Jump to content

In praise of AR3a's


Carlspeak

Recommended Posts

I remember that comment he made about the Quad ESL 63 and the impact of stereo speakers vs a mono one in one of his reports. I wonder if it would also apply to any of the Allison models, and most of all if it would apply to the Bose 901.

I think the point was that the ESL63, with the null towards the side walls, gave very little lateral reflections and so, in mono, the speaker was downgraded by listeners. With stereo, the second speaker could add some spaciousness and the listening impression was improved. As you have recently pointed out, adding channels can give a better compromise between spatial impression and image specificity, both from going from 2 to 3 speakers or even from 1 to 2 speakers. The 901 would be the opposite case, wouldn't it? Lots of lateral reflections would diminish the difference between mono and stereo. (The cynical might say there is no difference between mono and stereo with the 901.)

Finally, regarding floor bounce null artifacts, some clarification needs to be made. With floor bounce, at some frequencies the reflected signals get to the listener's ears at just the right time to generate a cancellation null right there AT THE EARS. Yep, it can happen and at what frequencies it happens at will depend upon the location of the listener in relation to the speaker and the floor boundary. Moving either the speaker or the listener will shift the location of the null.

HOWEVER, this is NOT the same thing as the Allison Effect. With the Allison effect, reflections from large, nearby room boundaries are reflected back to the WOOFER DRIVER itself, and cause the null at that point - generally in the mid-bass region. The location of the notch will depend upon the distances to the boundaries and not to the listener. As a result, the suckout notch is audible everywhere in the room and is related to the power input of the speaker into the room itself. Only moving the speaker will impact the broadness and the location of the null. Moving the listener will not do a thing. One way to deal with the situation is to locate the speaker at relatively unequal distances from each of the three nearest boundaries (floor, front wall, side wall), which will at least broaden it and make it more shallow. A further solution would involve electronic equalization. Of course, with his three-way models Allison located his woofers so that the notch was just above their crossover-controlled operating range and the location of the midrange drivers put the notch just below their operating range. The notching actually further steepened their electrically controlled crossover slopes, and allowed the phenomenon to work with the speakers instead of against them.

The Allsion effect can also impact the bass range below the mid bass, since it can also exist between spaced-apart woofer drivers on different systems in the same installation. For example, two woofer centers 12 feet apart will generate a suckout notch at about 56 Hz, just as will a boundary related reflection from a boundary 6 feet away from a woofer center. One obvious solution to this problem is to use an outboard subwoofer crossed over to the satellite somewhat above 56 Hz, getting any potential notch out of the operating range of the system, with the subwoofer itself located in relation to room boundaries in such a way that any cancellation notches it generates from the three nearest room boundaries would be above the selected crossover frequency.

Using a subwoofer and satellite speakers located in such a way that suckout notches are all outside of the operating ranges of the systems will allow the speakers to all input flat power to the room - assuming they are designed to do that, of course, under non-boundary conditions.

Howard Ferstler

This is "a distinction without a difference". In a three boundary case then reflections "back to the woofer" aren't the issue unless your ears are at the woofer. Like the floor bounce the rear and side wall bounce are only significant at the listener position. It really is best viewed as an image model, a speakers sitting in the corner of the intersection of three mirrors. (I'll elaborate for the benefit of others who may be reading along.)Along with the real speaker you will see 7 reflected image speakers. At whatever listener position you choose you will fix the distance (or time delay) of the 7 virtual sources relative to the one true source. So a sound wave from the real speaker is followed by 7 reflections. At lowest frequencies all sum in phase and at some higher frequency they are likely to cancel and create the "Allison dip". This can be easily modeled and a complete model will include boundary absorbtion, if significant, and directional properties of the speaker.

I don't see any distinction between a floor bounce and a rear wall bounce. The frequency of the floor bounce will vary with position, but the rear wall bounce will vary also. The speaker and its rear reflection may be more in line and geometrically you may see less variation (in null location) for small listener movements but that isn't a fundamental difference. If you moved around the speaker to the side wall then the speaker and rear reflection would come into time alignment and the dip would disappear.

The power response effect of the boundary is simply an area weighted energy (pressure squared) integration around a sphere (hemisphere, quarter sphere or eighth sphere, depending on number of boundaries modeled).

While we are on the subject, my one problem with the Allison papers is that they hide the effect of room standing waves. I did some duplicate measurements in a room a month or so ago and the boundary dips are quite noticeable, and per the paper, when the speaker is close to the back wall or side wall, say within 2 feet. As you move farther out then the boundary effects are quickly obscured by room standing waves and boundary reflections are lost "in the noise". The original Allison studies averaged response vs relative boundary position in a number (30 to 40?) Boston area homes. The act of averaging means that the standing wave effects of each room, different by virtue of different room dimensions, would disappear from the curves and only the boundary effects would remain. This gives the impression that the rear boundary effects are overiding when they might become fairly insignificant with speakers spaced, say 4-5 ft from walls.

For average consumers that place their speakers on a bookshelf, on the floor or in the corner (apparently quite prevelant in the Boston area study) then the Allison dip is a real issue. But if you follow the audiophile notion of "on stands and away from the boundaries", then the Allison dip is hard to find. Floor bounces seem to be always there, though. Bech and others found that they were likely to be audibly more significant, primarily because the timing was right (early enough) and also because we have poor directional acuity in the vertical direction. I.e. we can't seperate them from the original event like we can with side wall reflections. This is why I keep coming back to omni mic room averaged curves as being a not-so-good indicator of perceived frequency response!

Regards,

David

Link to comment
Share on other sites

  • Replies 240
  • Created
  • Last Reply
Let's do a "True Loudspeaker Love for Ken" poll.... :D

Obviously the reason Ken hasn't arrived at loudspeaker nirvana yet is because he never heard your design...or probably not even Geddes. He's your ultimate market. And you like the same kind of rock music so he'd likely see it or rather hear it your way. Don't you both live in California? Think he ever heard the Paragon? It might be slightly before his time. Who needs e-harmony, this could be a meeting of the minds. You could share eargasms together :rolleyes:

Link to comment
Share on other sites

I am about to go read a book (not about audio) and shut this computer off.

However, I do want to restate that with the "Allison Effect" what is impacted is the power input to the room from the woofer. And this happens when the quarter wavelength reflection gets back to the woofer driver after being reflected by a room boundary and nulls the output right there at the woofer. If the distances to the three nearest boundaries are all reasonably different the null is shallow and broad. However, if the distances to two or three boundaries are all pretty much the same the dip will be sharper and deeper.

"Don't confuse me with facts. My mind is made up."

Sorry, but the effect is all about the real loudspeaker, image loudspeaker reflections, and the relative distances between them from a particular observation point in the room. It is not simply about a 1/4 wavelength spacing between the system and the wall (or floor or side) creating a power response hole that is then observed through out the room.

But don't believe me, lets refer to Mr. Allisons 1974 paper "The Influence of Room Boundaries on Loudspeaker Power Output". On page 315 he describes the image model just as I had described it: (first for the single boundary case) "considering the source and its image beyond the boundary to be a pair of small sources vibrating in phase..." He defines the radiated power as I did: "squaring these pressure values, multiplying by cos theta, and summing the values thus obtained yields the total relative power radiated for the assumed value of x/lambda" (distance to boundary vs. wavelength).

This radiated power curve (Fig. 4) shows a 3dB rise at low frequencies and less than 1dB dip at a frequency of about .35 wavelength. (Not 1/4 wavelength!) Why only 1 dB dip and why not at 1/4 wavelength? This is because the frequency of the dip constantly changes with angle. As you move towards the sides of the speaker the relative distance between the real source and the virtual source contracts and the frequency of cancelation rises. In fact when you swing around to the side (touching the wall) the path lengths are ultimately equal and there is no cancelation dip. This is conveniently used for ground plane measurements where a speaker is placed on or above a hard surface and a microphone is placed in contact with the surface. Aside from a 6dB gain, the boundary has no effect. (See the JBL papers of Gander for more on this.)

Since the frequency of the dip varies with angle and is at the 1/4 wavelength frequency only for the straight forward case, the average (integrated) power effect is quite mild and centers on a higher than 1/4 wavelength frequency.

The reason why the power response isn't universally perturbed at 1/4 wavelength, even though the real speaker would experience the 1/2 wave delayed radiation from the virtual source behind, is that there is virtually no impact on cone motion by the delayed wavefront. This is an inherent side effect of all loudspeakers being relatively low in efficiency. Their own mechanical reactances are much greater than any acoustical loading effects and the delayed sound doesn't have a chance to modify the real speaker's acceleration. So it isn't a case of the speakers radiated power being modified for all points throughout the room.

Note that this is radiated power. Radiated pressure will vary with observation angle and the depth of the null can be quite deep. Indeed it can be near total concellation (an infinite notch) if the rear firing energy is equal to the front radiating component. (And wall reflectivity is 100%, etc.) Also the 3dB rise in power (double power) would be seen as a 6dB rise in pressure, since the power would be doubling while the radiating angle would be halving.

Allison's Fig 8 shows the power effect of 1,2 and 3 boundaries. The low frequency rises are 3, 6 and 9 dB (double those numbers for the pressure gain). I think you were confusing this rise when you refered to 3, 6, and 9 dB notches. The 3 boundary case shows a deeper notch than the 1 or 2 boundary cases. One way to think of this is that when constrained between 3 boundaries (1/8 space) your ability to swing around is constrained, so the possible path variation between the various virtual sources is limited and the dip becomes more severe. note that even with the 3 boundary case the deeper notch is at approx. 0.3 wavelength, not 1/4.

Regards,

David

Link to comment
Share on other sites

If someone handed me a one hundred million dollar project and said you are in charge of building a concert hall for our organization, I'd go to the number one man and ask him to design an exact copy of the number one hall right down to the last detail.

Wow, and how much would you charge the client for the above consulting work?

On the other hand, here's what I would do:

http://www.nytimes.com/2009/07/14/arts/music/14bach.html

-k

"Rock: Putting the 'Hard' back into Hard of Hearing!"

Link to comment
Share on other sites

Wow, and how much would you charge the client for the above consulting work?

On the other hand, here's what I would do:

http://www.nytimes.com/2009/07/14/arts/music/14bach.html

-k

"Rock: Putting the 'Hard' back into Hard of Hearing!"

"Wow, and how much would you charge the client for the above consulting work?"

My usual fee, $100 million.

OK, it's true I have yet to get one client....but after the first one....I intend to retire. :rolleyes:

Link to comment
Share on other sites

I see your point, but I might get with Allison about the details, just to make sure. I will point out that the formula he uses to determine null points is:

1130/d x .3. In this case, "d" id the distance in feet. Note that he did indeed use .3 as the multiplier instead of .25, beause as you indicated not all of the surface of a flat wall area will be the same distance from the woofer cone center. That multiplier is a compromise to account for the fact that room walls are flat.

Howard Ferstler

Not exactly what I indicated, but we'll leave it at that. The baffle dimensions do play a role since the sound from the rear firing image speaker behind the wall would have to get to the baffle edges before "spilling over" towards the front speaker. Perhaps that is an approximation used for the 0.3 factor? (I don't know if Allison's formula is for pressure at a listening point or for room integrated power.)

It would be very interesting if you could ask Allison about the subject and share his reply with the group. When I was first starting out in the business in 1977 I'd sent letters to many of the companies looking for a position. Allison didn't have one for me but he did take the time to send a very encouraging letter which I kept for a number of years.

A gentleman.

David

Link to comment
Share on other sites

Hi Howard,

Thanks so much for sharing that with us. Its most interesting to get Roy's personal views on the subject.

As to the endless debate, well its all in fun and thats why they put us in the kitchen rather than in the living room with the grownups.

Best Regards,

David Smith

Link to comment
Share on other sites

I thought you knew when I tweeked you about your 4435s?

I was too busy exchanging insults with Howard and Soundminded.

Some of my best work appears in that thread, e.g.:

Consider yourself summarily trumped, Howard.

[OH, and close the door on your way out, please; your reverberant field is leaking.... :rolleyes: ]

[This requires considerable concentration, y'know.... :D ]

Link to comment
Share on other sites

Ah, those other conversations away from the "Kitchen" are such dull things. Too bad Roy is retired and his companies are history. I'll bet you he would hire you in a minute if he was still in business. Anyway, he wanted me to post one more message:

Hi Howard,

If you are ever in contact with David Smith again. please tell him that I consider him to be a truly analytical thinker, that I admire his work for Snell, and that I really am grateful for his compliment quoted in your last message.

Peace -- Roy

That's it, and I am pretty sure that is enough.

Howard Ferstler

Thanks Howard, I'm beaming from ear to ear.

David

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.


×
×
  • Create New...