It's one of those things that if you cannot experience it yourself, nothing others describe will make sense.
Well isn't that just fine and dandy.It's one of those things that if you cannot experience it yourself, nothing others describe will make sense.
If you ever come to Taiwan, I'd be glad to arrange an anechoic chamber to conduct the tests.
Obviously you designed the cross and know the amplitude crossover point for sure
But looking at the resulting phase, it appears to follow the behavior of a LR 24d/oct filter at 1200Hz.
It is quite common that when choosing a given acoustical amplitude slope the phase seems to shift "faster" (earlier) than expected, because of fast amplitude variations down low. In a minimal phase system the phase at one frequency can be influenced by amplitude variations several octaves below, and this is often the case with horns (and waveguides
) that "load" the driver (even waveguide do, right?) to a point, and then cause abrupt amplitude variation below.In some of my tests I found that even after having obtained what looked like a textbook LR high pass filter amplitude-wise, phase seemed to rotate more than a textbook LR would have implied, and I had to use *heavy* (minimum phase) EQ one decade below to get the phase back to where it was supposed to be (Linkwitz Riley transform).
But of course all this is only academic and as long as the phase is coherent between drivers all is good
I don't think your designs would benefit a lot from phase linearization, but others will (those with bass or midbass crossovers, or higher order slopes), and this can certainly be heard.
Earl,
I am quit surprised that you are only using a first order filter that low on a compression driver on a fairly small horn and not having problems with over excursion at high levels? That almost sounds like a recipe for disaster but you obviously are not seeing that or you would be having a world of problems with warranty returns.
I am quit surprised that you are only using a first order filter that low on a compression driver on a fairly small horn and not having problems with over excursion at high levels? That almost sounds like a recipe for disaster but you obviously are not seeing that or you would be having a world of problems with warranty returns.
It is quite common that when choosing a given acoustical amplitude slope the phase seems to shift "faster" (earlier) than expected, because of fast amplitude variations down low. In a minimal phase system the phase at one frequency can be influenced by amplitude variations several octaves below, and this is often the case with horns (and waveguides
Yes, the phase variation at or near the "cutoff" (as much as I hate that term) can be quite different than what one finds in electrical circuits. That's because there is no electrical equivalent of the radiation impedance. All devices "load", my complaint is that the terms and beliefs that derive from Webster's approach are not accurate and should be abandoned (but alas "Horn Vampires" never seem to die.) The loading does not ever have as sharp a transition as Horn Theory would suggest. This can be shown in two ways - using a theory that is actually correct AND actually measuring the impedance of a real device. I have done both. The loading does steepen as the waveguide gets narrower, that much is true, but there is never a "cutoff" in the same sense as modal cutoff, where the mode really does stop completely (or nearly) below some frequency. (It actually becomes evanescent, i.e. a complex wave number.) The lowest modes in any device never do that.
The other big misunderstanding about "horns" is that they all "load" exactly the same above the "transition range" (AKA "cutoff".) Hence "loading" is only a factor at the extreme low frequency end of the devices passband. "Loading" and 'cutoff" are concepts that should die but have somehow continued to live well beyond their actual death.
Last edited:
Correct about failures - never seen one except in speakers in clubs. But those failures were from HF content heating the voice coil (probably a clipped amp) and burning it out. I have never have seen any excursion related problem.
But you need to understand something. A waveguide has a constant -6 dB/oct response in it pass band - it is not flat - it does not look anything like the response shown by B&C in their data sheets. So a 6 dB/oct HP filter set at the upper edge of the devices passband acts to flatten the response. Since the lower end of the passband is several octaves below the upper edge the actual attenuation of the excursion at the lower end is ten to a hundred times less than the upper end. The actual calculated filter frequency for the 6 dB/oct HP filter is actually about 12 kHz.
Designing filters for waveguides is quite a bit different than for horns and both of those are completely different than for direct radiators.
Last edited:
The same would be true coming back at you. The idea is to seek the truth. Unless the listening tests are blind and controlled there is no truth available to be learned. Anechoic chambers don't help, unless you turn the lights out and let someone else control the test.
Earl,
Isn't the actual attenuation of the excursion at the lower end ten to a hundred times more (not less) than the upper end?
Art
That is fine, we can have only the speakers in the anechoic chamber, and let someone else control the switch outside the chamber. We can both sit in the room and make judgement on each passage played and see how they match what the controller did. This will be interesting, you and I doing listening, let me know when you plan to stop by Taiwan.
Last edited:
Pos,
Phase response and amplitude response are locked together in minimum-phase systems, and this is what is normally considered a phase response of the system/loudspeaker.
There is only one "minimum-phase" phase response of a loudspeaker. You just need to determine this correctly. The SPL above 17kHz drops at the rate of ~120dB/oct and phase response must reflect this rapid change, regardless of the waveguide design.
"..."hiding" the low pass behavior of the compression driver"... as you put it, is nothing short of generating an incorrect phase response.
Best Regards,
Bohdan
Assuming the loudspeaker is MP, that is not always the case. I suspect that it is the case with my speakers. But it would probably not be the case if the horn were a diffraction device, because then the internal reflections could create a nonMP situation.
Hi Earl,
As you can see from other posts, phase response seems to be a “free for all” commodity these days. This is mainly because just about everybody experiences difficulties setting the correct start of the half-window during FFT process of converting IR to FR to get the true driver's minimum-phase response (you can translate this statement into the corresponding procedure of your individual measurement system)
This is not a trivial issue, and I have highlighted similar problem about 15 years ago in a short paper
http://www.bodziosoftware.com.au/Acoustic Center Evaluation.zip
This paper offers some guidance for setting the FFT windowing parameters, so that measured phase response is indeed as close the “minimum-phase” response as possible. As you can see from this paper, even including or excluding driver’s break-up region does not influence guiding principles.
In this current thread, I have posted a simple goal – to determine what is Abbey’s phase response. And as you can see – the same problem pops up immediately.
Your input would include providing enough crossover information, so that a theoretical model could be constructed first, and this model would be used as a guidance for setting the FFT windowing parameters. Providing IRs of individual drivers would be of great help.
Other than that, one can move the IR on the time scale to one’s heart content and claim, that every setting is the correct one, even though tweeter’s response would clearly violated minimum-phase principle.
Best Regards,
Bohdan
That is the goal isn't it? (It is mine.) Or do we need to include "Unless it sounds good to me some other way"?
But the phase would not be flat, it would have to have an every increasing slope with frequency or it would not be MP.
Hi Kindhornman,
Thank you for your comment.
This has been done already. http://www.bodziosoftware.com.au/Home_Theatre_Conclusions.pdf
Best Regards,
Bohdan
It is if you are not obsesed with absolute phase.
I am not willing to provide details on the crossover, but I am willing to provide IRs for the individual drivers.
But again, absolute phase for a MP system just does not interest me. The relative phase between the drivers (critical for crossover design) is maintained as long as the parameters for the two systems remain the same. Which I am always careful about.