Although I try to understand the theory as much as possible (I'm no mathematical poet) I like to back up my understanding with observational data as well - theory x says y should happen, do measurements back it up ?
Sure, I haven't made exhaustive study of drivers and baffles, but all the measurements I've made on drivers mounted on baffles using the excess group delay technique have shown that baffle diffraction as we usually talk about it is minimum phase.
Its an easy measurement to make too, anyone with software like ARTA can try it for themselves on any baffle they like, no need to take my word for it. (Just make sure room reflections are correctly windowed out...)
Room reflections can be (and usually are at some frequency) non minimum phase.
But understanding whether something is minimum phase or not let's you know whether it can be corrected, and how.
If diffraction is minimum phase then at least at one point in space it CAN be completely corrected (including phase) by standard minimum phase EQ. If its not minimum phase it can't be corrected in this way. That's a critical piece of information isn't it ?
Of course at high frequencies compensating for diffraction with EQ doesn't work because its only correct at one point in space and radically wrong at relatively small angles from the optimum point...
But it does have relevance for the baffle step correction at lower midrange frequencies. Baffle step is also a result of diffraction but at those low frequencies the change in response as you go off axis is very gradual due to the large wavelength.
If baffle diffraction is minimum phase then minimum phase baffle step EQ CAN perfectly correct the phase response through the baffle step region, at least on one axis, and unlike higher frequencies where the ripples in the polar response are dense, the comparatively gradual change with angle means that the corrected response is still acceptable reasonable amounts off axis.
So I would say the properties and nature of diffraction is useful stuff to know.
Last edited:
This is where I contend that the MP argument is misleading because what you say above is not really the case in general. Just because a source is minimum phase does not mean that its polar response can be "corrected", because it can't. I know what you meant and in that context you are correct, but the MP concept is not very useful in acoustics because it does not apply spatially (as you correctly point out). It is a concept rooted in one dimensional electronics and does not lend much assistance to the acoustics problem.
This is what (I think) Soongsc is trying to say (although he does it badly, with a dismissal of others arguments.)
It is not a matter of who is right or wrong, it's a matter of, so what if one is right or wrong. So what if other people think that diffraction is a minimum phase phenomena? I choose to apply design considering it not minimum phase, and that understanding works fine for me, but considering the other way does not. I just see the discussion not productive, so I choose not to continue.
DBMandrake, let's say you use an ideal impulse on a baffle to generate diffraction. On a rectangular board, because the diffraction occurs at different times, the data you capture is a collection of diffraction caused by the difference in timing. In reality, the total distance traveled is different from each diffraction point. Such difference in distance means that different time of flight removal is required for each point. So a single measurement cannot distinguish such difference, thus you never can measure true minimum phase.
However, since the diffraction at a single measurement point is a one to one relation, with the collective diffraction wave, you can still calculate a phase curve which may have no relevance with how you modify your design. But mathematically, if you assume 0 time of flight and calculate the phase curve, you can still call it a minimum phase curve. I never use this information in a design process because it does not correctly reflect the phenomena of what is happening.
However, since the diffraction at a single measurement point is a one to one relation, with the collective diffraction wave, you can still calculate a phase curve which may have no relevance with how you modify your design. But mathematically, if you assume 0 time of flight and calculate the phase curve, you can still call it a minimum phase curve. I never use this information in a design process because it does not correctly reflect the phenomena of what is happening.
Last edited:
We're looking at a multitude of linear systems (each point in space). If they are minimum phase and similar enough, EQ can be used to improve things. Isn't this useful?
We'll probably never know because instead of discussing the topic he keeps on repeating that he won't discuss the topic 🙂
NO! You are using exactly the misconception that I warned about. You can correct only one of that multitude at a time! You cannot correct them all at once. This is precisely the error that people make. You just did! I discounted "similar enough" because I don't know what that means. The response is either correctable or its not, "almost corrected" is what you will get with "similar enough". The fact is that the "points in space" are virtually never similar, let alone "similar enough".
Only a polar map can show what is correctable with EQ and what is not and minimum phase does not even enter into those calculations.
^
Guess we would need to look at real data which I don't have. Right now we're just having a philosophical discussion.
There's a paper by the Dirac guys where they did show how improving a certain number of points in space did in fact improve all other points too. This is of course related to in-room response optimization but I'd think the concept is transferable. There are points in space that are certainly more important than others. This could be one approach for such an optimization attempt. No?
Guess we would need to look at real data which I don't have. Right now we're just having a philosophical discussion.
There's a paper by the Dirac guys where they did show how improving a certain number of points in space did in fact improve all other points too. This is of course related to in-room response optimization but I'd think the concept is transferable. There are points in space that are certainly more important than others. This could be one approach for such an optimization attempt. No?
I was not too impressed with the Dirac guys.
Optimizing a number of points in space will optimize all points, but one cannot say that there is any solution which will improve all points simultaneously, that's quite a different thing, and it would be impossible.
Are we talking about the steady state field or the direct field here? Completely different situations. I was talking about the polar response and hence the direct field if we are talking about room acoustics. Since this thread is about "waveguides" consideration of a rooms steady state response is not appropriate.
Optimizing a number of points in space will optimize all points, but one cannot say that there is any solution which will improve all points simultaneously, that's quite a different thing, and it would be impossible.
Are we talking about the steady state field or the direct field here? Completely different situations. I was talking about the polar response and hence the direct field if we are talking about room acoustics. Since this thread is about "waveguides" consideration of a rooms steady state response is not appropriate.
I think one could be too focused on the technology of audio and forget about music. The only way you can tell whether they match up is to listen to the resulting system. I think we should really listen to a system based on Dirac technology.
Last edited:
Also FWIW, HOLM runs great for me under Windows XP. With 7 it crashes from time to time. Under Win 7, I save frequently. 😉
Markus,
Earl is right on here. Diffraction and real radiation of driver is not one dimensional. Degree to which it is one dimensional is degree to which EQ may be used to correct overall response.
Earl is right on here. Diffraction and real radiation of driver is not one dimensional. Degree to which it is one dimensional is degree to which EQ may be used to correct overall response.
I have no problem being wrong but did anybody actually try an EQ based optimization approach and looked at how good or bad it performed?
I tend to agree with this with "improve" being the operative phrase. As Earl points out diffraction effects vary in the polar sense, so no universal correction is available. Still, improvement from one vantage point is likely to be in the right direction from surrounding vantage points. This is a matter of using good judgment and not over correcting deep flaws.
We do the same for LF room correction: if response is measured over a broader listening area then the best correction becomes milder and less individually perfect, but gives general improvement for most seats.
As to baffle step, I would certainly assume this was minimum phase and can and should be totally corrected.
Its worth noting that nonminimum phase issues might benefit from correction as well as minimum phase issues. In the end you will have an all pass function (residual phase issues) but the system will be flat. That seems smarter than throwing up your hands and leaving response errors because a system is not minimum phase. Shouldn't let "the perfect be the enemy of the good".
David
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
The above measurements are measured about at 1M, one EQed based on the 1M measurement, the other EQed based on near field measurement. So one EQs the diffraction effects, the other does not. I have also tried EQing near field at center of dust cap, at the voice coil, and midway.
While not conclusive, the 1M EQ did sound a bit fuzzier than near field EQ in the transients, overall sound had darker background with 1M EQ. Ultimate Equalizer software EQ was used.
After this test session, the amp was improved to provide best performance without software EQ, and also implemented hardware EQ till most auditors felt it was optimum.
Last edited:
David
If baffle step is a diffraction issue, as I understand it is, then even if it is Minimum Phase, it cannot be corrected globally.
We're talking about an effect that is spread in space so looking at a single point is useless.
Why would you try EQ'ing the drivers near field response close to the dust cap ? 😕
A near field measurement is not even remotely valid at high frequencies even if the cone were a perfect piston and there were no baffle diffraction effects, let alone when there are cone breakup effects.
You need to be at least 3x the diameter of the cone away before high frequency data starts to become valid just for the driver by itself. To include baffle diffraction you need to be further away again. (In theory 3x the largest baffle dimension)
Automated EQ made at a single point in space is never going to work very well anyway because it can't tell what response aberrations are spatial (diffraction etc) and which are not (cone resonances etc)
Last edited: