I find that the room acoustics and loudspeakers are, by far the weakest links in home sound reproduction. The idea that you can process away nonlinear distortions of these under-appreciated limitations of small, low efficiency loudspeakers that perform poorly is odd to my way of thinking. I'd rather fix the problem at its source instead of trying to get very small speakers that can never do what you want them to do, but end up significantly suboptimizing the loudspeaker design in order to conform to preconceived notions of visual esthetics over acoustic performance. YMMV.
See this excerpt for discussion of the point above.
Chris
Chris, no doubt that loudspeakers and the room they are in are significant limiting factors. It's not entirely satisfactory (imo) to assign all important flaws to them.
Small speakers are small relative to the application.
If you set them up on a desktop and listen at modest volumes, small speakers are not particularly "small". So, there is a balance to be struck between the capabilities of a given speaker and a given room WRT the typical program material.
Point of fact is that most people, even here on DIY do not have systems with exceptionally low THD speakers, exceptionally high SPL capable speakers and amplifiers that absolutely will not clip. AND, that have said equipment that produces (let me say it this way) harmonic spectra that are non-objectionable. I suppose some may, but this is actually a non-trivial thing to accomplish.
Today you can buy something like an iNuke and clobber the headroom, but you'd have to find the iNuke to "sound good" too. I dunno, have heard one only once...
I just do not agree on the premise of the excerpt you cited. In order to do what the authors suggest, significant issues are created, including the need to do substantial equalization, and deal with the time difference/propagation delay between the subs and the mains. Probably fine for HT, not for me though.
Small speakers are small relative to the application.
If you set them up on a desktop and listen at modest volumes, small speakers are not particularly "small". So, there is a balance to be struck between the capabilities of a given speaker and a given room WRT the typical program material.
Point of fact is that most people, even here on DIY do not have systems with exceptionally low THD speakers, exceptionally high SPL capable speakers and amplifiers that absolutely will not clip. AND, that have said equipment that produces (let me say it this way) harmonic spectra that are non-objectionable. I suppose some may, but this is actually a non-trivial thing to accomplish.
Today you can buy something like an iNuke and clobber the headroom, but you'd have to find the iNuke to "sound good" too. I dunno, have heard one only once...
I just do not agree on the premise of the excerpt you cited. In order to do what the authors suggest, significant issues are created, including the need to do substantial equalization, and deal with the time difference/propagation delay between the subs and the mains. Probably fine for HT, not for me though.
The two corrections you mention are in fact trivial to apply: just use an active digital crossover (including such brands as ElectroVoice, Yamaha, Ashly, Rane, Marchand, Bryston, etc.). Here is an article highlighting why you would want to use one. They now cost about the same as well-designed and implemented passive crossovers, and have the distinct advantage being able to be reused over on completely new projects/setups. I'm not sure what point you are trying to make.
Getting rid of nonlinear loudspeaker distortion generated by poorly implemented small loudspeakers (mostly modulation distortion but also power compression) is another trick indeed.
I said nothing about loudspeaker THD. Did I miss your point?
The excerpt from the subwoofer article above was provided in order to encourage reading the JAES article: my excerpt is fair use only to encourage reading the actual article, including the data and information that the author provided.
Corner loading works spectacularly in my setup (see profile photo for details).
Definitely phase is at the top of the list. When I say phase, in multi-driver systems, the importance that at the listening position that all drivers sum and do so in a wide window, so off axis response is incredibly important.
I have said it many times, speakers have many frequency responses (change the mic position and you get any number of them), but it has only one power response.
Now when I say "phase" above, I don't just mean it as timing and summing 'upwards' off all the drivers, but also in the peculiar sense of "noise".
Here a small group of us are still arguing about calling it "noise" as it is not random noise - but it will do for now and has already stuck I suspect.
Most people just don't realise how much "phase noise" speakers produce - that the reason that some people like full-range speakers is the lack of any crossover, and they, crossovers, are a huge source of phase noise. Full-range drivers have an advantage, so do headphones, largely because they are single drivers and also because their impedance is higher.
Electrostatic have a huge advantage, at least on paper. They are 'voltage' devices and since amplifiers are also 'voltage' devices, phase noise is much lower.
Dynamic drivers are 'current' devices where the voltage across the voice coil ought to be a function of current 'through' the coil, but are not. So poor tracking of current creates noise since a 'voltage' source can produce current of any phase angle - and that is just a minefield and more so when you put a crossover in between.
I proved this very clearly recently - I acquired a pair of commercial speakers directly from the manufacturer (with who I have a working relationship) and asked several friends to come for a listen. As I already knew the drivers, I then put a totally different crossover in them - very precisely computer modeled and with care to keep voltage and current to track each other much more closely. The aim was to get close to pseudo current drive to suppress phase noise.
I then invited the same persons back. The reaction, even though externally the speakers looked exactly the same as before, this is not the same speaker.
The increase in clarity was immense, before the speakers sounded small and shut in - they now projected a huge soundfield. I might add that the drivers in the box were of very high quality and of the Danish DST school, designed in Denmark and made in China and the quality is of Scan-Speak level. Before that, the quality was totally wasted and now you could hear just how superb they were.
On the same topic, look at the last few years and the latest Dave Wilson designs and note that he says his crossovers are now 'anti-jitter' - and that is indeed just another reference to phase noise. He has his way of dealing with it, we have found our way, which is quite different (and yes, we know how Dave Wilson does it and that our way is a lot more intricate and elegant).
The loudspeaker that the above experiment was done on, is the Orpheus Apollo Loudspeaker made by Orpheus/Duntech here in Sydney, Australia. I only did this - and with their blessing, as the speaker is now no longer in production.
Re compression, absolutely, that is also part of it. But few understand the nature of compression in loudspeakers, how thermal effects creates seemingly only linear compression since all in-band frequencies are largely affected the same and the time constant is below the audio band, so how can it produce in band non-linear effects? Earl Geddes had a thread on this subject some time back. But some light is starting to emerge on that - but will that leave for another time.
Cheers, Joe
.
I have said it many times, speakers have many frequency responses (change the mic position and you get any number of them), but it has only one power response.
Now when I say "phase" above, I don't just mean it as timing and summing 'upwards' off all the drivers, but also in the peculiar sense of "noise".
Here a small group of us are still arguing about calling it "noise" as it is not random noise - but it will do for now and has already stuck I suspect.
Most people just don't realise how much "phase noise" speakers produce - that the reason that some people like full-range speakers is the lack of any crossover, and they, crossovers, are a huge source of phase noise. Full-range drivers have an advantage, so do headphones, largely because they are single drivers and also because their impedance is higher.
Electrostatic have a huge advantage, at least on paper. They are 'voltage' devices and since amplifiers are also 'voltage' devices, phase noise is much lower.
Dynamic drivers are 'current' devices where the voltage across the voice coil ought to be a function of current 'through' the coil, but are not. So poor tracking of current creates noise since a 'voltage' source can produce current of any phase angle - and that is just a minefield and more so when you put a crossover in between.
I proved this very clearly recently - I acquired a pair of commercial speakers directly from the manufacturer (with who I have a working relationship) and asked several friends to come for a listen. As I already knew the drivers, I then put a totally different crossover in them - very precisely computer modeled and with care to keep voltage and current to track each other much more closely. The aim was to get close to pseudo current drive to suppress phase noise.
I then invited the same persons back. The reaction, even though externally the speakers looked exactly the same as before, this is not the same speaker.
The increase in clarity was immense, before the speakers sounded small and shut in - they now projected a huge soundfield. I might add that the drivers in the box were of very high quality and of the Danish DST school, designed in Denmark and made in China and the quality is of Scan-Speak level. Before that, the quality was totally wasted and now you could hear just how superb they were.
On the same topic, look at the last few years and the latest Dave Wilson designs and note that he says his crossovers are now 'anti-jitter' - and that is indeed just another reference to phase noise. He has his way of dealing with it, we have found our way, which is quite different (and yes, we know how Dave Wilson does it and that our way is a lot more intricate and elegant).
The loudspeaker that the above experiment was done on, is the Orpheus Apollo Loudspeaker made by Orpheus/Duntech here in Sydney, Australia. I only did this - and with their blessing, as the speaker is now no longer in production.
Re compression, absolutely, that is also part of it. But few understand the nature of compression in loudspeakers, how thermal effects creates seemingly only linear compression since all in-band frequencies are largely affected the same and the time constant is below the audio band, so how can it produce in band non-linear effects? Earl Geddes had a thread on this subject some time back. But some light is starting to emerge on that - but will that leave for another time.
Cheers, Joe
.
Last edited:
I've generally found that wide beamwidth correlates to a 'big' sound.
One exception is omnipolar speakers; basically you often get comb filtering, so the beamwidth is only omni in the midrange and bass.
Dipoles are a nice compromise; they have about 180 degrees of coverage, but there are nulls to the left and to the right, which work nicely in real rooms.
TLDR: If you want a big sound, use a wider beamwidth. But be careful with omnipolar speakers, they can suffer from comb filtering at high frequencies.
One exception is omnipolar speakers; basically you often get comb filtering, so the beamwidth is only omni in the midrange and bass.
Dipoles are a nice compromise; they have about 180 degrees of coverage, but there are nulls to the left and to the right, which work nicely in real rooms.
TLDR: If you want a big sound, use a wider beamwidth. But be careful with omnipolar speakers, they can suffer from comb filtering at high frequencies.
Noise is the real parameter to look at, that and of course to some extent the tonal balance, when the speakers are noisy the music can't really escape the box to fill the room. Noise is the compound product of turbulence, breakups and resonance generated from cabinets airflow and pressure based elements. Better to have a noiseless frequency limited system than a big noisy one. Our brain is really good at filling in the missing, if not preoccupied with processing noise and other non content information from the speaker.
Here's a start:
Car Audio | DiyMobileAudio.com | Car Stereo Forum - View Single Post - midbass arrays revisited
(actually that whole thread by DIYMA member Lycan is an interesting read)
I have noticed that the Cornu wall mount sounds much bigger than its little 4 in driver would suggest and this is mostly the result of having a 360 deg array of four back loaded horns splayed out radially to fill the supporting back wall plane with sound. The other thing that makes the Cornu sound big is the extra mid bass near 200 Hz to 300 Hz that emanates from the back horn. This range is low enough to not pinpoint directivity but provides room filling ambience through reflections off the back wall.
With a speaker near the close walls : Is the boomy +3 db to +6 db in bass due to the reflexions + a stacked added signal out of time (phase?) with the one which radiate directly?
Like that, there are often a boomy bass which seems big but in fact uncoralate with the main signal ....which maid it boomy ? The bass harmonics are added togethers and the listener is decieved ? (I mean with passive filter not time aligned for design to be near the walls)
I have got a pass band bass box wich radiate from behind to 120 hz and the bass are tighter and less big sound (in the sense of boomy) at 1.1 meter (sorry for feet?)
Always ask how the famous old english enclosure speakers works... but seems to be very good !
Like that, there are often a boomy bass which seems big but in fact uncoralate with the main signal ....which maid it boomy ? The bass harmonics are added togethers and the listener is decieved ? (I mean with passive filter not time aligned for design to be near the walls)
I have got a pass band bass box wich radiate from behind to 120 hz and the bass are tighter and less big sound (in the sense of boomy) at 1.1 meter (sorry for feet?)
Always ask how the famous old english enclosure speakers works... but seems to be very good !
Much of this is also in my experience too. That's why I wanted to second what this gentleman is saying.
Getting the phase correct in the crossover regions between drivers is a really big deal ( and so is time alignment of drivers/horns): the sorry part of this is that every loudspeaker can be implemented by the builder with these sort of corrections built-in, and especially if the loudspeakers are sold as a commercial product. The manufacturer should take the time and effort to complete the design and implementation.
Of course there is manufacturing drift in the supplied drivers and even in electrical devices, so 100% anechoic testing of these performance factors before delivery is required. I know of at least one manufacturer that does this (Klipsch).
Good writeup on a sore subject-one well within the capabilities of even the most humble of DIYers.
Chris
Last edited:
Big sound is tightly linked to the size of radiating area. For example scale of soundstage of 7 inch FR + radiating surface in midrange equal to such horn mouth = very BIG sound, nothing to compare with direct radiation without horn. Also wide dynamic range is important for the sense of big sound scale
Attachments
Last edited:
I think he just means ripple in the phase response, as opposed to a nice monotonically changing phase wrt frequency. Incorrect use of the term 'phase noise' for lack of a better term for the above phenomenon. If you look at a graph of the phase response, it looks like someone superimosed "noise" on the curve.
"Phase noise" is actually phase (and therefore frequency) jitter of what is supposed to be a pure tone. It's important to us RF guys, and anyone trying to improve their ADCs/DACs.
"Phase noise" is actually phase (and therefore frequency) jitter of what is supposed to be a pure tone. It's important to us RF guys, and anyone trying to improve their ADCs/DACs.
Cask05, WRT to THD (and IM) I was merely saying that this is a prerequisite, and that most systems don't really have low THD and IM. The other issue was headroom.
As far as applying a digital xover/processing is concerned there are some issues with that. It would seem impossible to put a subwoofer in a corner and correct for its frequency response AND correct for the delay and/or phase relationships without also using the same correction for the rest of the speakers as well.
In a non-corner placement, using digital correction for the bass only becomes possible, assuming one can place the sub in appropriate proximity to the main speakers. One can't use a delay to place a sub in front of the mains, only move it behind...
So, this leaves us with a delay for the mains, IF the subs are placed behind the mains. The farther back, the greater the delay, and the greater the distance between the two the more limited the positions where the two are in proper relationship. Back to the "approximation of a point source" issue.
So even if one accepts the idea of placing the subs with the mains in appropriate proximity, the next layer is the idea of having to listen to the mains with the added processing and D/A steps. If the digital processors have managed to become "blameless" or "neutral" at this point in time or not becomes an issue.
Of course one could argue that the compromises of passive xovers in a given system are less than that of the compromises of the digital xovers. A bit depends on one's own perspectives and perceptions of sound and sound quality.
I expect that eventually, maybe even soon, digital correction will reach a level where the compromises are extremely limited compared to the benefits.
The other aspect that is interesting to me is the idea that computer control has made things like high-performance jets that are using otherwise unstable and un-flyable airframes very flyable and very high performing. I don't know if this is something that is even possible with audio, (at least in this sense) but it certainly shows that digital processing and control can have some real world benefits that can't be realized by other means.
As far as applying a digital xover/processing is concerned there are some issues with that. It would seem impossible to put a subwoofer in a corner and correct for its frequency response AND correct for the delay and/or phase relationships without also using the same correction for the rest of the speakers as well.
In a non-corner placement, using digital correction for the bass only becomes possible, assuming one can place the sub in appropriate proximity to the main speakers. One can't use a delay to place a sub in front of the mains, only move it behind...
So, this leaves us with a delay for the mains, IF the subs are placed behind the mains. The farther back, the greater the delay, and the greater the distance between the two the more limited the positions where the two are in proper relationship. Back to the "approximation of a point source" issue.
So even if one accepts the idea of placing the subs with the mains in appropriate proximity, the next layer is the idea of having to listen to the mains with the added processing and D/A steps. If the digital processors have managed to become "blameless" or "neutral" at this point in time or not becomes an issue.
Of course one could argue that the compromises of passive xovers in a given system are less than that of the compromises of the digital xovers. A bit depends on one's own perspectives and perceptions of sound and sound quality.
I expect that eventually, maybe even soon, digital correction will reach a level where the compromises are extremely limited compared to the benefits.
The other aspect that is interesting to me is the idea that computer control has made things like high-performance jets that are using otherwise unstable and un-flyable airframes very flyable and very high performing. I don't know if this is something that is even possible with audio, (at least in this sense) but it certainly shows that digital processing and control can have some real world benefits that can't be realized by other means.
This is also my experience.
Horn-loaded drivers give you that dynamic range without accompanying non-harmonic (FM and AM) distortion. Toole even mentions the "dangers of horn-loaded loudspeakers" on page 382 (first paragraph) of his book due to the absence of modulation distortion and the tendency to turn the loudspeakers up until they are "very much louder than is commonplace with consumer loudspeakers".
I always chuckle when I read that commercial announcement by Toole: at least he's honest.
Chris
Last edited:
I think that day is in the past. I cannot help you with anti-digital decision bias. My ears are the final judge, and I've found that "golden ears" have nothing on mine in this regard. My ears also tell me that the trade is clearly on the side of correcting for time alignment of drivers, separate channels of loudspeakers, and below-Schroeder-frequency room EQ.
I do have pretty extensive experience in this area for both things that fly and preprocessing correction of nonlinear distortion of vibratory sources:
- In the former case it works if well designed and sufficient real-time control laws are well designed/implemented with corrective feedback loops and robust real-time sensor suites (including redundant angle of attack and dynamic pressure sensors).
- In the latter case, I've found that the corrections are not robust with thermal heating of components, i.e., low efficiency direct radiating drivers and passive crossover networks that heat under load. Also there are issues of in-room listener positions that are restricted due to the artifacts of nonlinear response correction, and amplitude nonlinearities (i.e. direct radiators with dynamic limitations due to their inefficiencies, including dipole drivers).
Chris
Last edited:
Ok Chris, this is going a little OT, but the OP seems to have either gone away or been overwhelmed/run over by the herd in full stampede!
So, which digital xovers would you suggest giving serious consideration to?
I do not have an anti-digital bias, rather a realistic appraisal of positives and negatives. Or at least I try to keep my mind open to new developments.
So, which digital xovers would you suggest giving serious consideration to?
I do not have an anti-digital bias, rather a realistic appraisal of positives and negatives. Or at least I try to keep my mind open to new developments.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.