90 degree phase shift
Hello,
Relative 90 degrees phase shift between two channels is easy to do in analog domain using couple of all-pass filters set up properly. It can achieve band width of more than a decade with phase error of only couple of degrees. Analog phase shifters like this used to be popular in the 70's with Ambisonics systems.
gedlee said:
Hello,
Relative 90 degrees phase shift between two channels is easy to do in analog domain using couple of all-pass filters set up properly. It can achieve band width of more than a decade with phase error of only couple of degrees. Analog phase shifters like this used to be popular in the 70's with Ambisonics systems.
john k... said:
Yes.
It gets tiresome.
..Apparent Format:
1. An objection is raised.
2. See my work on (..which often requires an expenditure).
3. That response isn't worth anything.. *OR*
4. (..if the response does seem to meet an academic standard and at least initially deemed worthy), change the subject (..often with another objection).
"rinse and repeat."
(..sometimes adding in the occasional defensive posturing.)
I found it was annoying enough to warrant the "inclusion" of the forum's IGNORE LIST. Ultimately the format above is essentially doing just that - ignoring the respondent (..though in a "strung-out" manner).
BTW,
(..trying to get the topic on-track
)What about the "velocity" operation that some cardioids and dipoles provide?
What about cross-correlation ("reduction") with wavelengths that ordinarily sum? and in keeping with this..
What about side-wall ("reduction") summation.
IMO the best "contrast" to all 3 questions is with an in-wall true infinite baffle. It tends to highlight whats right and whats wrong between your standard monopole and most implementations of cardioids and dipoles.
But why 90 degrees - which is quadrature ?
Even when the drivers/cabinets are specified we still have significant transduction phase variation through the resonant frequencies as well; whereupon any circuitry must be driver specific.
This is becoming more complex than audio is meant to be.
Yes phase/frequency tracking filters are possible, or drivers can be used in resistively small enclosures below their imposed system resonance only, but a filter which develops a quadrature output (90 degrees represents amplitude differential in time) takes time (variable with frequency) to modify output. The waveforms/pressures are modified in 'music time' after the music event has driven the filters, and this cannot fail but to distort any eventually reproduced leading edge impulse events, as well as muddy a composite in-room response.
The same would apply when a separately boxed monopole is used to generate cardioid when the monopole waveform is filtered to phase match the output of a dipole. Only an optimum range of frequencies for the cardioid pattern can be assured, which must then become 'cottage loaf' like either side of that optimum frequency.
I cannot help become distracted by un-neccessary argument, and I think it was Earl who broached the subject that cardioid can only be LS-position-room specific.
Cardioid is no more than a balancing of figure of eight plus 2pi output to null the rear output. The rear output of a dipole is going to be completely different in an open field compared to when in a room with corner/wall reflections modifying level/phase of rear output at the very point where it these must combine and null to generate the cardioid response.
However, when the cardioid is passively generated without electrical signal mixing between a dipole and a monopole then the rear null can be physically adjusted to suit any room corner, or even a field, and I think JohnK uses his circuit for the LF amplifier crossover and corner response filtering only.
Also I believe that JohnK's 'mechanical' cardioid (no electrical filter induced leading edge distortions) is optimised for lower frequencies only, such that the response of his design gradually reverts back into full dipole operation with increasing frequency.
Maybe JohnK's cardioid dipole plus one of Earl's separately positioned bass monopoles would be an excellent working compromise ?
Cheers .......... Graham.
Even when the drivers/cabinets are specified we still have significant transduction phase variation through the resonant frequencies as well; whereupon any circuitry must be driver specific.
This is becoming more complex than audio is meant to be.
Yes phase/frequency tracking filters are possible, or drivers can be used in resistively small enclosures below their imposed system resonance only, but a filter which develops a quadrature output (90 degrees represents amplitude differential in time) takes time (variable with frequency) to modify output. The waveforms/pressures are modified in 'music time' after the music event has driven the filters, and this cannot fail but to distort any eventually reproduced leading edge impulse events, as well as muddy a composite in-room response.
The same would apply when a separately boxed monopole is used to generate cardioid when the monopole waveform is filtered to phase match the output of a dipole. Only an optimum range of frequencies for the cardioid pattern can be assured, which must then become 'cottage loaf' like either side of that optimum frequency.
I cannot help become distracted by un-neccessary argument, and I think it was Earl who broached the subject that cardioid can only be LS-position-room specific.
Cardioid is no more than a balancing of figure of eight plus 2pi output to null the rear output. The rear output of a dipole is going to be completely different in an open field compared to when in a room with corner/wall reflections modifying level/phase of rear output at the very point where it these must combine and null to generate the cardioid response.
However, when the cardioid is passively generated without electrical signal mixing between a dipole and a monopole then the rear null can be physically adjusted to suit any room corner, or even a field, and I think JohnK uses his circuit for the LF amplifier crossover and corner response filtering only.
Also I believe that JohnK's 'mechanical' cardioid (no electrical filter induced leading edge distortions) is optimised for lower frequencies only, such that the response of his design gradually reverts back into full dipole operation with increasing frequency.
Maybe JohnK's cardioid dipole plus one of Earl's separately positioned bass monopoles would be an excellent working compromise ?
Cheers .......... Graham.
Well, Grahm and Earl have touched on my next question or suggestion. Earl's set up seems to consider placement, LF cut, amplitude and phase, anything else?
So, why not add source directionality to the list. Say we position 3 cardioids or dipoles so that they are at the intersections of floor/side wall, floor back (or front) wall, and side/back wall. All sources would be away form the corners and would have their axis aligned with the line formed by the intersection of the surfaces. Using dipoles, for example, since they couple strongly only to modes which propagate in the direction of the dipole axis such a set up would allow individual control of front to back, side to side and vertical modes. Cardioids would couple differently.
So, why not add source directionality to the list. Say we position 3 cardioids or dipoles so that they are at the intersections of floor/side wall, floor back (or front) wall, and side/back wall. All sources would be away form the corners and would have their axis aligned with the line formed by the intersection of the surfaces. Using dipoles, for example, since they couple strongly only to modes which propagate in the direction of the dipole axis such a set up would allow individual control of front to back, side to side and vertical modes. Cardioids would couple differently.
@ John k, is there a difference in exciting all modes just as gedlee describes and your idea to controll all modes? do you mean no mode is excited with this setup? And if thats the case, what will be the difference in perceived sound? The double bass array i posted earlier also seems to want to controll the modes instead of exciting them. Maybe my question boils down to something like "Is it possible to excite no mode in a room?".
@ Gedlee, i would really like to see a closer description of your woofer setup process. intuitively i would put the woofers in the room like you described. then at first let the one in the corner play a pink noise sample at full volume and look at the rta. while the first one plays, i would raise the second subs volume until its about the same at listening position and play with the phase, until the rta shows the smoothest response and finally repeate that for the third sub. Another question would be, do you mean a simple allpass phase controll like most cheap subwoofer amps have or something more sophisticated?
@ Gedlee, i would really like to see a closer description of your woofer setup process. intuitively i would put the woofers in the room like you described. then at first let the one in the corner play a pink noise sample at full volume and look at the rta. while the first one plays, i would raise the second subs volume until its about the same at listening position and play with the phase, until the rta shows the smoothest response and finally repeate that for the third sub. Another question would be, do you mean a simple allpass phase controll like most cheap subwoofer amps have or something more sophisticated?
MaVo said:
John - I have said many times, but I guess that I'll say it again. Talking about directionality of the source in a frequency range of the room where the sound CANNOT travel in arbitrary directions is pointless. You can talk about LF sources in a free field or in a room, but the two things are completely different and there is no reason to believe that what happens in one will happen in the other. Thus, I think that your "test" is based on the false assumption that the free-field directivity will still hold in the closed room. It won't.
MaVo your procedure is pretty close. I am reluctant to to give a "simple" overview of the technique, but here it is (there will be some details missing and the devil is in the details). The one source (two or three actually) that has to be fixed in the room are the mains. They have to be positioned for best MF -> HF response at the seats. Thus their response, phase, and amplitudes are a given. The rest of the subs need to augment those.
Playing a noise signal I measure the sound field by spatially averaging (simply move the mic by hand over the seating area while the Analyzer averages. This give a good approximation of the power response at LF, but will not be accurate at HF (unless your analyzer can do a dB average as opposed to a spectral average). I add one sub at a time, usually starting with the lowest frequency one. I just use simple plate amps with a phase switch, gain control and LF cutoff control - nothing fancy. Try adjusting these settings until you reach the best response with the first sub and the mains all playing at the same time and the response spatially averaged.
Then add one sub at a time. Adjust the gain and the LF cutoff for best overall response and try the phase switch to see the diffrence. The phase switch should be a large difference for the first sub (pick the "better" one), a lessor difference for the second and so on. Add one sub at a time.
While this procedure cannot be proven to lead to the optimal response, it always leads to a very good response. Remember that the subs and the mains all overlap each other to some extent.
On occasion we have seen modes in the room that we just cannot get the response down at without deflating the neighboring response. In these cases some EQ helps. I have never used more than two bands of EQ and even then only less than 1/2 of the time. If you are using more EQ than this then something is wrong.
It is also possible to start all over if the end response is not as good as you think that it should be. If you end up with the same response two or three times then you have it as good as it gets and EQ is going to be required, or maybe another sub.
MaVo said:
Its obvious that the above procedure could be automated and improved by just taking transfer functions from each source to the listening positions. Then the computer picks the best settings. This is what I had hoped to do some day. A computer with sound card and a mike would be all that was required. Everyone has these tools these days.
Hi John,
I can see you are chewing over the kernel of an idea there, but I am puzzling to understand what it might be.
We should not alter the music in an attempt to correct a room response which arises after the loudspeakers have reproduced sound, especially as room modes are in three dimensions plus volumetric, whereas stereo has only two channels.
Cheers ....... Graham.
I can see you are chewing over the kernel of an idea there, but I am puzzling to understand what it might be.
We should not alter the music in an attempt to correct a room response which arises after the loudspeakers have reproduced sound, especially as room modes are in three dimensions plus volumetric, whereas stereo has only two channels.
Cheers ....... Graham.
FrankWW said:
It wasn't intended to be a personal attack, rather stating a degree of displeasure with a type of forum "response".
As far as getting to some interesting things.. well, yes - but that was a couple pages ago. If you really enjoy things being "sorted out" I'd recommend the Enable "Technical" discussion thread.
If you find me tiresome, by all means make use of the forum's IGNORE LIST.
recently I had to pull my system apart and hence put it all back together. Anyway, turns out somewhere that I had one of my bass drivers out of phase, what I mean is that when I fired it up the bass was obviously 'wrong' and a simple flick of the switch and it was fine and back in phase.
But the point I'd like to make backs up the observation made a page or two ago, that the diffuse bass soundstage given when out of phase WAS extremely enveloping and enjoyable in it's own right.
I'm sorely tempted to set up an extra sub or two, deliberately out of phase (and may even try the crossed 'x' mentioned earlier) and see if by judicious juggling of levels etc try and maintain the in-phase accuracy and imaging, yet superimpose upon it sufficient out of phase content to get the best of both worlds. It may even help by the distributed nature of the subs to iron out some of the FR anomolies.
I did try the Harman Kardon paper suggestion of distributed subs before, but found that it didn't really work for my ears. However, by lucky experimentation much later I think I found out why.
My theory to explain it (and it hasn't been touched upon here yet, so I'd love to see what others think) is that the 'trouble' arose because the subs were actually at different distances from the LP. So I figured that the dislike I experienced could have been down to the different arrival times from the different subs. You know, instead of one 'thump' it was kinda 'thump thump', or worse, At the very least, I spose it was 'smeared'.
On the lucky experimentation I mentioned before, I used a DCX 2496 to run the seperate sub channels, and could dial in reasonably closely I think the appropriate delay for each sub according to it's distance from the LP, with the aim that all the signals would arrive to coincide with the signal from the mains.
It certainly made a difference to my ears.
Anyway, assuming some quasi random placement of the subs, are there any comments on the audible effects of the different arrival times, or did I just manage to convince myself of the importance of something that is essentially unimportant?
I too would love to see a write up of the procedure that would help me do it properly, I already have had an interest in it.
And, I too would think it wonderful if you made available the HT writeup you speak of Earl, there would surely be a lot of valuable data in it.
But the point I'd like to make backs up the observation made a page or two ago, that the diffuse bass soundstage given when out of phase WAS extremely enveloping and enjoyable in it's own right.
I'm sorely tempted to set up an extra sub or two, deliberately out of phase (and may even try the crossed 'x' mentioned earlier) and see if by judicious juggling of levels etc try and maintain the in-phase accuracy and imaging, yet superimpose upon it sufficient out of phase content to get the best of both worlds. It may even help by the distributed nature of the subs to iron out some of the FR anomolies.
I did try the Harman Kardon paper suggestion of distributed subs before, but found that it didn't really work for my ears. However, by lucky experimentation much later I think I found out why.
My theory to explain it (and it hasn't been touched upon here yet, so I'd love to see what others think) is that the 'trouble' arose because the subs were actually at different distances from the LP. So I figured that the dislike I experienced could have been down to the different arrival times from the different subs. You know, instead of one 'thump' it was kinda 'thump thump', or worse, At the very least, I spose it was 'smeared'.
On the lucky experimentation I mentioned before, I used a DCX 2496 to run the seperate sub channels, and could dial in reasonably closely I think the appropriate delay for each sub according to it's distance from the LP, with the aim that all the signals would arrive to coincide with the signal from the mains.
It certainly made a difference to my ears.
Anyway, assuming some quasi random placement of the subs, are there any comments on the audible effects of the different arrival times, or did I just manage to convince myself of the importance of something that is essentially unimportant?
I too would love to see a write up of the procedure that would help me do it properly, I already have had an interest in it.
And, I too would think it wonderful if you made available the HT writeup you speak of Earl, there would surely be a lot of valuable data in it.
Interesting.
Back in the mid 70's I made a pre-amp with extended LF response which did not roll off at 50Hz as per RIAA. Obviously the bass was something else, especially as I was using Celestion Ditton 66 loudspeakers at that time.
This pre-amp offered option for switching the primary of a miniature push pull driver transformer between channels at interstage points where the falling inductance could cross-couple the LF only to make the bass increasingly mono with falling frequency.
Thus when music was playing (loudly) and the bass notes were becoming extendedly boomy due to LF SPL feedback (even with a genuine brick wall mounted turntable) then the switch could be flicked and reproduction (disco partying) could continue.
The enemy had been the differential LF pressure waves energising room modes, NOT the overall LF response.
Thus I can now appreciate the validity of this out of phase woofer because it must reduce boomy LF room excitation, though, it might induce differential excitation at some frequency and thus positioning need to be careful arranged.
Also, if the LS 'suck' was not equidistant from the listener to the driven sources, then surely there might be additional 'thump-smear' as experienced by Terry.
Just a few thoughts.
___________________________________________________
JohnK has optimised single driver cardioid bass, and there has been comment that it cannot be any better than monopole bass in relation to room modes.
My reason for not wanting to use the single driver bass is because I prefer the cleanliness of OB driver mounting even at LF; ie. no delayed damping.
John's last investigations have suggested that the radiating energy for either dipole, cardioid or monopole at LF does not match driver energies at higher frequencies, and (if I read correctly) he has already suggested a need for balancing this with a monopole.
So maybe extra monopole placements will be validated no matter what type of main drivers are being used, though I have noted that the distance between any Sub driver and listener can be critical unless frequency dependent phasing circuitry is introduced, even below 50Hz, and thus I prefer fully integrated main drivers only.
Cheers .......... Graham.
Back in the mid 70's I made a pre-amp with extended LF response which did not roll off at 50Hz as per RIAA. Obviously the bass was something else, especially as I was using Celestion Ditton 66 loudspeakers at that time.
This pre-amp offered option for switching the primary of a miniature push pull driver transformer between channels at interstage points where the falling inductance could cross-couple the LF only to make the bass increasingly mono with falling frequency.
Thus when music was playing (loudly) and the bass notes were becoming extendedly boomy due to LF SPL feedback (even with a genuine brick wall mounted turntable) then the switch could be flicked and reproduction (disco partying) could continue.
The enemy had been the differential LF pressure waves energising room modes, NOT the overall LF response.
Thus I can now appreciate the validity of this out of phase woofer because it must reduce boomy LF room excitation, though, it might induce differential excitation at some frequency and thus positioning need to be careful arranged.
Also, if the LS 'suck' was not equidistant from the listener to the driven sources, then surely there might be additional 'thump-smear' as experienced by Terry.
Just a few thoughts.
___________________________________________________
JohnK has optimised single driver cardioid bass, and there has been comment that it cannot be any better than monopole bass in relation to room modes.
My reason for not wanting to use the single driver bass is because I prefer the cleanliness of OB driver mounting even at LF; ie. no delayed damping.
John's last investigations have suggested that the radiating energy for either dipole, cardioid or monopole at LF does not match driver energies at higher frequencies, and (if I read correctly) he has already suggested a need for balancing this with a monopole.
So maybe extra monopole placements will be validated no matter what type of main drivers are being used, though I have noted that the distance between any Sub driver and listener can be critical unless frequency dependent phasing circuitry is introduced, even below 50Hz, and thus I prefer fully integrated main drivers only.
Cheers .......... Graham.
terry j said:
I'm inclined to think that time delays are not a big issue as audible events in and of themselves. The reason should be kind of clear. AT LFs these delays are only a small portion of the signal period. It just takes too long for our ears to detect a LF signal. We can't detect the presesnce of a LF signal in a few ms let alone have these kinds of delays be audible. Detection of arrival times is very complex, but no researchers that I know of have found any effect at LFs.
With several of subs arriving at different times the net result is a single tone which is the complex sum of the individual signals. Hence even small changes to the delays can have big changes to the net summation of the LF sources. So you are changing the FR at LFs when you change the delays or the phase to the subs and this can sometimes be quite a lot. As to its being an improvement or not only some measurements would sort that out.
I am going to post my HT book, a chapter at a time. I will review each chapter for updated data and then post it. It will take a while like this, but it will happen. Stay tuned.
gedlee said:
In response to these comments I spend some time yesterday and this morning preparing some calculation of the behavior of a dipole and a monopole source is a rectangular room when the dipole axis is aligned with one of the room directions. The results clealy show the effects of directionality of the source on excitation of the room modes in the direction aligned with the dipole and in the directions perpendicular to the dipole axis.
Enjoy
john k... said:
John
I appreciate all your work, but I don't see where this really shows anything. Its ideal to make your point and I can show an example, not so ideal, which makes mine.
Do you deny that a sound wave in a room with widely dispersed modes (little modal interaction) can ONLY travel along the wave vector defined by that mode? This is the key because you either agree or disagree. If you disagree that this is true, then I can easily prove that it is and if you agree that its true then you cannot contend that directivity is a factor. Whats your position?
The two directions that you have choosen to look at in your example are aligned with the mode directions so of course this works in this case.
I am not promising anything as I have a lot to do, but maybe I will do a full frequency response simulation of examples of dipole cardiod and monopole. I will do several situations as one is never conclusive. This will show that, in general, all that one achieves by the dipole component of any source is a lower SPL, not a smoother SPL.
Do you agree that in a simulation like this a cardiod is modeled as two monpoles seperated in space, out of phase, with one twice the strength of the other? It also possible to do three points, but this is just more work and I wouldn't expect a much different answer.
gedlee said:
My position is that you stated, more than once, quite emphatically, that directivity doesn't matter. The fact is that it does matter. This point has been raised in this discussion, by several people, and is related to Kate's and Salmi’s dipole work, etc. I say this, not to make any claim of which is better or worse, but only to show that there is a difference. Take a dipole, a cardioid and a monopole woofer; place them in a room on a Lazy Susan. How the different sources couple to the room, what resonances are excited, and by how much, and how they will sound will vary as the dipole and cardioid are rotated. This will not be the case for the monopole.
I do not agree that a cardioid is two monopoles separated in space, out of phase, with one having twice the strength. That will not produce a cardioid response.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.