To paraphrase, you have dips and peaks in the frequency response of your speakers with the natural associated minimum phase response, and rather than fix the frequency response (and therefore the phase response) you elected to just fix the phase response leaving unnatural lumps and bumps in the f response...........why would you do that ?
No, of course not. The speakers I used measure flat well within +/- 1.5 dB, so this had been fixed already.
Dips and peaks in the FR should be corrected with minimum phase methods, which I did with some simple analog corrections. You could also use a MiniDSP to do that, which is the advice I gave to Kindhornman. But this still leaves you with the phase shift associated with enclosure and xover.
I wanted to hear if a good pair of speakers would get better by flattening the phase. The one man jury is still out on this. In spite of the prelimenary results I reported that appeared to indicate readily audible differences, I must have been driven by expectation bias, because in the end I could trace these differences back to other factors.
A fun thing to do with RePhase is to introduce inordinate amounts of phase shift and listen to the results. The only conclusion I can draw at the moment is that the sort of phase shifts LR4 xovers or bass reflex enclosures introduce are not something to worry about.
Dips and peaks in the FR should be corrected with minimum phase methods, which I did with some simple analog corrections. You could also use a MiniDSP to do that, which is the advice I gave to Kindhornman. But this still leaves you with the phase shift associated with enclosure and xover.
I wanted to hear if a good pair of speakers would get better by flattening the phase. The one man jury is still out on this. In spite of the prelimenary results I reported that appeared to indicate readily audible differences, I must have been driven by expectation bias, because in the end I could trace these differences back to other factors.
A fun thing to do with RePhase is to introduce inordinate amounts of phase shift and listen to the results. The only conclusion I can draw at the moment is that the sort of phase shifts LR4 xovers or bass reflex enclosures introduce are not something to worry about.
A few pages back we had some squeaks about a new thread banning mentions of DBT, yet the biggest scam in Audio the BQP still rears its ugly head and now there are even crystal technology product available......
Then the phase response will also be flat, by definition, if the system is already minimum phase.......they are inevitably linked.
Improving time alignment of speakers does not change a lot the "sound" envelop of the instruments. (On my point of view)
The main difference is with details or fast transients. We will notice details like nails on acoustic guitars that you never noticed before, more realistic and fast drums attacks etc...
About the general discussion in those last pages, it would be a mess if "science" became a religion.
The scientific approach is necessary, and near all what science tells-us is true. But not always sufficient ... do not explain everything.
So, i believe we have to take great care of what say our senses, ensuring they do not fool us ;-).
About feedback, make yourself your opinion with this preamp/headphone amp http://www.diyaudio.com/forums/solid-state/235695-no-nfb-line-amp-gainwire-mk2.html#post4356396
Can be set both no GFB and CFA. Same components, distortion numbers under any audibility threshold... And IT DO sound different.
Samething about power regulated supplies. I don't know exactly what happens, but i avoid them if i can, while i know why i prefer batteries ;-)
The main difference is with details or fast transients. We will notice details like nails on acoustic guitars that you never noticed before, more realistic and fast drums attacks etc...
About the general discussion in those last pages, it would be a mess if "science" became a religion.
The scientific approach is necessary, and near all what science tells-us is true. But not always sufficient ... do not explain everything.
So, i believe we have to take great care of what say our senses, ensuring they do not fool us ;-).
About feedback, make yourself your opinion with this preamp/headphone amp http://www.diyaudio.com/forums/solid-state/235695-no-nfb-line-amp-gainwire-mk2.html#post4356396
Can be set both no GFB and CFA. Same components, distortion numbers under any audibility threshold... And IT DO sound different.
Samething about power regulated supplies. I don't know exactly what happens, but i avoid them if i can, while i know why i prefer batteries ;-)
Last edited:
Njet.
As I mentioned, the enclosure on the low end and the xovers around the xover frequencies also introduce phase shifts.
On the low end in the case of a BR enclosure, 360 degrees (180 in the pass band), and at the xover frequency in case of LR4 also 360 degrees, 90 degrees per filter order. The LR4 system behaviour, except for the bass, is that of an all pass network with 360 degrees phase shift around the xover frequency. This in spite of the fact that the individual drivers are minimum phase; the addition of their outputs is not.
Last edited:
The forum does not and will not EVER restrict technical discussion. Ever. Lots of other places on the intertubes where people's delicate sensibilities can be protected from any cold dose of reality. So don't worry about that.
45degrees at the F-3dB of a single pole filter and ultimately 90degrees.
LR4 will be 180degrees at crossover, because the two B2 from which it is made will be 90degrees at the F-3dB.
Andrew, that is in and by itself correct, therefore just to prevent confuddlement the following.
At crossover LR4 has 180 degrees, this is what the low driver does up till the xover point; the high driver adds another 180 degrees afterwards for a total around the xover frequency of 360 degrees.
The following shows the bode plots of an LR4 filter @ 2.5KHz.
The phase plot of the high driver is invisible because it is covered by that of the low driver.
It is nice to show because it illustrates the double flash of insight Linkwitz and Riley had that 1) coincident drivers add with 6 dB, so that the filters have to be down 6 dB at the xover frequency. 2) with well choosen Q, .5 in the case of LR4, the phase plots of the two drivers will align perfectly. This prevents bumps or suck outs in the FR around the xover frequency.
So in practice with a two way loudspeaker with vertically aligned drivers, the acoustic amplitude remains constant wrt to frequency, but there will be a vertical directional change/steering through the crossover band ?.
Dan.
Dan.
I believe yes, but what is important in the matter is total energy.
The vertical accidents around the crossover energy will only affect reflections, it is not so important as long the listeners stay in the good vertical axis. Reflections will be randomized and will blur the problem with room own resonances.
If one use a manual focusing camera with a very large aperture and play with the focus one might miss the very fine state where the object become pin sharp. All other focus positions, front or back, are blurry... Did you acquire "focus" in your test?
//
My Behringers run LR4 active crossovers.
In practice, whether in nearfield or farfield listening position the two drivers merge completely seamlessly and the cabinets subjectively appear as single sound sources.
Carpet damped floor and shallow tweeter horn conspire to dispense with room reflection issues.
Any vertical dispersion issues related to crossover/driver phasing in this case is effectively a non issue, and the in room FR is very nicely subjectively flat.
I have spent extensive time listening to Duntech Sovereigns with their 6dB crossovers and driver positional time alignment which combine to give wonderfully flat power response and depth imaging.
IME the LR4 approach is another solution that gives subjectively excellent power response and depth imaging also.
Dan.
Last edited:
Thanks, but all thanks go to Barleywater, who put me on the trail of this exiting possibility to do FIR filtering for no money at all.
TNT, the pin point accuracy of my speakers does not seem to be influenced by them being in minimum or linear phase mode. Enclosure diffraction, or rather the absence of it, and a flat FR play more important roles.
If you have a speaker near a large surface.... such as a floor, wont that coupling modify the cross-over response -- amp/slope/phase? You dont get the sim response with ideal roll-off slope/rate. Instead of 3db/oct it will be 2.5 etc.
The final adjustment of the cross-over values needs to be done with modelling that includes a surface near it. Now tall cabinets with drivers higher and further from the floor will see less affect.... only the bass/mid will have to be tweeked for the room interaction on the cross-overs.
What I did was to make a cross-over (analog at the time) with variable freq and variable slope and then measured the result of fine tuning the cross-over in the chosen room and placement. Its the only way you can truely get those 90 and 180 degree phase and exact slopes needed. I found tuning the cross-over in the room for best square wave worked well.
[sorry. its just the finer details matter to me]
THx-RNMarsh
The final adjustment of the cross-over values needs to be done with modelling that includes a surface near it. Now tall cabinets with drivers higher and further from the floor will see less affect.... only the bass/mid will have to be tweeked for the room interaction on the cross-overs.
What I did was to make a cross-over (analog at the time) with variable freq and variable slope and then measured the result of fine tuning the cross-over in the chosen room and placement. Its the only way you can truely get those 90 and 180 degree phase and exact slopes needed. I found tuning the cross-over in the room for best square wave worked well.
[sorry. its just the finer details matter to me]
THx-RNMarsh
Last edited:
Richard,
That is fine in the case of the diy builder or someone who has the test gear to measure in room response. Most manufacturers don't have that freedom of design, you can't predict the placement in the room or even the size of the room. What do you do? I know I determined that the L/R 4 passive network had the least issues for something where you had not further control once they left your hands. Even with a dsp based network with all the bells and whistles as a manufacturer you will still have to make a choice that may be great in one room and not when somebody places the speaker in a corner or somewhere else that changes so many things. I guess if you were extremely adept at all this on the digital side of things and I know in Pro audio things are different, but on the consumer side of things you could include a microphone that would be placed in the listening position and would auto tune the system, but even here things can be tough as no two rooms are going to react the same. I don't think on the consumer side of things there will ever be a perfect answer. In your case you have the advantage of all the test gear and the knowledge to change things to suit your particular tastes in what you want to hear.
That is fine in the case of the diy builder or someone who has the test gear to measure in room response. Most manufacturers don't have that freedom of design, you can't predict the placement in the room or even the size of the room. What do you do? I know I determined that the L/R 4 passive network had the least issues for something where you had not further control once they left your hands. Even with a dsp based network with all the bells and whistles as a manufacturer you will still have to make a choice that may be great in one room and not when somebody places the speaker in a corner or somewhere else that changes so many things. I guess if you were extremely adept at all this on the digital side of things and I know in Pro audio things are different, but on the consumer side of things you could include a microphone that would be placed in the listening position and would auto tune the system, but even here things can be tough as no two rooms are going to react the same. I don't think on the consumer side of things there will ever be a perfect answer. In your case you have the advantage of all the test gear and the knowledge to change things to suit your particular tastes in what you want to hear.
All depends too of of the FC frequency. When the FC is low enough and a large band medium, i founs no problem with vertical placement If Richard is right for little enclosures like satellites, we usually measure and tune columns "on the ground".
That is a great paper, tantalizing close to the end when he speculates on application to nonlinear systems. Thanks!
Sims are immensely helpful but treacherous unless you know pretty well what you are doing already. I resisted simulation for many years, but that time calculating "by hand" and building lots of stuff was fruitful.
I had a moment to go back and plug some of the stranger designs into the simulator a few years ago. It was reassuring to find I'd done a pretty decent job of optimization back in the 70's and early 80's.
Nice to see you are exploring.
How about some details on measurement conditions, and a posting of IR_before.wav and IR_after.wav?
Once FR is relatively flat most noticeable improvements with phase are in roughly 20-500Hz region. Attack of acoustic instruments and voice become more realistic.
In near-field consistent recognizable square waves from 70Hz-3kHz; should be observable. Convolution of windowed IR from further away with square waves should also produce decent observable result over broad range.
Perhaps you are ready to play around with DRC as presented by gmad in A convolution based alternative to electrical loudspeaker correction networks? No more difficult than re-Phase.
- Status
- Not open for further replies.