 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
219.113.104.20
Please help me to think about the issue(s) of time delay/phase coherency in rear-loaded horns.Do sound waves traveling through a horn propagate at the same speed as sound waves in open air? Faster? Slower?
The sound in a rear-loaded horn is coming off the back side of the cone, so at the point of sound generation it is already 180 degrees out of phase with the sound coming off the front of driver. Ideally the sound upon emerging from the horn would be perfectly in phase with the sound coming off the front of the driver. In other words, ideally it would have traveled through half a wavelength by the time it emerges from the front of the horn. Or at least that is what my noggin' is telling me as I think about it. Obviously perfect phase alignment with the front of the transducer can only be accomplished for a single frequency, but I would suppose that if properly implemented, the results would be close enough for rock-n-roll, and also for just about everything else.
If this line of thinking is correct, then, for example, the Yamamura-Churchill Dionisio 27 with its horn length of 5.4 meters would seem to be pretty good at turning that bass around and getting it more or less properly aligned with what's coming off the front of the driver.
Am I on the right track here, or am I overlooking something important?
Next question:
Even assuming that the sound coming off the rear of the driver can be returned to the front in proper phase alignment with the sound coming off the front of the cone, obviously there is a certain time delay -- the amount of time that it takes to travel through the horn. Taking the example of the Yamamura-Churchill Dionisio 27, this is 5.4 meters. Is this a delay that the human mind can detect? In other words (prejudices aside -- let's posit a blind listening text) is it something that some people with good ears could detect and find objectionable? Or is it something that ought (prejudices aside) to be acceptable to the ears of any listener? Again, this assumes that the horn is not engendering other problems such as resonances, etc.
Thanks for enlightening me!
Chris
Follow Ups:
Nice topic Chris,Made me think of this wonderful early system which had exactly the opposite problem. I suspect that midrange lagging the bass may be worse.
Recordings of tap dance routines may be a good test for phase discrepancies.Quote from an article on the site linked below:
"The most significant deficiency, in John Hilliard's mind, was the phase discrepancies between the WE555's and the other drivers. The 12' horn paths so delayed the midrange response that monitoring tests of a tap dance routine resulted in two taps being heard for every one recorded."Regards Philip
"The 12' horn paths so delayed the midrange response that monitoring tests of a tap dance routine resulted in two taps being heard for every one recorded."That would be the famous Jane Powell case, circa 1937 or thereabouts, which showed that even very short midrange delay times could be perceived. It was the impetus behind the earliest time-align work, achieved by moving the midrange radiator forward or backward until the 'double hit' disappeared.
When I tried to improve a rear-loaded horn design I was working on a few years back, I quickly went to using as much stuffing as I could manage without blocking the horn path in an attempt to smooth out the notches. It seemed to me that the half cycle delayed rear wave in phase with the front wave provided less than the most convincing sense of transient impact for its frequency extension (at approx 50 hz), although still better than a bass reflex port tuned to the same frequency. OTOH, the relatively small BR port itself provided better backwave filtering into the lower mids with a small amount of stuffing than the back horn with a lot of stuffing, deteriorating the soundstage less. I've since gone to BR tuned well below 40hz and down about 6db at Fb to minimize the audibility of its transient shortfalls. With my passive bass boost circuit, I can pull this off with decent bass weight using a 2220 in a modest sized box (~3.5 ft 3 ).
Maybe someone has the article referred to in this text?http://www.trueaudio.com/post_010.htm
Chris- You may find the link below interesting as it talks about the issue you raise. The reviewer attempts to make a positive out of the "staged" attack of back loaded horns but I share your concerns.
Again, to take a "best case" example like the Yamamura Churchill Dionisio 27, the low frequencies being amplified by the horn are more or less in phase with the sound coming off the front of the speaker cone, but they are 0.016 seconds "behind." If steady signals were being reproduced I am sure this would not be a problem, but I worry that it might be in the dynamic context of musical reproduction. If -- and due to my inexperience and lack of study I can't say one way or the other with any confidence -- if the human brain is capable of perceiving such a time delay with regard to low frequencies, then we are faced with the problem of the music being continually muddied or smeared on account of this constant delay. On the other hand, if the nature of psychoacoustics is such that we can't process such information at low frequencies, then the fact that the two waves have more or less proper phase alignment may make for a truly superb fullrange horn loudspeaker. I would love to hear from a source that could speak with some authority on this subject! Thanks!
ChrisThe Haas effect (or preceedence effect) is appropriate to your questions. The Haas effect states that delayed sounds (and direct sounds) are integrated with each other by our psycho/acoustic hearing aparatus if they are perceived together within 20 to 40 milliseconds of the direct sound. That is to say the delayed sounds which fall within this window are perceived by us as artifacts of the original sound, while sounds delayed outside of this window are perceived as being connected with artifacts of the acoustic space the direct sound occured in (i.e.: reverberation). In a back loaded horn, the main area of concern in this discussion is the crossover region where the output of the direct radiator front and the output of the rear loaded horn overlap each other. I've never noticed a particulary "reverby" sound with a backloader, and I have never seen this referenced as an artifact or fault in any review of one. This does not preclude this being inaudible, as some people just don't care for the sound of a ported box and prefer a closed box direct radiator. Also, don't forget that a direct radiator and a horn loaded driver have a constant phase difference of 90 degrees between the two drivers transfer characteristics, and the driver in a back loaded horn with direct radiator front has this integration problem as well. So how important is all of this in the real world? Well, one can expect some ripple in the crossover region with a backloader, but clever design can minimize this, and anyway the whole thing will hopefully get lost in what the room does with the sound wave after it moves some distance away from the speaker. You might like the resultant speaker (and accept it as a fair compromise), or it might drive you nuts if you are of such a tempermant. The real thing to obsess about is the midrange IMHO.
Paul
PaulFirst, I like your last statement "The real thing to obsess about is the midrange IMHO" as to me this is all too true.
As to the first question asked. The wave speed in a horn does appear to be variable at the throat, being faster in the throat and slowing down to be the standard speed at the mouth. Interestingly, Websters EQ. (the wrong one) predicts that an exponential horn has an infinite wave speed near its LF cutoff - curious.
Now to the audibility of a delay. The Haas effect applies to the apparent location of the sound source with two signal, one dealyed, where the first arrival has "precidence" over later arrivals less than about 20 ms. Above 20 ms (or so) the second sound is heard as an echo.
A delay of 16 ms. is enourmous and would be perceived as strong sound coloration at low frequencies and a marked audibity at HF as group delay distortion even down to a few ms. The situation of audibility changes dramatically above and below about 1 kHz because the way the ear processes signals is completely different in thes two regions. So below about 100 Hz 16 ms would probably not be a problem, maybe some, but in any real room the room would mask everthing any way. Between 100-500 Hz this amount of delay would cause a strong coloration and above 500 Hz would begin to sound very bad as it became many periods of the sound long. It would begin to actually sound like an echo about about 2 kHz.
So there is no simple answer.
Earl Geddes
Dr. GeddesMy comments were in the context of the crossover region of a backloaded horn and the direct radiator front of the driver, corresponding to (very) roughly the region of 100 to 200 Hz. You are entirely correct to point out that the audibility of the delay is much greater at higher frequencies. As to the "window" of the Haas effect, I have seen it being referenced differently by different sources. I reached for Everest's "Master Handbook of Acoustics, 4th ed.", mainly because it was handy. Everest defines it as being 20 to 40 msec, but I have also seen it as 20 to 30 msec. It seems that even the experts can't completely agree on this. Your comments seem to indicate that you consider the effect to be acceptable in audio if it occurs between between 0 and 20 msec., which is quite a bit more conservative than the 20 to 30 msec range, but this is fair enough in the context of whether this is audible in the performace of a speaker. Most of the backloaded horns referenced in this discussuion (with the exception of the giant Dionisio) have horn lengths in the 6 to 9 ft. range and (assuming that sound travels roughly 1 ft. per msec.) it would seem that most popular backloaders fall well within your 20 msec. range. Some have complained that they can hear a "reverby" sound in a ported box, but oddly enough this critisizm is not ususally leveled at the backloaded horn, though perhaps following this discussion it will be. Hopefully no one began construction of a horn where the tweeter is 16 ft. in front of the midrange horn following my first post.
I have a superficial familarity with Webster's horn equations, but I'm quite surprised that an infinite wave speed is assumed at the low freq. cuttoff (of an infinite exponential horn presumably). Infinities are usually to be avoided in physics, and generally imply that there is some fault with the theory. At some point the wave must exceed the speed of light (before it reaches infinity) which would cause time reversal effects according to Relativity Theory. The wave would then return back to the driver and cancel itself out before reaching the cutoff. Therefore I predict that no sound could be produced by such a device ; )
I have to confess that I have a short attention span, and am easily distracted by the music coming out of any speaker, unless the speaker is doing something bad.
Paul
Regarding the Haas effect, or the precidence effect as it is also called, has to do with image location and the fact that as far as the location of a source is concerned, the first arrival dominates if the second is within 20-30 ms. (The time depends on a great many factors which is why it is so variable.) The Haas effect says nothing about coloration of the sound which will certainly occur under 20-30 ms. and is a very big problem. So I must not have been clear on that point."Your comments seem to indicate that you consider the effect to be acceptable in audio if it occurs between between 0 and 20 msec."
Clearly this is not the case at all. I would prefer to not see any secondary sound, reflections, diffraction, delayed rear, whatever in this time band - it is not "acceptable". That of course is really difficult! but less is better.
"I'm quite surprised that an infinite wave speed is assumed at the low freq. cuttoff"
Actually its not assumed, it falls out of the results. These details get difficult to discuss in posts, but suffice it to say that what we think of as "normal" wave propagation "probably" does not occur in the throat of a horn or waveguide. That is because it appears that the wavelengths of the sound are changing as the sound propagates. Now we can either consider this a change in wave speed or a change in wavelength, but clearly something odd is going on. In an exponential horn at "cutoff" the wavelength goes to infinity. This is shown in Wood "Acoustics", page 117. he shows that c' the wave speed in the device goes as
c'=c * k /(k^2 - m^2/4)^1/2
where k is the wave number and m the flare rate. When m/2 = k, then c' goes to infinity. Of course we know that this does not happen, but somehow these details of Websters Equation seem to have been overlooked.
Earl Geddes
Dr. GeddesI'm afraid that your position (i.e.: "The Haas effect says nothiong about coloration of the sound...") puts you at odds with Everest and Haas. From the Glossary of The Master Handbook of Acoustics: "Haas effect See fusion zone. Also called the preceedence effect. Delayed sounds are integrated by the auditory aparatus if they fall on the ear within 20 to 40 msec of the direct sound. The level of the delayed components contributes to the apparent level of the sounds, and it is accompanied by a pleasant change in character." The entry for Fusion Zone also contains the phrase "...pleasant change of character." Haas himself is quoted on p.74 concerning early reflections which "...result in a pleasant modification of the sound impression in the sense of broadening of the primary sound source while the echo source is not perceived acoustically." Whether this "modification" and "broadening" is a big problem in the context of the backloaded horn will eventually come down to an argument of design trade-offs and aesthetics. I think it's a non-issue in the typical backloader, but you of course may disagree, and require a more purist design aesthetic.
Thankyou for correcting my misunderstanding of the consequences of an infinite wave speed in the Webster horn. So when the flare rate divided by two is equal to the wave number, the wave speed goes to infinity. So one does'nt really need an infinite exponential horn, as I had assumed, to reach a wave speed equaling the speed of light with Webster's equations. An interesting critique of Webster.
Paul
I would have to say that someone may be taking the Hass quotes out of context. The description in Blauert "Spatial Hearing" Pg. 226 states that "At delay times less than 50 ms. echoes are no longer perceived as annoying even if the reflection is considerably stronger than the primary sound" as the Haas efect. This does not mean that these echoes are perceived as beneficial and in fact Bluert makes a strong point of this (could not find the page) that the studies of the perception of an echo were quite independent of questions of image shift and coloration. The later effects were not factors in their experiments. Subjects were not asked is these effects occured, only if they could detect an echo.In room acoustics it is generally held that reflections < 30 ms. - the "sound fusion" region - are generally bad things because they cause coloration (from comb filtering) and image shifts since the ear cannot sort out the refection from the direct sound.
The quotes from Haas appear completely contradictory to this. Pleasantness, in Haas's context, may be actually be sound that is "colored" with a wide "image blur", although these things are widely held by acousticians to be negative aspects for good "sound reproduction".
I agree that from what you say I may be at odds with Everest and Haas, but modern thinking along this line is pretty much along the lines that I have quoted. At least it is for me.
Earl Geddes
Dr. GeddesThe Haas quote was in the context of Everest's sub chapter "The Precedence Effect" (spelling correct this time, as we both have mispelled it). Haas' translated JAES paper "The influence of a single echo on the audibility of speech" is referenced. Everest and Haas are describing a percieved effect, and not necessarily promoting the said "pleasant" effect as an acoustical design tool. Everest notes that the transition zone between the areas where the delayed sounds are integrated with the direct sound, and the point where they are perceived as discrete echos is "...gradual, and therefore, somewhat indefinite." He states that some define this transition as 80 msec, and others at 100 msec as the point where there can be no question that a discrete echo is perceived. He settles on the first 30 msec to define the area of integration. It's interesting that Blauert uses 50 msec and introduces an "annoying" factor. Even more interesting is that the Haas' paper was originally published in
German in 1951, but was'nt translated to English (in the JAES version) untill 1972. Blauert's book "Spatial Hearing, 1983" is referenced by Everest in a discussion of image localization and how the perception of which changes considerably with frequency. Everest notes that "This is an active area of research that is being cotinually refined." Haas' focus was speech inteligbility so one can assume that recorded speech was used in the original study, while Blauert seems to have been using test signals (100 to 10K Hz are mentioned by Everest). Considering the different test situations and areas of focus, it does'nt seem to me that these experts are really at odds with each other.Paul
but it seemed to me that you were saying that early reflections made the sound "better" or "more pleasant" and were using Haas and Everest as support for this position. Did I misunderstand you?
Earl Geddes
Dr. GeddesI was not attempting to take either position, that early reflections made the sound better or worse. I don't believe Everest and Haas were taking one position over another here either, though their comments could probably be used to support either position. If "pleasant" colorations were found, I believe Haas was just reporting this. "Pleasant" in this context speaks for itself. However if someone observed that "pleasant" early reflection caused colorations in the mid-bass had a tendency to smear the stereo image, I would see no point in arguing against this either. Ed Dell once said "He who awaits perfection is constantly imobilized", or...
"I'm not perfect, but I'm perfect for you": Grace Jones
Paul
I am not seeking perfection, and I don't think that a discussion of early reflections in small rooms is academic or arbitrary. But if there is a better or worse way to do something then I certainly want to know.
Earl Geddes
nt
I've played around with horn subs and delay lines and have found that below 80 Hz the delay of the wave from a sub horn isn't audible. At even 100 Hz the effect is very hard to discern, due to room interactions that mask any delay. Moreover, what also must be considered is the 'first in first served' psychoacoustic effect, that whichever frequency the brain first processes gets preferential treatment, as it were. If the midbass and higher frequencies lag the low fundamentals, as is the case with a direct radiator sub and a horn-loaded midbass, then the sound would be perceived as muddy or indistinct. When the midbass and higher frequencies lead the fundamentals the effect is a more immediate sound, a fact not lost on the folks at Barcus-Berry. Once you enter the realm of the midrange Paul is right on, at 500 Hz even 7ms of misalign in either direction is quite audible. But losing sleep over 16ms delay at 80 Hz one should not do.
BillI think you meant to say "Once you enter the realm of the midrange EARL is right on...". He introduced the concept of how the various effects are heavily frequency dependent to the present discussion. I had been talking only about the bass and midbass region.
Paul
This is the specific phrase I was referring to, and based on my own observations agree with:"So below about 100 Hz 16 ms would probably not be a problem, maybe some, but in any real room the room would mask everthing any way. Between 100-500 Hz this amount of delay would cause a strong coloration and above 500 Hz would begin to sound very bad as it became many periods of the sound long. It would begin to actually sound like an echo about about 2 kHz."
Although 16ms of delay will do the Jane Powell thing a lot lower than 2kHz, as we well know. But back to the matter of what's going on below 100Hz, not only does the path length derived delay of a horn sub not create audible problems, neither does group delay in the same frequency range. Too many people waste too much time and effort trying to arrive at less than ten degrees GD at 20Hz when in truth it really doesn't matter anyway.
BillAs always the short posts often prevent a full discussion, because some of what you same comes with a lot of caveats. Like "first in first served" or the "precidence" effect is mostly an image location phenomina and does not work beyond certain various times at variuos frequencies, and is overshadowed by other effects at much shorter times at higher frequencies. There is almost nothing about hearing that one can say "the ear does this" and have it applicable in all circumstances. It may do something at LF but not at high, or at higher SPL but not at low, or visa versa - in both cases. That's why we can all get into trouble here. Because we make statements knowing ourselves what the limits were in our minds, but others may not.
7 ms. at 500 Hz being audible would be on the edge of what I have read. Higher frequency than that - certainly, longer times - certainly, but 7 ms at 500 Hz would be pushing the limits. It is also SPL dependent.
Earl Geddes
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]
