A few questions:
What is an optimal listening window, flat with on axis, or a slight decrease down to say -2dB at 20kHz?
What methods can be used to widen dispersion from 10-20kHz, and or to flatten the horizontal listening window (0-30°) and vertical listening window (0-10°). And can these methods apply to tweeter waveguides, although any other methods would also be interesting.
What is an optimal listening window, flat with on axis, or a slight decrease down to say -2dB at 20kHz?
What methods can be used to widen dispersion from 10-20kHz, and or to flatten the horizontal listening window (0-30°) and vertical listening window (0-10°). And can these methods apply to tweeter waveguides, although any other methods would also be interesting.
An acoustic lens can accomplish this. Not sure how well it works on a direct radiator vs a horn if thats your application.
Similarly, a dispersion slot can also widen the angle. I remember Tannoy used this on their older speakers. With HF applications its a bit more tricky to implement, as the wavelength is very short in the HF region.
Similarly, a dispersion slot can also widen the angle. I remember Tannoy used this on their older speakers. With HF applications its a bit more tricky to implement, as the wavelength is very short in the HF region.
I wonder if anyone has tried to 3d print some DIY acoustic lenses, so far I haven't seen anyone try to simulate it in ABEC/AKABAK.
Waveguide mainly improves the lower frequency
not higher frequency band of a tweeter.
To help with crossover design. Or improve distortion if the system
requires the tweeter to be crossed lower.
Waveguide being a larger area will reduce high frequency off axis.
The only way to improve high frequency above 12 kHz off axis.
Is to use a smaller driver, or smaller radiating area.
All simplicity when the frequency wavelength is smaller than
the driver SD, you get beaming. no way around it.
Most typical domes are 1" or around 25mm = SD is around 6 to 7 cm2
Going down to a smaller 3/4" or around 19mm = SD is around 3 to 4 cm2
Smaller tweeter will help push rolloff off axis from 10 to 12 kHz to maybe 13 to 15 kHz
Far as I remember 10 Khz is around = 3.4 cm2 wavelength
So if tweeter SD is 6 or 3 cm2
off axis high frequency is hard to do regardless.
The sound waves above 10 kHz will always be smaller than a tweeter
no way around it, unless you make a very very small tweeter.
Far as real world solutions. A very small tweeter wants 4 to 5 kHz crossover
So you need a smaller tweeter and you need a smaller midrange
To keep crossover at 4 to 5 kHz not typical 2 or 3 Khz
Reason being a tweeter more likely to hit distortion at lower frequency.
if crossed to low. it is done often of course, so distortion prevails
Hence " waveguide" if you improve efficiency down low.
Instead of using a 2k crossover which risks distortion. to sum flat.
You can still use the 3k crossover but everything sums flat
without risk or lower risk of distortion.
Unfortunately the tradeoff for a larger area from the waveguide
is beaming at high frequency just like any horn / waveguide
not higher frequency band of a tweeter.
To help with crossover design. Or improve distortion if the system
requires the tweeter to be crossed lower.
Waveguide being a larger area will reduce high frequency off axis.
The only way to improve high frequency above 12 kHz off axis.
Is to use a smaller driver, or smaller radiating area.
All simplicity when the frequency wavelength is smaller than
the driver SD, you get beaming. no way around it.
Most typical domes are 1" or around 25mm = SD is around 6 to 7 cm2
Going down to a smaller 3/4" or around 19mm = SD is around 3 to 4 cm2
Smaller tweeter will help push rolloff off axis from 10 to 12 kHz to maybe 13 to 15 kHz
Far as I remember 10 Khz is around = 3.4 cm2 wavelength
So if tweeter SD is 6 or 3 cm2
off axis high frequency is hard to do regardless.
The sound waves above 10 kHz will always be smaller than a tweeter
no way around it, unless you make a very very small tweeter.
Far as real world solutions. A very small tweeter wants 4 to 5 kHz crossover
So you need a smaller tweeter and you need a smaller midrange
To keep crossover at 4 to 5 kHz not typical 2 or 3 Khz
Reason being a tweeter more likely to hit distortion at lower frequency.
if crossed to low. it is done often of course, so distortion prevails
Hence " waveguide" if you improve efficiency down low.
Instead of using a 2k crossover which risks distortion. to sum flat.
You can still use the 3k crossover but everything sums flat
without risk or lower risk of distortion.
Unfortunately the tradeoff for a larger area from the waveguide
is beaming at high frequency just like any horn / waveguide
Last edited:
Unlikely since SD is graciously larger
many drivers go well above 20 kHz 40 kHz
off axis same old 10k
10 kHz wavelength is about 3mm
So until you make a tweeter smaller than 3mm
Off axis you will never beat.
It is ok cause the classic argument is men dont hear much above 8 kHz to 12 kHz
sometimes as low as 6 kHz being realistic. Depends
Then again even men that admit or aware of hearing losses.
A common HiFi system that goes above 12 Khz
Still has more clarity and more detail even with hearing losses
More important is quality of the tweeter. Not the pretty graph.
I have "seen" planars or ribbons with not so great mega perfect graph
But in reality have actually " heard" nice sound and clean on attack transients.
many drivers go well above 20 kHz 40 kHz
off axis same old 10k
10 kHz wavelength is about 3mm
So until you make a tweeter smaller than 3mm
Off axis you will never beat.
It is ok cause the classic argument is men dont hear much above 8 kHz to 12 kHz
sometimes as low as 6 kHz being realistic. Depends
Then again even men that admit or aware of hearing losses.
A common HiFi system that goes above 12 Khz
Still has more clarity and more detail even with hearing losses
More important is quality of the tweeter. Not the pretty graph.
I have "seen" planars or ribbons with not so great mega perfect graph
But in reality have actually " heard" nice sound and clean on attack transients.
Its 344/10000 = 0.0344 m = 34.4mm
17.2mm for 20kHz, one thing I have attempted to do is to cover part of the tweeter surround with the waveguide to obtain a smaller initial apeture, its marginal but may help and let's me get the waveguide contour closer to the tweeter dome. The SB19 tweeter has a small enough dome that it has minimal off axis funny business below 20kHz.
I have also seen an image somewhere on the forum about optimal dome shapes, and degrees of convex/concave-ness on the dispersion, if anyone has the image/s it would be much appreciated.
Do you think it would be possible to create a profile similar to OS, that has an optimal radius of curvature for sound diffraction? Say the radius of curvature is inversely proportional to the distance down the profile of the horn, so that you effectively have a diffraction slot immediately after the tweeter dome, and at somepoint along the profile you transition into a spiral, to get the right termination into free-space like alot of the ath-waveguides?
Let me address these points separately..
1. I'm not sure why you mention a diffraction slot because the OS profile doesn't need help to widen the top end. It is often incorrectly assumed that a waveguide can only contain and limit radiation.
2. The OS profile does take a plane wave out to spherical, as it needs to be, with less diffraction than any other regular profile.
3. You can begin with OS then switch to a termination because OS preserves the shape until you're ready to do so. You could also integrate the termination like the ath-waveguides but you'd need a larger device for the same control because they trade low end directivity against the potential for reflection.. something you could take upon yourself to handle if you choose.
1. I'm not sure why you mention a diffraction slot because the OS profile doesn't need help to widen the top end. It is often incorrectly assumed that a waveguide can only contain and limit radiation.
2. The OS profile does take a plane wave out to spherical, as it needs to be, with less diffraction than any other regular profile.
3. You can begin with OS then switch to a termination because OS preserves the shape until you're ready to do so. You could also integrate the termination like the ath-waveguides but you'd need a larger device for the same control because they trade low end directivity against the potential for reflection.. something you could take upon yourself to handle if you choose.
looks about right, for 3/4 inch tweet.
as mentioned you get about up to 15 kHz instead of 12 kHz.
there is a few dual ring 3/4" which might help.
far as I remember most the dual rings can trade off
little more distortion oddities here and there from the
typical nipple shapes used.
Waveguides are very much fun.
But it is usually going to lead to a larger area.
hard to simulate a splay baffle, would have to do in experiments.
eventually to fix " off" axis it would require being " on" axis
to another device.
Using dual offset tweeters is a rabbit hole, unfortunately
as mentioned you get about up to 15 kHz instead of 12 kHz.
there is a few dual ring 3/4" which might help.
far as I remember most the dual rings can trade off
little more distortion oddities here and there from the
typical nipple shapes used.
Waveguides are very much fun.
But it is usually going to lead to a larger area.
hard to simulate a splay baffle, would have to do in experiments.
eventually to fix " off" axis it would require being " on" axis
to another device.
Using dual offset tweeters is a rabbit hole, unfortunately
Last edited:
In my case I'm not looking for CD but rather a sloped DI to give roughly a predicted in room with a -1dB/Oct slope down to -8dB from on axis at 20kHz, harmon curve style.
Ahhh nice, lol darn decimal point= I thought it was 3.4 as opposed to 34
either way tweeter has to be smaller.
Unfortunately there is plenty of " car stereo" tweeters which are very small.
yes I know horrible word.
They are very small ,usually Mylar, which in theory should work
size wise.
But they are just average ( horrible) 3 way Mylar domes
Surprised nothing HiFi has been done.
The concept is there. Sensitivity likely the issue
The waveguide iteration I have so far (for SB26ADC, I know augerpro WGs exist, but I wanted to try the process for myself and make it specific to my box design) has a small radius in the throat, I will try a version with the actual OS equation.
I'm trying to widen up the very top octave to flatten out listening window, and push up the power response a dB or 2, at this point I'm very close to having something final, just thinking about any other possibilities that could be put on the table.
Horizontal:
Vertical:
Power/DI:
Ignore below 1kHz.
I have a suspicion that the slight peaking off axis at 20kHz, thats naturally present in the SB26 might basically fill in that 15kHz and up bit but I really need to do a real measurement in the waveguide.
I'm trying to widen up the very top octave to flatten out listening window, and push up the power response a dB or 2, at this point I'm very close to having something final, just thinking about any other possibilities that could be put on the table.
Horizontal:
Vertical:
Power/DI:
Ignore below 1kHz.
I have a suspicion that the slight peaking off axis at 20kHz, thats naturally present in the SB26 might basically fill in that 15kHz and up bit but I really need to do a real measurement in the waveguide.
Maybe try waveguide for a 13mm instead of 19mm
since it reduces the starting point.
Far as I know Tang Band makes a few silk domes
That are 13mm voice coils.
TB datasheets usually kinda suck.
But using basic diffraction math im sure
you could push things up to closer to 18 kHz
as opposed to 15 kHz
I guess if you really wanna hear Class D comparator noise
or compressor hiss in a recording off axis.
during silence
Otherwise I think 3/4" more than fine.
other than bragging rights or very fun challenge.
since it reduces the starting point.
Far as I know Tang Band makes a few silk domes
That are 13mm voice coils.
TB datasheets usually kinda suck.
But using basic diffraction math im sure
you could push things up to closer to 18 kHz
as opposed to 15 kHz
I guess if you really wanna hear Class D comparator noise
or compressor hiss in a recording off axis.
during silence
Otherwise I think 3/4" more than fine.
other than bragging rights or very fun challenge.
You can achieve that in room response with CD as well, so while it’s your choice and it isn’t wrong, I don’t understand your assumption that that way is needed to achieve it. As it turns out I would expect to equalise any waveguide every time so it is what it is.
I have been using this little cheap planar above 7kHz
https://www.diyaudio.com/community/...bbons-what-quality.113262/page-2#post-6738896
https://www.diyaudio.com/community/...bbons-what-quality.113262/page-2#post-6738896
Wouldn't this require sloping of the on-axis response to compensate for the power response being roughly flat from CD, the power response in CD will decrease a little bit, as less and less sound power diffracts round the edges, but the majority of the power remains in the designed beam width.
I'm aiming for flat on-axis, if what I said above isn't true then it could make WG design more interesting in terms of designing soundstage width.