together; we know that greater amounts of diffraction in the device will create greater amounts of HOMs; we know that they do exist in horns; and we know they have an audibility that rises with level. This all says that there is very strong likelihood that diffraction related HOMs may result in an undesirable increase in poor sound at higher SPLs, particularly in a devices which have more internal diffraction.
After this I built several systems to test these ideas - minimizing all forms of diffraction in the system - but there are no scientific tests after that point (except the B&C versus TAD test which would not comment on the audibility of HOMs). So there is only circumstantial evidence at this point. But all that evidence is in agreement with the hypothesis. Nothing that I have seen to date refutes it.
I did a test of the foam that I use - intended to further reduce the HOMs - and it did show a reduction of later aberrations on the impulse response even when the two devices were equalized for the same response. But this could be either an HOM reduction as well as a standing wave reduction. It is impossible to tell without a full mapping of the impulse response across the polar plane. (That and a heck of a lot of programming to extract the relevant data.)
Now, one of your key points is quite valid and that is that perhaps all or many effects in a loudspeaker which we think of as linear have a nonlinear perception. I made this exact same observation in the paper when it was first presented. This could very easily be true and I expect that it is. But then think about the light this sheds on any and all studies, scientific or not for which the playback level was not controlled or investigated as a variable. It can bring their results into question.
Toole and Olive have long contended that nonlinearity is not a relevant factor. Toole barely even mentions it in his book. I agree that it is not a dominate factor, but I also do not believe that it is irrelevant and can be ignored. But I know what it is not and that is THD or IMD or whatever you want to call it as no study has ever shown a correlation between those measures and perception. I at least have some, albeit small, relevant data that supports my position. Only time will tell, but I am not worried.
Yes, too many question, I can't address them all, but there are several misconceptions that seem to be present, that if corrected might make the whole idea somewhat more palatable.
Take your diffraction slot question as a good point to start. When the wave front reaches this slot two things happen, but you only acknowledge one. First there is a reflection of some of the wave (as you state), but there is also transmission and part of this transmission are waves diffracted from this aperture. The diffracted waves do not travel along the axis of the device, but travel at angles to the axis, bouncing off of the walls of the device. These diffracted waves arrive at the listening, any listener anywhere, delayed in time from the direct wave at delays that depend on location. It is this delayed diffraction signal that is audible beyond its measure.
The reflected wave back down the device creates standing waves which themselves are audible, but correctable in that they are one dimensional. But the diffracted wave that is transmitted is not correctable except at a single point, but then only by creating a worse effect - two diffraction signals - at every other point.
The signal diffracted off of the discontinuity - any discontinuity - can only travel down the device to the mouth if they are allowed by the solution - these are called higher order modes (HOM). The diffraction at the discontinuity goes in all direction, but only that direction that aligns with an HOM will propagate to the mouth. All the other directions will decrease exponentially with distance as what are called evanescent wave. These waves CAN reach the mouth if the device is short enough and the frequency low enough, but the exponential decrease in amplitude is pretty steep. At some frequencies the "gain" of an HOM can be greater than the "gain" of the main mode - which you might recognize as a form of 'coincidence'.
Now, what of this has been tested? Not as much as I'd like, but then I am working alone with only my office as a lab and very limited resources.
HOMs have been shown by Makarsky (SP?) at Aachen in his Phd thesis. He concluded that they have little effect on the polar response and I have no issue with this conclusion, but he did not study the perception of them. He only confirmed that they exist.
At B&C we showed that compression driver nonlinearity was not a major issue (although there might be some evidence that at very high SPLs they may be audible, but clearly at normal home room levels they will not be audible).
Yet, horns all seemed to have a distortion that made them sound worse as the level went up. I found this most puzzling at first. Much as many readers are likely to now.
When Lidia and I did our study, we showed that diffraction like effects do increase in audibility as the level went up despite the fact that the effect does not change with level. Moore also noted a similar effect in another context.
Put this all together; we know that greater amounts of diffraction in the device will create greater amounts of HOMs; we know that they do exist in horns; and we know they have an audibility that rises with level. This all says that there is very strong likelihood that diffraction related HOMs may result in an undesirable increase in poor sound at higher SPLs, particularly in a devices which have more internal diffraction.
After this I built several systems to test these ideas - minimizing all forms of diffraction in the system - but there are no scientific tests after that point (except the B&C versus TAD test which would not comment on the audibility of HOMs). So there is only circumstantial evidence at this point. But all that evidence is in agreement with the hypothesis. Nothing that I have seen to date refutes it.
I did a test of the foam that I use - intended to further reduce the HOMs - and it did show a reduction of later aberrations on the impulse response even when the two devices were equalized for the same response. But this could be either an HOM reduction as well as a standing wave reduction. It is impossible to tell without a full mapping of the impulse response across the polar plane. (That and a heck of a lot of programming to extract the relevant data.)
Now, one of your key points is quite valid and that is that perhaps all or many effects in a loudspeaker which we think of as linear have a nonlinear perception. I made this exact same observation in the paper when it was first presented. This could very easily be true and I expect that it is. But then think about the light this sheds on any and all studies, scientific or not for which the playback level was not controlled or investigated as a variable. It can bring their results into question.
Toole and Olive have long contended that nonlinearity is not a relevant factor. Toole barely even mentions it in his book. I agree that it is not a dominate factor, but I also do not believe that it is irrelevant and can be ignored. But I know what it is not and that is THD or IMD or whatever you want to call it as no study has ever shown a correlation between those measures and perception. I at least have some, albeit small, relevant data that supports my position. Only time will tell, but I am not worried.
I'll see if I can find it. That was ten years ago, however. It never made it into the journal and we could not even give it in person owing to a death in the family, so its dissemination has been very sparse.
- Status
- Not open for further replies.