Baffle Diffraction

[scottjoplin]

Or embrace the reflections with dipoles for example

 

[gedlee]

No thanks. I prefer the better imaging.

 

[DBMandrake]

I designed my speakers to have high directivity to avoid early reflections and an absolute minimum of nearby diffraction from the cabinet or nearby objects. And still, the more damping I added to the wall behind the speakers the better the imaging became. I now have about a foot of foam absorption on that wall (and almost none anywhere else.) One can do only so much with the speakers themselves, you must consider how the room interfaces with them as well.

What is the baffle width of your speakers and therefore how low is the baffle step transition frequency ?

Adding foam to the wall behind the speakers will not just affect the first reflection from the speakers, it will have an effect on the overall absorption and decay time in the room, so how would you separate out this effect on the perceived sound from any effect that was due only to the initial reflection from the rear wall ?

I agree foam on the wall behind the speakers would improve the overall sound, but I'd never get that past the boss. In a previous house I had large, thick, folded floor to ceiling curtains on the wall behind the speakers (and partially up the side walls as well, just enough to intersect the early reflections, with the other half of the room more live) and the difference was huge, but it's hard to say how much of that improvement was just from increasing the total absorption in the room when theoretically with wide baffle speakers not much sound above the baffle step frequency is radiated towards the wall behind the speaker.

 

[33Polkhigh]

I still don't understand how FR affects imaging. For example if you used edge diffraction to flatten a FR then I assume you'd lose some imaging.

Early reflections might smear the time difference between direct and indirect sound, so its more like an early echo that occurs. I don't know how FR factors in, but it seems were very sensitive to locating sounds by changes in reverberation.

They didn't have tweeters with them and the woofer foam was shot, so cheap and ideal for this project. The baffle did have some felt on it that surrounded where the tweeter had been. I removed it intact. I then fitted the PHT409 tweeter (small compression driver with waveguide) and measured it.

Those result are very impressive. I might try some rockwool cubes or strips placed on the baffle. Rockwool outperforms any other common material for sound absorption. Hiding it under a grill isn't a bad idea.

 

[33Polkhigh]

What is the baffle width of your speakers and therefore how low is the baffle step transition frequency ?

Adding foam to the wall behind the speakers will not just affect the first reflection from the speakers, it will have an effect on the overall absorption and decay time in the room, so how would you separate out this effect on the perceived sound from any effect that was due only to the initial reflection from the rear wall ?

I agree foam on the wall behind the speakers would improve the overall sound, but I'd never get that past the boss. In a previous house I had large, thick, folded floor to ceiling curtains on the wall behind the speakers (and partially up the side walls as well, just enough to intersect the early reflections, with the other half of the room more live) and the difference was huge, but it's hard to say how much of that improvement was just from increasing the total absorption in the room when theoretically with wide baffle speakers not much sound above the baffle step frequency is radiated towards the wall behind the speaker.

I would think a horn wouldn't need foam behind it. Reflecting more of the sound forward does seem to have more energetic and impactful sound, whereas dispersion gives a more atmospheric sound.

 

[Juhazi]

Princeton University's Acoustic lab has a project of measurring 3D directivity of speakers. GedLee Nathan included!

Most measurements start from 500Hz sadly. Directivity below 500Hz is rather tricky to measure reliably, you must go outdoors and place the speaker high from ground.

Dipole or cardioid radiation are options for a wide baffle, to achieve directivity control below 500Hz. Even a tall vertical line does it only vertically.

3D3A Lab at Princeton University

 

[gedlee]

What is the baffle width of your speakers and therefore how low is the baffle step transition frequency ?

Adding foam to the wall behind the speakers will not just affect the first reflection from the speakers, it will have an effect on the overall absorption and decay time in the room, so how would you separate out this effect on the perceived sound from any effect that was due only to the initial reflection from the rear wall ?

I agree foam on the wall behind the speakers would improve the overall sound, but I'd never get that past the boss. In a previous house I had large, thick, folded floor to ceiling curtains on the wall behind the speakers (and partially up the side walls as well, just enough to intersect the early reflections, with the other half of the room more live) and the difference was huge, but it's hard to say how much of that improvement was just from increasing the total absorption in the room when theoretically with wide baffle speakers not much sound above the baffle step frequency is radiated towards the wall behind the speaker.

I am very leary of the concept of "baffle step transition frequency." I know where it comes from and what it is supposed to mean, but with the way I design speakers it is not a factor. When measured free field in the enclosure and designing a crossover from those measurements the BSC becomes integral with the crossover and not a separate thing. As IMO, it should be.

My speakers basically don;t have a baffle. the drivers are 15" and 18" and the baffle is 18", with roundovers there is basically no "baffle left".

It is true that the heavy absorption wall will affect the entire sound field, but it will dominantly affect the lower midrange reflections for several reasons. First, at LFs my room is highly absorbent from its construction - heavily damped floating walls. (Sorry, I was misleading when I said that there was no other absorption, I was referring to MFs -> HFs.) Hence, the added absorption in this frequency range is not a huge factor. Above about 500 Hz there isn't much rear sound radiation, so this wall is again not much of a factor. Its major effect would be reflections from about 200 - 700 Hz. Not a dominate region for imaging, but still a factor. The overall absorption increase would then be a secondary factor, I agree.

 

[gedlee]

I still don't understand how FR affects imaging. For example if you used edge diffraction to flatten a FR then I assume you'd lose some imaging.

Its is not the changes in FR that are the issue, it is the time delay in the diffractions and the reflections that matter most. Hence with diffraction the FR changes are not huge, but the time delay is highly offensive to imaging.

 

[DBMandrake]

I am very leary of the concept of "baffle step transition frequency." I know where it comes from and what it is supposed to mean, but with the way I design speakers it is not a factor. When measured free field in the enclosure and designing a crossover from those measurements the BSC becomes integral with the crossover and not a separate thing. As IMO, it should be.

I don't give baffle step correction any "special" treatment in the crossover design either.

Both drivers were measured in the final cabinet, using a combination of a gated high frequency measurement taken at about 1 metre using a 4ms gate time, which is the longest I can measure in my currently available measuring location, spliced together with a bass measurement up to about 300Hz, to get as close as I can to a free field measurement. (The overlap between the two measurements is marginal, but its the best I can do at the moment so there is a little uncertainty in the lower midrange region in my measurement)

The rise in response from bass to midrange is then just treated the same as any other response "aberration" in the midbass driver when it comes to designing the crossover to achieve as flat a response as possible. So baffle step correction is automatically "baked into" the crossover design not treated separately. It's just another wiggle in a line to be corrected.

The complication though is that when you rely on splicing a nearfield measurement with a gated measurement, you have to apply "inverted" baffle step correction to the nearfield measurement before splicing it to the gated high frequency response to get a true far field response on which to base the crossover design.

To do that you have to simulate the baffle step response of the speaker, and depending on the model used to do that errors will creep in. Many baffle step simulators show a lot more severe ripples above the transition region than I have ever measured in practice - for whatever reason these higher order effects don't seem to manifest as so many simulators suggest they will.

The only effect of significance I've seen is that above the transition region there will be a modest peak, so on my speakers with 39cm wide baffles and a transition frequency of about 300Hz, I see about a 2dB peak around 500Hz in the raw response as a result of this. However this too is corrected for in the network. The higher order peaks and dips beyond this predicted in sims don't seem to appear, perhaps due to increasing directivity of the driver itself no longer illuminating the baffle edges.

Rather than use a "high resolution" baffle step model that predicts all kinds of horrible peaks and dips I decided to use the simple shelf model recommended by the author of ARTA, and this has been surprisingly accurate.

My speakers basically don;t have a baffle. the drivers are 15" and 18" and the baffle is 18", with roundovers there is basically no "baffle left".

Whether a panel extends significantly beyond the edge of a driver or not doesn't really matter. An 18" woofer on an 18" wide baffle is still an 18" baffle, and behaves as such. When the driver is nearly as wide as the baffle it is effectively "self baffling".

But the baffle step transition frequency will be the same if you put an 18" driver on an 18" wide baffle with the frame coming right to the edge, or an 8" driver in the middle of that 18" baffle.

I'll assume you have maybe an inch either side of the driver frame (?) so lets call your baffle width 20" or 500mm - which is a transition frequency of about 250Hz.

So your speakers will maintain some directivity right down to around 250Hz, whereas a 150mm wide baffle will only maintain some directivity down to 600Hz. Quite a big difference in practice, with the wide baffle being much less sensitive to placement relative to the wall behind it.

I've built and tested speakers with 500mm wide baffles and had excellent results with that width. It's my favoured baffle width for a large 3 way system - wide enough to count as "wide" for baffling purposes and obtaining a low baffle step frequency, but not so wide as to be impractical.

It is true that the heavy absorption wall will affect the entire sound field, but it will dominantly affect the lower midrange reflections for several reasons.
First, at LFs my room is highly absorbent from its construction - heavily damped floating walls. (Sorry, I was misleading when I said that there was no other absorption, I was referring to MFs -> HFs.) Hence, the added absorption in this frequency range is not a huge factor. Above about 500 Hz there isn't much rear sound radiation, so this wall is again not much of a factor. Its major effect would be reflections from about 200 - 700 Hz. Not a dominate region for imaging, but still a factor. The overall absorption increase would then be a secondary factor, I agree.

I think you'll find that you won't get much rearward radiation from your speakers above about 300Hz if they are nearly 20" wide, especially with 18" drivers...

If the rear wall is effective at absorbing 200-700Hz it may be that what you're noticing is not the first reflection from the speaker but forward radiated lower midrange that has bounced off the wall behind the listener then gone back to the wall behind the speaker to be absorbed.

 

[mark100]

The complication though is that when you rely on splicing a nearfield measurement with a gated measurement, you have to apply "inverted" baffle step correction to the nearfield measurement before splicing it to the gated high frequency response to get a true far field response on which to base the crossover design.

Hi, have you been able to compare your gated, corrected, then spliced near-field measurements; against a single full-system far-field measurement?

If you have, how do they compare?
And may i ask what your far-field measurement technique is?

Thx !

 

[33Polkhigh]

Does anyone if baffle dimension ratios matter with respect to better or worse diffraction. So would a 19/10 baffle work better than a 20/10?

 

[gedlee]

I think you'll find that you won't get much rearward radiation from your speakers above about 300Hz if they are nearly 20" wide, especially with 18" drivers...

The woofer is 15" as I said. The waveguide is 18".

If the rear wall is effective at absorbing 200-700Hz it may be that what you're noticing is not the first reflection from the speaker but forward radiated lower midrange that has bounced off the wall behind the listener then gone back to the wall behind the speaker to be absorbed.

Remember that I cant my speakers. Lots of radiation still to the rear wall.

 

[gedlee]

Does anyone if baffle dimension ratios matter with respect to better or worse diffraction. So would a 19/10 baffle work better than a 20/10?

Ratio won't matter until it gets high symmetric, like square, or even worse a circle with driver in the center. Otherwise, only the roundover radius matters.

 

[Juhazi]

Baffle width has other effects besides step. Edge diffractions are more benign in narrow baffles. Softly rounded wide baffle could be optimal, eg. Sonus Faber Stradivari. But I still wonder where that wiggle just above 1000Hz comes from...
https://www.stereophile.com/content/sonus-faber-stradivari-homage-loudspeaker-measurements

 

[DBMandrake]

Baffle width has other effects besides step. Edge diffractions are more benign in narrow baffles.

Based on what ?

As already mentioned by someone else, with a wide baffle by the time the wave travelling along the front baffle reaches the edge it will have travelled further and been attenuated more than on a narrow baffle.

Also it lowers the frequency of the first peak/dip, which is always the worst one - a narrow baffle puts that first peak/dip within the frequency range of the tweeter, with a wide baffle it can be below the operating range of the tweeter!

Don't place too much faith in simulations performed by The Edge. They really aren't very accurate at all and I have given up on using it a long time ago because it always predicts far worse perturbations of the frequency response than actual measurements show! (It doesn't seem to take driver directivity and reduced baffle illumination caused by it into account for example so beyond the first peak/dip real results are much smoother than predicted)

 

[Juhazi]

^Some notes. Yes Edge is just a quick and dirty little simmer, but:
- I was focusing at what happens in midrange, including baffle step and baffle's first edge null.
-At this F range a 150mm driver radiates evenly, it starts narrowing above 3kHz
- The Edge models driver directivity based on flat disc model. The user can set number of driver surface and edge points, accuracy.

There isn't a single ideal baffle dimension and form actually. Music has very wide frequency/wavelength range which makes this topic so challenging.

 

[33Polkhigh]

Baffle width has other effects besides step. Edge diffractions are more benign in narrow baffles. Softly rounded wide baffle could be optimal, eg. Sonus Faber Stradivari. But I still wonder where that wiggle just above 1000Hz comes from...
https://www.stereophile.com/content/sonus-faber-stradivari-homage-loudspeaker-measurements

Baffle diffraction is more of a tweeter issue. Though even full range drivers have off axis lobes that form and so aren't completely directional up high.

A bigger box with more space around a woofer allows more sound absorption and a lower fs, which sounds better to me. But skinny speakers with a subwoofer is good too.

 

[jmproductions]

Diffraction Be Gone pads

An externally hosted image should be here but it was not working when we last tested it.

I thought Jim Goulding's wool pad tweeter solution didn't look too bad. Although I'm sure a lot of people on here would find it unacceptable. He used to cut them to your custom specs. Doesn't seem like he is in business anymore, so nobody out there doing this or any products available?


Thick wool is not the easiest thing to cut accurately. I've been trying to source a punch tool to produce a clean circle but not much luck.

 

[bentoronto]

And still, the more damping I added to the wall behind the speakers the better the imaging became. I now have about a foot of foam absorption on that wall (and almost none anywhere else.) One can do only so much with the speakers themselves, you must consider how the room interfaces with them as well.

First, 78 arm-chair speculative posts and not a single bit of evidence until the nice post 79 (not counting Juhani's data in post 60, who always says smart things). Shame.

For sure, theory and sims always look far more drastic than real-world results prove to be. And when music is the source, often terrible sim'ed defects prove to be unhearable.

Unlike Earl's observation for direct speakers, some panel and dipole users who tried making the front a dead-end found it harmed the sound quality. The room isn't necessarily "the enemy" but it is a crucial influence.

But for sure, rooms matter a whole lot more than tinkering with crossover phase (and maybe cabinet diffraction too). This is doubly important today with sparsely damped furnishings the mode.

B.

 

[cyberstudio]

As already mentioned by someone else, with a wide baffle by the time the wave travelling along the front baffle reaches the edge it will have travelled further and been attenuated more than on a narrow baffle.

I am not so sure - attenuated by what? Air is a poor absorber of sound even at relatively high frequencies. The baffle itself does not absorb sound but reflects it - but here you could apply felt which would help.

The only thing left to "attenuate" is the inverse square law but that is not really attenuating but merely spreading it out over a larger area. The wider baffle now has longer perimeter so we are just trading more diffraction along a shorter edge for less diffraction along a longer edge.

Also it lowers the frequency of the first peak/dip, which is always the worst one - a narrow baffle puts that first peak/dip within the frequency range of the tweeter, with a wide baffle it can be below the operating range of the tweeter!

That is certainly true.