On the bottom right hand corner picture on page 8 of the wavelet representation of the impulse response, you can see higher order modes.
The vertical division is one millisecond and you can realize the picture is possibly coarser than the reality it presents: http://forums.melaudia.net/attachment.php?aid=1760
I think what's notable is that the time scale seems pretty close to that in Geddes's paper: CiteSeerX — Audibility of Linear Distortion with Variations in Sound Pressure Level and Group Delay
That paper might be a simple piece of work but it does show that sensitivity to delayed phenomena such as HOMs does in increase with increased SPLs. It's interesting that the actual tests were done at relatively low levels (71, 74, 77 and 80 dB SPL).
I'm not sure what you mean here but in my view the real problems with diffraction phenomena are (1) as SPL increases, timbre changes towards the unnatural (eg voices get darker), and (2) the experience tends to TOO LOUD, even nasty. As you know, a good speaker doesn't sound loud or nasty even at SPLs much higher than Geddes's test.
Earl will probably tell you it can be done at great cost in time (which is money, or at least, lost opportunity in some other area).
Very good catch! You see this test was done at a public university which had to conform to OSHA (and some others I seem to remember) standards. believe it or not, to have gone above 80 dB would have required a separate set of approvals and a higher level of testing oversight which we simply could not accommodate. One can only imagine how audible things would get at 90, 100 dB!
10 dB more would have been doable with far greater effort.
20 dB more could not be done in a public setting period. It couldn't even be done at any company that feared a potential law suite. It could only be done at that level by some rogue audiophile!!
Bear - we do not seem to be on the same page. There is no "second" diffraction source, only the original one.
There is a single pulse, then a smaller pulse that follows which represent the diffraction. These two pulse arrive at different separation times depending on spatial location. OK, so far. An FFT of these two pulse would yield the frequency response effect of the diffraction.
Now, lets say that we want to correct that response so that there is no effect of the diffraction. Ok? Its easy to see that if there wasn't a second pulse - the diffraction part of the signal - that there wouldn't be any effect. Right?
So lets hypothesize making an equalizer that adds in the exact reverse of the second impulse at exactly the right time - kind of like active noise control. Wahla! The diffraction pulse is cancelled by the EQ pulse in exact anti-phase and the resulting frequency response is flat as a board.
Not so fast! Move the mic a few inches away. Now the difference in arrival of the diffraction signal and the direct signal has changed. The EQ filter still adds in the phase inverted impulse, except that now the two - the original diffraction impulse and the EQ impulse don't exactly line up anymore. They create a doublet response which gets added to the direct signal. the end result is that the response with the added EQ is now worse than it was before, not better.
The response can only be exactly corrected at a single point and gets worse everywhere else.
I cannot explain it in any more detail. If this still doesn't work for you then I suggest we just move on.
Jeepers! What a lot of name-dropping here. Could be just two guys with catalogs chatting back and forth. And horny? Horny for what? More horns--see, it is arrational!
This.
An FFT correction that is only valid for the tiny point in space represented by the 1/2" microphone capsule is not useful, no matter how impressive the measurements are. At an absolute minimum, the FFT correction would have to equally good at another point in space a few inches away ... the location of the other ear. By "equally good", I mean precisely phase-matched, preferably within 5 to 10 degrees, and precisely amplitude-matched, within 0.5 dB.
That's a tall order when there are ripples in amplitude, phase, and time across the frontal arc. If the ripples are several feet across, no problem, but if they are only a few inches across, then fine-grained equalization is a lost cause.
Guess where small spatial ripples come from? Diffraction and reflection, or drivers working in the breakup region. Part of the reason to minimize spatial ripples is to make mike measurements less sensitive to exact microphone location. The old-timers here know that small wiggles and ripples as you sweep the microphone spatially is a bad sign for the loudspeaker ... an indication of crossover phasing problems, driver roughness, cabinet-edge reflections, or similar problems.
Last edited:
Ok, language is an interesting thing.
My question revolved around your use of this phrase earlier:
In specific "two diffraction signals".
But based on your last post #244, there is nothing unusual going on, nothing that at least I did not know before.
Back to our irregular programming...
Tnx.
My question revolved around your use of this phrase earlier:
In specific "two diffraction signals".
But based on your last post #244, there is nothing unusual going on, nothing that at least I did not know before.
Back to our irregular programming...
Tnx.
My direct driver with 845 DHT 20W/20W can do over 100db without a sweat. The sound for me is not perfect but is still exceptional, so why bother with any horn, if it cannot at least give the same quality at least with a higher dynamic range say 120 dB I cannot go horn and pick up a whole load of problems that direct drivers avoid. There is no dirty secret with simple direct drivers
I am bothering because it interests me, but if I am paying for it, it has to be better. I look at ALL the technical papers and consider ALL the parameters and place them in my mind in order of importance.
Of course we can do what is frequently done in room correction: take some spatial averaging and equalize the average rather than response of any exact location. I don't know why we couldn't take a horn and average the response across a +-15 degree lateral and vertical window. If the diffraction effects are very space specific they will disappear in the average and no harm will be done. I would think the HOM effects would be pretty consistent over such a range of angles as they are all coming from the throat region, as I understand. Diffraction from the mouth may be a different matter.
Beyond that, if we feel we can confine ourselves to one point in space, why not just DSP correct to perfection? Wouldn't that get rid of any response artifact, whether caused by classical effects or modern effects? If diffraction is in the impulse response, and since it is a linear effect it doesn't vary with level, why not simply EQ it out?
David
I've heard systems that claimed to do that. Very good indeed, but not the best systems I've ever heard. Being out of the EQ'd spot was still OK, but not as good as in it.
As always, the polar performance is the key. If a horn has very good polars then the EQ will hold for a greater angle off of the listening axis.
In fact, if a horn has really good polars you have to wonder why you don't see those nasty diffraction effects as you shift measurement position?
David
In fact, if a horn has really good polars you have to wonder why you don't see those nasty diffraction effects as you shift measurement position?
David
Attachments
Do or do not the diffractive components vary 1:1 with amplitude?
Or do they conform to a relationship that is not linear with amplitude?
If the former, EQ may bring some benefits I'd guess. If the latter, then standard EQ will not work except around a particular range of SPL... some sort of DSP solution might be useful.
I'd rather try to eliminate or mediate the sources of undesirable diffraction if possible.
_-_-
Or do they conform to a relationship that is not linear with amplitude?
If the former, EQ may bring some benefits I'd guess. If the latter, then standard EQ will not work except around a particular range of SPL... some sort of DSP solution might be useful.
I'd rather try to eliminate or mediate the sources of undesirable diffraction if possible.
_-_-
Earlier I said Peavey horns might not be ideal. Reading over Million, Keele , Dave Smith (our Speaker Dave AES papers. Scratch that. EV JBL Peavey combo big CD horns are paramount period.
Last edited:
On this forum, hampered by phone, I sometimes give the intellects here credit for deciphering my "stream of thought" shorthand, this question was not necessarily aimed at you, sorry, I'll be clearer. Perhaps real time diffraction models could be like computer weather models, real time "rough" then clarity later with rendering time. There ya go.
Last edited:
The thought of not averaging the EQ of a big wide CD horn seems bizarre to me. But again, here from the swamp common sense being either ignored or hailed triumphantly seems odd. Peavey said. something to that effect at Namm a while back .
I have some big EV horns with EV drivers that are really flat if you don't try and use them too high. 650 to around 4K no EQ is needed, then hand off to a small format 1" driver/horn with the same pattern. Sounds great. Same can be done with your Peavey, I have the old 22 model with the salt shaker phase plug/ square magnet and in the right horn it's really flat 1.2 to around 8K - just add a tweeter
Yeah Pooh. It's my conclusion that the EV-Peavey, and many other butt type, Exponent/Radial combo horns and other horns of this type are in fact, if you are not going the whole synergy/conical route with extra drivers, the way to go. I just do not like pushing that xover high as required with far flare rate cones. I feel like the large horns with smooth cheeks respond to EQ very well. One more issue, cutting between direct radiatior B&W 801 s3 and Peavey Sp-1 EQed, biamped time aligned (LR-24) at around 100 db quite clearly sounds clearer more derailed, ie you hear reverb decay and timing everything is tighter. I would expect some of these qualties are inherent to most compression driver speakers even the compromised shallow ones that do not load the low EGs well and require higher xovers. What I an saying and this come as no news to people like SpeakerDave, is that compression drive speakers are IMO in almost ever case better ref. monitors. I might also mention that whatever the non-linear compression distortion component is, it is a damn site less than a direct radiator trying to go to those levels. No contest. Note to grumpy Englio-philes , jotted from HTC Desire. Hope this is coherent.
Last edited:
- Status
- Not open for further replies.