That's the whole deal with asymmetrical, non-phase matching xo's, to modify lobing and vertical beam shape for an optimized power response. Kimmo's by now famous 1.2-lambda rule is all about it.
Naaah, I know but wanna keep my tight driver spacing instead, to have a more point source.... 😛
It can be phase mached too, but that does require some luck or planning
https://www.diyaudio.com/community/threads/vituixcad.307910/post-6529910
Your individual driver measurements are interesting.
I’m a bit surprised by the directivity of the SB15. I thought it would be usable up to at least 3KHz.
It looks like 2.5Khz may be the upper limit. And on my 6” screen it looks like the RS270-4 is good to almost 1Khz.
A 10” way with a Waveguided tweeter? 😎
I’m a bit surprised by the directivity of the SB15. I thought it would be usable up to at least 3KHz.
It looks like 2.5Khz may be the upper limit. And on my 6” screen it looks like the RS270-4 is good to almost 1Khz.
A 10” way with a Waveguided tweeter? 😎
Both. A DSP can also help:
With wider mid/tweeter diver spacing following Kimmo's proposal, the power drop around mid/tweeter xover frequency might had been reduced:
(from +/-90° hor + ver FRD data @1,5m dist.)
But then I get sound origins more away from each other, and I hear midfield at ~2-3m distance, and therefore wanna keep the sound origins close to coming from a single direction. I trade the smoother power response in sum off for that feature. In addition, with wide driver spacing, you get a more comb-filtering like vertical lobe pattern with some lobes - whereas the close spacing, together with a LR4, gives you one good main lobe to your face. Check with Tolvan XDir. I'd take that one lobe instead of the comb. But this is the good thing when you have options and models for that measurements, and make your choices.
Best regards
Peter
Last edited:
I think it is the result of the baffle, there is bunching at 2kHz, which means things are usually worse at higher frequencies.
tktran303:
I’m a bit surprised by the directivity of the SB15. I thought it would be usable up to at least 3KHz.
tktran303:
I’m a bit surprised by the directivity of the SB15. I thought it would be usable up to at least 3KHz.
Or right click the impedance graph in Vituix and change it to vertical directivity 🙂
I haven't made enough different verions of these sort of speakers to have an opinion on how they sound. Jim has most of the options under his belt now though 😉
Hi,
thanks mentioning listening distance for context! I think that this kind of stuff depends also on room acoustics though, and that it would be more relevant to talk about in relation to shift in perception rather than with absolute distance.
In my place with my speakers sitting at sofa, about 3m away, there is so much early reflections the sound is kind of hazy perceptually. If I move closer to speakers, there is very obvious shift in perception where the hazy sound moves away and clarity happens, which I think is due to getting over some threshold in hearing system, brain is able to make sense of the direct sound over all the early reflections I have here with my system. This is the transition I mentioned, and I speculate this transition distance depends on speakers, positioning and room acoustics. If the transition is something that exists in all systems it would be nice to talk with that as context, as it abstracts speakers, room and perception to one relatively easily detectable audible phenomenon. Absolute number of listening distance could mean either side of the transition for all of us, and thus hard to relate as both sides are very different sounding and would benefit different kind of compromises in my opinion. Knowing which side of the transition one is referring to would help everyone to relate their own situation to yours perceptually.
You've provided nice example I'll use to demonstrate what I mean:
I doubt I would hear any difference in c-c at 3m in my place as it's all single blob of sound, and it seems lobes are inaudible basically. Direct sound seems to be perceived together with early reflections as one so that the sound is hazy in general, full of dips and peaks measuring with a mic. On the other hand when I move little bit closer over the transition, at 2m with my system, I'm positive c-c would matter more in the sense the lobes (nulls) could be heard more readily(*).
So, takeaway is that listening beyond the transition like most people seem to do, it would make sense to have nice power response and not worry about lobe nulls too much which would make 1.2x c-c nice trick to get better sound as it helps smooth DI, despite fear of lobes who the room sound helps to disguise. On the other hand, if listening closer than the transition brain would pay attention to the direct sound and somewhat suppress the reflections, so here it would make sense to optimize for the direct sound, for wide main lobe if you must vary listening height for example. By the way, smooth DI would allow optimize for both sides of transition, as it is relationship between power and direct sound. I also think that room acoustics and positioning adjust how far the transition happens.
For these reasons to me it would be more meaningful to talk about which side of the transition you listen, because that would incorporate all features of your system to perception and it would be more easier to me to relate to your post. To be more literate, what you say would make sense to me if you listen closer than the transition, and not make sense if you listen beyond the transition, and not knowing which side you listen makes it irrelevant information to me.
So, it would be very interesting if you detect the transition with your system and are willing to post about it, and whether your statement holds true on the otherside and not the other. Would be very interesting data point to discuss about, as this kind of stuff would help communication through forum a lot if I'm right track with this.
* I tried Harman How to listen listening test recently and found out EQ dips are the hardest excersise for me, one of the least audible phenomena in the test. Even though dips on a frequency plot, like lobe nulls, are alarming to eye on a polar map they aren't likely very noticeable when listening, at least that's what my listening test results tell me. In fact, if I try to hear lobes with my speakers it is very hard no matter how far I'm, even up close. DI in general is easy to perceive in comparison, high frequencies increase while on axis and decrease when on axis so there is clear shift in perceived spectrum, but it is very hard to hear when ear is at the null. My c-c is also past 1, here vertical normalized plot from simulation of current system.
Last edited:
The raw measurements of SB15. tktran303 mentioned the directivity of SB15 off axis.
I meant the area of sudden directivity change, the widening, the responses bunch together.
^ Hi, that is baffle edge diffraction, it makes null (destructive interference) on axis about at the 2kHz and some widening of pattern, but also narrowing below the 2kHz. The narrowing is the "main diffraction hump" that reduces with highly asymmetric baffle for example, and goes away with a spherical enclosure. Both the hump and null very easy to identify on the graphs and together pretty much are a fingerprint of baffle edge diffraction. Baffle edge round overs mellow them out, so does minimizing flat baffle area around the transducer which can get rid of the null but not the hump. This anomaly on the graph would be much bigger, if the box they are measured had no bevel and was more symmetric.
As there is null about at 2kHz, which has about 17cm long wavelength, and since diffraction related back wave is opposite polarity and responsible for this interference we can rather easily approximate size of the baffle just looking at the graphs. Destructive interference would happen when there is another sound source opposite phase at some particular wavelength, so half wavelength late if it's in same polarity, or full wavelength late if the secondary sound source is in opposite polarity. Since diffraction back wave is in opposite polarity, relatively long portion of baffle edge (diffracting feature) must be roughly about 17cm away from center of the driver to make such interference null on-axis. If you look at photo of hifijim test box it seems to be about foot wide, so pretty close.
If you want to move this null to 3kHz, you must shrink the baffle, bring the edges closer. If you wish to drop it to 1kHz you'd increase size of the baffle making the edge further from the source. These change path length difference between direct sound and this secondary sound source, which would shift at which frequency interference ripple shows up. If you want the diffraction ripple to disappear, use round over that starts immediately beside the transducer.
Anyway, important bit here is that this is not feature of the driver, but something that is caused by the construct the driver is attached to.
As there is null about at 2kHz, which has about 17cm long wavelength, and since diffraction related back wave is opposite polarity and responsible for this interference we can rather easily approximate size of the baffle just looking at the graphs. Destructive interference would happen when there is another sound source opposite phase at some particular wavelength, so half wavelength late if it's in same polarity, or full wavelength late if the secondary sound source is in opposite polarity. Since diffraction back wave is in opposite polarity, relatively long portion of baffle edge (diffracting feature) must be roughly about 17cm away from center of the driver to make such interference null on-axis. If you look at photo of hifijim test box it seems to be about foot wide, so pretty close.
If you want to move this null to 3kHz, you must shrink the baffle, bring the edges closer. If you wish to drop it to 1kHz you'd increase size of the baffle making the edge further from the source. These change path length difference between direct sound and this secondary sound source, which would shift at which frequency interference ripple shows up. If you want the diffraction ripple to disappear, use round over that starts immediately beside the transducer.
Anyway, important bit here is that this is not feature of the driver, but something that is caused by the construct the driver is attached to.
Last edited:
tmuikku: exactly my thoughts.
With 3way with smaller mid + tweeter the common trouble is that baffle geometry often fits just one of the drivers, and is not good for the other one. So one has to play carefully with diffraction simulator and balance between diffraction signatures to tweeter and the midrange. And moreover, the baffle and drivers arrangement must also look good, which further complicates the things and leads to compromises.
With 3way with smaller mid + tweeter the common trouble is that baffle geometry often fits just one of the drivers, and is not good for the other one. So one has to play carefully with diffraction simulator and balance between diffraction signatures to tweeter and the midrange. And moreover, the baffle and drivers arrangement must also look good, which further complicates the things and leads to compromises.
Yes, compromises are unavoidable in every project, and all one has to do is be mindful what is it that you want to achieve. If one wants good looks, then there might be some ripple in the plots and it's fine, a conscious and acceptable trade-off taken, aesthetics prioritized over measurement graphs. If one wants the ripple go away, then perhaps trade-off the aesthetics, prioritize measured graphs over the aesthetics. Any project is successful when a goal is met, so either is fine as long as compromises aligns with the project goal at hand 🙂
Last edited:
Interesting diffraction response for the SB15NBAC in this cabinet.
I have SB15CAC30 drivers. The datasheet response of NBAC and CAC looks almost identical
SB15CAC
SB15NBAC
And here is the response for the SB15CAC I measured on a cabinet like below
The 2kHz bunching was not there for me..
As others have pointed out, baffle width was much less for me
I have SB15CAC30 drivers. The datasheet response of NBAC and CAC looks almost identical
SB15CAC
SB15NBAC
And here is the response for the SB15CAC I measured on a cabinet like below
The 2kHz bunching was not there for me..
As others have pointed out, baffle width was much less for me
Attachments
Last edited:
+1 👍 yea its also a piece of furniture!
@PKAudio : thanks for clarification, thought it was related to my graphs.
One addition to the 1.2 x fs wavelength driver spacing; This is the resulting lobe in theory:
And this approximately with my tight spacing:
I'll rather take that last shape for what I want to have the sound coming out the speaker gererally.
But I admit the overall area within the red line gets larger for the wider 1.2x fs spacing, this is what results then as more sound power in the spinorama.
EDIT: @tmuikku :
Had to think about that transition you speak of, it may have a lot to do with the critical distance. I must admit I haven't consciously tried it myself yet, thanks for the hint to test that more systematically.
I have two rooms where my speakers are playing. One big living room ~40m² where I sit at >3 m distance, and my home office room ~14m² where I listen from <2m distance. So I think in one room I'm sitting more out of that transition radius you speak of, and in the other room more within.
Last edited:
@vineethkumar01 and @PKAudio, for comparison:
SB15NBAC30-4 in 31,5cm wide monkey coffin, placed top of baffle above tweeter, side offset ~golden ratio:
SB15NBAC30-4 in 18cm wide compact speaker, rounded edges:
(wiggle @800-900Hz is port length resonance influence)
Looks good @2kHz, something to observe @3kHz after the response drop at wider angles. I lowpass that driver around 2,2kHz LR4 incl. passive series notch for 10kHz breakup in both concepts.
SB15NBAC30-4 in 31,5cm wide monkey coffin, placed top of baffle above tweeter, side offset ~golden ratio:
SB15NBAC30-4 in 18cm wide compact speaker, rounded edges:
(wiggle @800-900Hz is port length resonance influence)
Looks good @2kHz, something to observe @3kHz after the response drop at wider angles. I lowpass that driver around 2,2kHz LR4 incl. passive series notch for 10kHz breakup in both concepts.
Last edited:
Yeah, I've been using term "audible critical distance" as I'm not sure if it coincides with the actual critical distance you linked, and is perceptually analogous in my imagination, simplified it feels like either direct or room sound dominates. I've found the perception correlate well with what David Griesinger talks a lot about, what he calls Limit of Localization Distance, but not sure if it's that exactly, but there abouts and likely for same reason, brain's ability to lift the direct sound to a separate foreground audio stream based on how well original harmonics are preserved. Being due to what ever it seems to be on/off kind of phenomenon at some distance from speakers, and it makes quite a stark difference perceptually, literally one step and it's like getting "into the sound 3D clarity stuff" or "out into my room with frontal hazy image" kind of difference by using colorful language. Quite easy to hear anyway, and hazyness vs. clarity of phantom image is perhaps the simplest way to describe the difference.
Yeah I think it's fundamental property of home stereo systems, which to me marks hifi sound, being closer to the transition is what I call hifi sound and beyond it's good as well but not hifi to me in that sense. I've been using listening test like so:
put mono noise playing, try to get as strong phantom image as possible. Now start at far distance, walk closer to speakers staying equidistant and concentrate on listening the phantom center. It's kind of a hazy blob when you are far away and then gets into focus when you get close enough. Try to detect these two states, and mark the distance you perceive where the transition between the two happens. Closer than this, your brain pays attention to the sound and lifts it into focus above noise in the room. Further than this and it's just noise somewhere in direction of the speakers. Now test with music 🙂 I do not know if this happens with all rooms and speakers, but I'd believe it does, further or closer. Once I heard it, it has been easy to detect even with all kinds of loudspeakers, or talking to someone in same room even.
Last edited:
get a grip, uhm .. I mean, grill! 😆
I must admit that with a two-way, it's sometimes possible to localize the woofer with a wider driver spacing: in the event when a section of music which was predominantly above xo frequency changes to another section that is predominantly below.
But most times, the wider driver spacing is not perceived at all.