Sure, felt would work well to reduce the diffraction. I might still be tempted to use a radius on the edge however. But even though it does reduce the diffraction, the felt still does not make the tweeter into a constant directivity device. I would be interested in how the felt affects the polar response of a direct radiator.
And of course many of us will remember foam and felt being used at the mouth edges of some older horns to reduce the diffraction from them.
Wool Felt Sheet Specifications, SAE Table:
SAE felt specification chart from Sutherland Felt Company - Detroit area, Michigan
Wool Felt Sheet Sources:
Materials Converter | Slitting | Die Cutting | Waterjet Cutting | Laminating
Welcome To Boston Felt Online
General recommendation for acoustic applications is F-11 or F-13 SAE felt. Use a bare minimum of 1/4" thickness, preferably 3/8" or greater. 1/2" is better. Acoustical foam is not very effective below 3000 Hz for reducing edge diffraction. Felt absorbs the energy from the impinging wave.
Dennis Murphy post on diffraction changes with 3/4" baffle edge round:
MurphyBlaster Productions
SAE felt specification chart from Sutherland Felt Company - Detroit area, Michigan
Wool Felt Sheet Sources:
Materials Converter | Slitting | Die Cutting | Waterjet Cutting | Laminating
Welcome To Boston Felt Online
General recommendation for acoustic applications is F-11 or F-13 SAE felt. Use a bare minimum of 1/4" thickness, preferably 3/8" or greater. 1/2" is better. Acoustical foam is not very effective below 3000 Hz for reducing edge diffraction. Felt absorbs the energy from the impinging wave.
Dennis Murphy post on diffraction changes with 3/4" baffle edge round:
MurphyBlaster Productions
OK ... it is what it is but it would be nice to know what it was before you start chopping the box 😉No I never measure...I know, I did this in a wrong way...I should first measure then build...but then I would not build anything 😀 ... Now I have build, I like the results...think it have potentials 😀
The drivers are 2904/7000 , 18W8545, 21W8555...
The 2904/7000 have some dip in that area, but it's like -1db and I have like -5db...As I don't have experience, I don't know how bad it is...I see some graphs from other loudspeakers and most of them have those dips...
Well I have to put my PC back together as without it's hard to make any progress 😉
This is dinemsions of front buffle, for 2904/7000 and 18W8545...
Maybe after you have your ducks all lined up (individual responses measured) you could start a thread about it and work from there.
One method to determine if a particular edge(s) is giving an anomally in the frequency response involves extenting the effective baffle edge(s) using cardboard and masking tape. Re-run the frequency response test and compare against previous readings. You should see a shift in the frequency of the anomally center if diffraction is an issue.
I just got back to the thread, but I see Remlab has responded to the query about my testing and articles.No I never measure...I know, I did this in a wrong way...I should first measure then build...but then I would not build anything 😀 ... Now I have build, I like the results...think it have potentials 😀
The drivers are 2904/7000 , 18W8545, 21W8555...
The 2904/7000 have some dip in that area, but it's like -1db and I have like -5db...As I don't have experience, I don't know how bad it is...I see some graphs from other loudspeakers and most of them have those dips...
Well I have to put my PC back together as without it's hard to make any progress 😉
This is dinemsions of front buffle, for 2904/7000 and 18W8545...
To Dr. Geddes about polar response, one of the primary benefits of felt, even more than modest roundovers, is the overall improvement of polar response. A partial benefit for those who insist on using single axis measurements for design is that the on-axis is greatly improved such that the power response is improved because the on-axis does not have serious non-linearity in the on-axis SPL that causes some DIYers to try to correct, thus screwing up the off-axis and power response.
Felt is most effective (for tweeters) when used with offset, the distributive effect (discussed by Charlie). Because the wave must pass through the felt, at lower frequencies it becomes less effective, due to the varying effectiveness with frequency and because the baffle dimensions often place the peak diffraction effects a bit lower in frequency. But within its controlled range, felt is very effective in polar response irregularity control.
Only mentioned briefly is the proximity of a midrange or midwoofer. The first felt piece I use, always, is a think piece between tweeter and adjacent driver(s). The cavity of a driver is often far more of an issue than an edge, depending on the specifics, but another driver is always bad.
I have not posted anything on the many tests I've made off-axis, but over time I found a consistency such that I can generally judge the off-axis from seeing the on-axis for most situations. I did, though, post some tests I made in using the Soundeasy Ultimate Equalizer for the crosover. This has a 1" Seas DXT tweeter, centered horizontally near the top on a modest baffle (8" x 14"). I was surprised at the diffraction. I had to use a fair bit of felt on it, though I've found that a full face application is often not necessary. Triangular placements of 1" wide strips often do nearly as well. Each combination of driver and baffle is unique.
Below is a normalized horizontal polar response set. Ignore below 800 and 20K+, those are measurement artifacts (windowing and 24K upper limit of measurement system). It has Scan-Speak 15W and Seas DXT drivers crossed at LR8 @1200.
The first 30 degrees has limited variance and it's reasonably controlled to 60 degrees.
But that also effects baffle step and can cause humps or dips depending on the crossover design. The worst hump is the first zero in the network.
That's true, with baffle step effects being more gradual and wider, as it is basically a 2 dB/octave impact.
I was looking at ATC site today, and almost all of their products seem to ignore the baffle diffraction, except their discontinued cheap models that had round edge. Their pro models have tiny round edge,. It seems like their decision is not really based on acoustic theory.
http://www.atcloudspeakers.co.uk/hi-fi/loudspeakers/discontinued-loudspeakers/scm11/
http://www.atcloudspeakers.co.uk/hi-fi/loudspeakers/discontinued-loudspeakers/scm11/
Other respected company, Genelec moved to round edge baffle about 10 years ago, but they lost popularity since then. Many golden ear professional engineers still prefer their obsolete sharp edge Genelecs and Yamaha for some reason. One of the most popular studio monitor today, Barefoot MM27, has round edge, though.
I feel that there must be some kind of "strength" with the the sharp edged baffles, (or weakness with round ones)...
I feel that there must be some kind of "strength" with the the sharp edged baffles, (or weakness with round ones)...
Last edited:
Preference and "correct" are not one and the same. McDonalds is not the best burger but it sure is the most popular!
It's popular because it's cheap and a lot of calories. But just like a roundover, if you grind the beef fresh from good cuts of meat, it's more money, tools, and work, but you get a better product.
It's popular because it's cheap and a lot of calories. But just like a roundover, if you grind the beef fresh from good cuts of meat, it's more money, tools, and work, but you get a better product.
I don't think popularity among respected professional engineers and among audiophiles means the same, but yes, McDonalds must be popular among both parties!
But that's the analog- it's not that it's the best, it's just that it's the choice. If a better burger were available for the same cost, of course everyone would get it. But speaker designs are all constrained by cost or we'd have space-age enclosure tech and silver shorting sleeves on all of our high grade neo underhung motors with permendur.
The engineers don't have those either- it's not that they wouldn't be an upgrade, it's just that it would be costly and so the product selection, both by the speaker manufacturer and the studio, is more value oriented.
For us as DIYers, labor is free, so why not do the easy-win things like serious roundovers or bevels?
Good analogy, but perhaps the reasoning is not so sound. McDonalds is far from cheap (not in Australia anyway), and I can't see anyone buying it to deliberately consume more calories... but I can see them eating it because it's quick and easy (or put simply they are lazy).
I can also see them eating it because it is an alternative to the everyday (unless McDonalds is your everyday) and it may be just a styling change for the sake of it. But totally agree different / popular doesn't necessarily equate to better, it could simply be cost or laziness.
Last edited:
Agreed.
Pros are not necessarily any better judges than audiophiles if they are relying on subjective auditions. Measurements will clearly show that sharp edges are a problem, but then most just discount those and rely on their own preferences.
Sharp edges give some "color" to the sound and differentiates it in the marketplace.
Hi Earl, I have learned a lot from your posts about waveguide.
What I don't really understand (and I want to know) is why so many respected companies like ATC, Westlake, Harbeth, Focal, KEF etc still ignore the baffle edge diffraction. I think it's really easy to fix for them, if they really want to. Any manufacturing difficulty or any special marketing reasons?
Also I wonder what kind of psychological reasons behind, if some people simply prefers the edge diffraction. (coloration). This is probably the same question as OP did at first.
What I don't really understand (and I want to know) is why so many respected companies like ATC, Westlake, Harbeth, Focal, KEF etc still ignore the baffle edge diffraction. I think it's really easy to fix for them, if they really want to. Any manufacturing difficulty or any special marketing reasons?
Also I wonder what kind of psychological reasons behind, if some people simply prefers the edge diffraction. (coloration). This is probably the same question as OP did at first.
Last edited:
Earl, one more question (thank you in advance).
Do you think attaching round edged side panels to existing sharp edge speakers is an effective solution to improve the sound? If so how thick the board should be?
Do you think attaching round edged side panels to existing sharp edge speakers is an effective solution to improve the sound? If so how thick the board should be?
Narrow speakers often accept sharp edges, both for style and for minimal width.
One thread discussed a narrow 3-way speaker using side-side counter-force 8" woofers. A simple simulation was done which illustrated significant SPL smoothing from using a 1.5" edge radius.
Naturally, the box width had to increase to accomodate a 1.5" edge radius. The proposal was to use this width to provide soundstage advantage by mounting a 6.5" midbass like the Satori MW16P which covered a wider bandwidth with a Xover under 100Hz, instead of the 5" midrange used in a popular product.
One thread discussed a narrow 3-way speaker using side-side counter-force 8" woofers. A simple simulation was done which illustrated significant SPL smoothing from using a 1.5" edge radius.
Naturally, the box width had to increase to accomodate a 1.5" edge radius. The proposal was to use this width to provide soundstage advantage by mounting a 6.5" midbass like the Satori MW16P which covered a wider bandwidth with a Xover under 100Hz, instead of the 5" midrange used in a popular product.
Attachments
I can't speak for the others but certainly in Harbeth's case, it would be impractical. One of their selling points is the use of thin-wall cabinets (12mm or 9mm).
I haven't checked the whole thread yet to see if this has been referenced, so apologies if it has. This is Grimm Audio's design process which includes wide radius edges - basically the depth of the cabinet
http://www.grimmaudio.com/site/assets/files/1088/speakers.pdf
The curved edges are added after the basic box is built and form the stand and enclosures for the electronics.
Last edited:
- Status
- Not open for further replies.