I think that you might be missing the fact that in the near field the waves still propagate in all directions and along the baffle. It is not until some distance away from the source that the directivity begins to take shape.
I'd have to say however, that how much a waveguide reduces the edge diffraction is an unknown. As I have said, measuring it is extremely difficult. I started putting large radii on my cabinets a long time ago and just kept doing it because the results of the system were so good. How much of that is due to precisely what aspects of the design is far too difficult to *****, but based on my experience I would think it a huge risk to take not doing as much as one reasonably can to improve the sound quality.
Initially I made my cabinets with Constrained Layer Damping (CLD) throughout. This was a big PITA. So I did some tests, found that it wasn't making a significant difference and so I deleted the CLD in favor of a few well damped cross braces - much easier to fabricate. The radii on the cabinets however was never that difficult so I never had an inclination to test the effectiveness of them.
Hence, you may be right that with a waveguide it isn't such a big thing, but then the woofers do act up to about 1 kHz. It's the right thing to do, so why not do it? How effective it is is TBD.
I'd have to say however, that how much a waveguide reduces the edge diffraction is an unknown. As I have said, measuring it is extremely difficult. I started putting large radii on my cabinets a long time ago and just kept doing it because the results of the system were so good. How much of that is due to precisely what aspects of the design is far too difficult to *****, but based on my experience I would think it a huge risk to take not doing as much as one reasonably can to improve the sound quality.
Initially I made my cabinets with Constrained Layer Damping (CLD) throughout. This was a big PITA. So I did some tests, found that it wasn't making a significant difference and so I deleted the CLD in favor of a few well damped cross braces - much easier to fabricate. The radii on the cabinets however was never that difficult so I never had an inclination to test the effectiveness of them.
Hence, you may be right that with a waveguide it isn't such a big thing, but then the woofers do act up to about 1 kHz. It's the right thing to do, so why not do it? How effective it is is TBD.
Not actually that difficult, depending on the type of driver. I've done this test before, comparing three tweeters.
One a standard soft dome tweeter with wide dispersion, the second a waveguide loaded soft dome tweeter (basically the same motor and dome but with a waveguide face plate) and finally my ribbon tweeters, which have small waveguides. The first two tweeters had identical diameter face plates, while the ribbon tweeter was slightly bigger, by about 15mm.
Two on axis measurements were taken of each driver where diffraction should be worst - one with the drivers flush mounted on a large baffle as the reference measurement, (and windowing time sufficiently short to exclude diffraction from the edge of this large baffle to approximate an infinite baffle) and a second measurement with the drivers held in mid air from behind with a clamp with no baffle other than that afforded by the face-plate itself. All three tweeters are closed back with no rear radiation.
The two measurements for each driver were then overlaid to see how much the baffled and un-baffled measurements deviated from each other. It stands to reason that the less radiation there is from the driver at 90 degrees which would travel along a baffle to diffract of the edge of the cabinet, the less the un-baffled frequency response would depart from the "infinite" baffle measurement.
The results were just as you'd expect - the standard done tweeter's frequency response was horrifically non-flat without a baffle with huge peaks and dips in the response amounting to over 4-5dB, especially in the 6-10Khz region which corresponds to the diameter of the face plate. Clearly a lot of energy diffracting off the edge of the suspended face-plate.
Both the waveguide loaded dome and ribbon tweeters fared dramatically better. I don't recall which one showed the least deviation as I did this test 15 years ago and the measurements are now lost. But suffice to say that both the waveguide tweeters were minimally impacted.
There was still some change, but if I recall right the ribbon tweeter only showed a broad a 1dB dip from about 8-10Khz as the only significant change in response from sitting in free air without a baffle, until you got down below about 3Khz where the waveguide stopped being effective. In fact I had those ribbon tweeters just sitting on top of a cabinet without a baffle for normal use for quite a while until they eventually went on a baffle as the effect of not being on a baffle was so minimal.
Last edited:
To me what you describe is the definition of "difficult" - it took a lot of work. I think that your results and conclusions make sense: of course a waveguide does prevent some, maybe even most, sound from radiating to a diffracting edge, but there still is some and a radii edge will benefit that.
I would think it wouldn't be that hard to measure, how much of a benefit perceptually is the bigger issue.
One question I have is with waveguides in the crossover region. Does directivity help reduce the off axis issues introduced by a typical crossover?
Also as far as cabinets, you can put a speaker in the window and put some pillows around it and hear the noise the cabinet makes.
I think the key to a musical cabinet is too push the resonances into the midrange with bracing but avoid standing waves or modes.
That is a good question that I have wondered for years, I really don't know for sure, but I would say "probably not" - they will be about the same. And here is why I think that. The nulls occur mostly due to off axis displacement differences between the sources and the resulting time differences that result. This does not change with the directivity, nulls are nulls, zero is zero, all situations have them. But they are most likely at a different angle in the case of the waveguide. Whether they move in or move out, I am not sure, but I know that they are always there and there is not much that I can do about them so an in depth study would not likely raise any new solutions, hence I never did one.
Found this study about diffraction, it's pretty good. It sounds like the larger the curve the lower down in response you get a benefit. But then it doesn't look at off axis effects.
https://www.audioxpress.com/assets/upload/files/203moriyasu2151.pdf
https://www.audioxpress.com/assets/upload/files/203moriyasu2151.pdf
OK, we have writer who is doing his best to flog his new speakers which are distinctive looking due to curved edges. And the worst he can demonstrate is plus and minus an unhearable 2.5 dB, that would be totally lost in the reality of any room, manufacturing tolerances, throw-rug slightly off-centre on your floor, picture of your grandmother on the left wall but not the right wall, etc.
Yes, to the eyeball it looks dramatic as FR plots. But just how does your ear say to you, "Gosh, I was sure that singer was 1 dB louder at 1,257 Hz in the complex note she is singing, I am sure of that".
Do you possess absolute judgment and perfect memory of pitch, of loudness, and what the source sounds like in your room if she were singing there?
B.
Last edited:
I have looked at that article before.
Too bad the graphs are almost impossible to read
I still think diffraction has a big effect on sound. But it needs to be considered as part of the whole along with cone break up and baffle step.
Using a complicated scientific approach (putting wool socks on speaker) I can hear big differences and resonances.
I have to repeat what I said before that certain areas like 2-4K are more important than others and sometimes positioning a driver to create a bump or dip in this area can be a good idea.
In particular if you use a baffle that is about 4x the width of the cone you can get a target diffraction notch that sits in the spot of cone breakup. So diffraction can be used to offset cone breakup.
Then a roundover or absorption can be used to deal with the upper frequencies. You want the initial baffle step peak to come before cone breakup.
I think this wide baffle technique is the best, and allowing cone break up to coincide with a diffraction bump in a serious recipe for a torture machine speaker.
Using a complicated scientific approach (putting wool socks on speaker) I can hear big differences and resonances.
I have to repeat what I said before that certain areas like 2-4K are more important than others and sometimes positioning a driver to create a bump or dip in this area can be a good idea.
In particular if you use a baffle that is about 4x the width of the cone you can get a target diffraction notch that sits in the spot of cone breakup. So diffraction can be used to offset cone breakup.
Then a roundover or absorption can be used to deal with the upper frequencies. You want the initial baffle step peak to come before cone breakup.
I think this wide baffle technique is the best, and allowing cone break up to coincide with a diffraction bump in a serious recipe for a torture machine speaker.
Last edited:
For example, if you use 5.5in driver, do you mean the width of the baffle should be 5.5 x 4 = 22in. That is quite wide.
For 8in, it would be 8 x 4 = 32in.
Unless I am misreading your comments?
Yes it is quite wide. But the issue is the cone breakup and where it falls. So a typical 3 way would have maybe a 10" baffle with an 8" woofer and a 2.5" midrange. Cross to a 3/4" dome around 3000 to 4000 and a simple roundover will take care of dome diffraction.
But the point is to complete baffle step before cone break up and create a diffraction notch. A speaker that is becoming directional around the point of cone shriek (assuming there is cone shriek) is IMO flawed even though this is common practice.
Last edited:
For example, for a typical 3in. mid, the break up should be well above 7K or more (unless there's something wrong). If you have a baffle around 8in wide, the diffraction is about from 4KHz - 5KHz or around there(of course the narrower the baffle, the higher the diffraction freq). My point is usually the cone break up frequency is usually higher than the diffraction frequency.
If you have a 6in driver then the break up freq may be lower, but the baffle should be a bit wider than 8in which will result in lower diffraction freq.
But one thing I would agree is diffraction is one tough nut to crack, no pun intended. One way to minimize diffraction is to use higher order filter which will filter out the driver response before it runs into diffraction. But higher filters have their own issues.
How on earth would one be able to do that... diffraction isn't something that only happens at a single frequency. Plus it will change when you move due to the changing pathlengths. Look up the Olson graphs. Even though they were just hand crafted lines on paper, they should tell you what diffraction does (or what it looks like) and give you pointers how to reduce it's effect.
Who said diffraction only happens at one frequency? Actually I don't know what the hell you're talking about. Are you arguing with me or something in your own head? By the way, the guy in your sig. looks like a freak.
Last edited:
One thing I should point out is that 4x the baffle width is just an estimate that seems to work. Also this works better with an mtm because it removes some of the top and bottom diffraction.
People have been putting bbc dips into speakers for years but offsetting the baffle and cone resonances is an elegant solution, especially since diffraction is harder to deal with down low,
Break up typically starts around when the frequency is about 4x the cone width. A 3in diameter cone will be beaming pretty good by 7k.
Cone shriek is often earlier in break up because as the cone beams it loses total energy. Alot of this is about adjusting the power response.
Here is woofer that could use some diffraction adjustment. It has a noticeable peak around 1500 hz which is not uncommon for woofers of this size.
https://www.parts-express.com/pedocs/specs/297-435-hi-vi-f5-specifications-44671.pdf
People have been putting bbc dips into speakers for years but offsetting the baffle and cone resonances is an elegant solution, especially since diffraction is harder to deal with down low,
Break up typically starts around when the frequency is about 4x the cone width. A 3in diameter cone will be beaming pretty good by 7k.
Cone shriek is often earlier in break up because as the cone beams it loses total energy. Alot of this is about adjusting the power response.
Here is woofer that could use some diffraction adjustment. It has a noticeable peak around 1500 hz which is not uncommon for woofers of this size.
https://www.parts-express.com/pedocs/specs/297-435-hi-vi-f5-specifications-44671.pdf
Last edited:
Diffraction is ALSO a multi-source "model" so it's really a bunch of discreet values (frequencies) from a huge array of sources - from radiator to any coupled/bounding change in angle.
The net result is usually a pass-band of effect, with some amount of combing from the diffraction and direct sound generating a resulting ripple above this pass-band. (..though you can "average" this out depending on the graphical display of the Impulse response.)
FREAK.. what the he!!
-that's Diamond DAVE of the original Van Halen!
YouTube
-ok, maybe he is a little "freaky" looking.
(..though that "freak" managed to "accommodate" a great deal of "tail" in his prime.)
Last edited:
-that can work well, particularly with driver off-set on the baffle.
See DLR's work below (..from quite some time ago):
David Ralph's Speaker Pages - Felt Effects on Baffle Diffraction
-btw, I like neither off-set drivers nor felt that covers large areas of the baffle.
The felt tends to subjectively "slow" the higher freq. response with respect to perceived very short-term dynamic changes. Cleaner? Yes, but at some of the expensive of perceived "speed" and a little bit of "air" (which in most designs like this is also a diffractive effect).- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.