You have lots of experience and good communication skills and logical thought processes, I'll follow you!😀
Open baffles are very interesting in this regard. Arrays have their issues like you named, so does OB:s and most possibly every other candidate for reduced vertical early reflections. It is a nice thing you finally got results, after so long time searching, I'm happy for you!
I'll post my ramblings on suitable threads as I don't have dedicated time for my project, it has taken few years to get here and probably takes some more to get to an end. Will make a thread eventually, hopefully 🙂
ps. regarding temporal masking and stuff, my current though process is this: vertical reflections are seen as interference pattern at listening spot (VCAD in-room estimate), hence their magnitude can be evaluated. IF, the early vertical reflections are detrimental to sound experience, then minimizing them would yield better sound experience and this can be evaluated looking at the interference pattern. Only thing left is to know what is enough and if there are some bandwidth that benefits most etc. I guess you seem to have landed on a happy place now, so maybe that is enough. Maybe target what OB can provide for typical listening living room listening situation. Need to checkout studies on this at some point, what they have concluded. Certainly it is very hard to make wide bandwidth narrow vertical pattern, while still having nice wide horizontal pattern. Perhaps there is a nice balanced compromise for all of this. Usually enough is best option, it leaves room for other compromises find their place in the system 🙂
Last edited:
Always good to check your simulation based epiphanies first 😉
Here is a very good start for research
https://www.audiosciencereview.com/...nd-measurements-audioholics.23519/post-791482
Hah, thanks! It is so much fun I'm not afraid doing mistakes, learning for granted. Making it all public hopefully helps others as well, and gets others help to spot the errors 🙂 Alright time to move on, lets not divert the thread to something else.
Thanks Ken for accommodating the visit!
Thanks Ken for accommodating the visit!
Thanks for the update.
Comments in response
The revised vertical directivity plot shows that the floor bounce has an innocuous spectrum, while the ceiling has a large dip - however at a relatively low frequency. I'll comment on the interference process below.
I note that the ceiling dip may not be as bad as it looks, if the bass-floor-ceiling reflection which arrives shortly after the main ceiling reflection should be counted - I don't know if it should. Also, I note that at 400Hz, the midrange is only down a few dB in the ceiling direction, whether or not that's a critical few dB, I don't know.
Stepping back from the detail
My hearing finds the lower crossover (together with redone DRC) improved wth respect to spatial cues.
I know that's not a high-quality observation, and isn't much to go on.
I plan to look at the DRC impulses for clues as to why this might be, but expect that to be hard due to the Gabor limit.
It's reasonable to expect that mammal/... /human hearing evolved to pick out the direction of sources and that low-level processing (including temporal masking) contribute to this ability. Our aim is then to fool ourselves into making speakers that are hard to identify. It would be consistent with my experience that crossovers at 1-2kHz are the most troublesome, and could be convinced that an imperfect crossover at 450Hz could still be less trouble in a typical small room, than a better-designed crossover at 700Hz.
After revising my reading of Bech (thanks fluid), I am still doubtful about how to interpret the relevant corner of the time-frequency plot (i.e. sub kHz, and delays of a few ms). At this point I'd not say that I know a good recipe to follow. In future, I'd likely play safe and exclude crossovers between around 400Hz and 3kHz (unless the upper crossover is for co-axial drivers). That's about three octaves, so a stretch for a dipole midrange.
Comment on interference
When thinking about interference I keep in mind detection of the superposed waves. A microphone detects a sum which would correspond to a power spectrum like those in your simulations. The structures in the ear and spatial arrangement of sensory hair cells sample a set of superpositions - producing a more complicated bundle of detected signals. Then cochlear and downstream neurons process this bundle ( eventually binaurally) in a way that I expect isn't straightforward, e.g., https://journals.physiology.org/doi/pdf/10.1152/jn.00160.2012 which I guess is still far simpler than reality.
My point - I can't say whether "this" makes sense to me or not.
I look forward to your thread whenever it appears.
Ken
A good 8 inch fullrange driver can do that. In fact it can do more than that. I'm using one from 150Hz up to 3,5kHz and have no problems whatsoever with beaming. For a planar driver it is not an easy task.
I disagree: even with ideal drivers, the dipole peak with an 8" with zero surrounding baffle is problematic around 2kHz.
Hence my choice of a 4" wide unit for reasonable horizontal radiation pattern around the dipole peak.
Of course, that's by my criteria which are apparently different from yours. The advice to cover no more than 2.5 octaves per band in a dipole system is often put forward by folks far more experienced than I am.
Ken
Partial retraction/correction of post 52.
My comments about the 450Hz crossover remain as stated, but the starting point with the 650Hz crossover was probably not set up as intended. In brief, I had uploaded the wrong filter profiles for the 650Hz crossover, i.e. probably not the final version, and if that's the case, there was a phase mismatch (not enough to cause an obvious dip, but enough to tilt the main lobe significantly). In some quick comparisons, it appears that the change due to correcting the phase mismatch likely explains most of the difference I reported in the switch to 450Hz crossover. I'd no longer make a strong argument that the 450Hz crossover is better per se., though I'm still reasonably sure it's not worse than the higher crossover. The difference is now subtle rather than stark.
I discovered the error when I tried to see if I could push the crossover even lower. Based on measurements of the midrange, I decided it could go down to 400Hz without pushing the drivers too hard (even with loud music). When I switched back to 650Hz, to redo the comparison, I spotted that the saved version of the crossover wasn't as expected. I can't be certain, but based on file dates, the version installed shortly before post 52 was also likely to have had the same error.
For HFD users: at least on my computer, the Hypex software only works properly with filter profiles in the default directory. Initially, I kept them elsewhere (in a folder related to these speakers, which seemed a good idea at the time). Ibelieve this caused problems with creating and saving new profiles and hence version control. I've reverted to keeping profiles in .../Users/..../FilterData and haven't had any trouble since.
Apologies to those who spent time puzzling about the misleading results.
Ken
No problem from my part at least!🙂 it is quite easy to make mistakes and suffer misbehaving tools, things happen! Good you found it!
I read some papers fluid linked here and adventured further following citations and got impression the <1khz stuff is not very critical considering imaging (spatial perception). It was already few weeks ago and forgot most of it 😀
ps. I noticed few hours ago that my simulator DSP setings was not configured for my DSP, no wonder there was always something to tweak by ear. well, live and learn, currently configuring same xo again, hopefully there is no need to tweak by ear anymore 🙂
I read some papers fluid linked here and adventured further following citations and got impression the <1khz stuff is not very critical considering imaging (spatial perception). It was already few weeks ago and forgot most of it 😀
ps. I noticed few hours ago that my simulator DSP setings was not configured for my DSP, no wonder there was always something to tweak by ear. well, live and learn, currently configuring same xo again, hopefully there is no need to tweak by ear anymore 🙂
Last edited:
Thanks tmuikku,
regarding sub-1kHz: I suspect that, when everything is behaving well above 1kHz, there's still something to be gained by obtaining good performance at lower frequencies. Certainly, the crossover error at 650 Hz affected imaging in a significant way, that's a counter example (for my hearing at least). After all, the vast majority of speakers are more-or-less broken in the 1-2 kHz range.
Ken
Early reflections: Part 1: introduction and vertical reflections
Since long before I started with these speakers, I had in mind the problem of controlling early reflections. Experience led me to think that side wall and front wall reflections have an effect on imaging clarity. I’ve been puzzled for years about the significance of floor and ceiling reflections, and until now unable to find correspondence between measurements or models of reflections and timbral or imaging differences.
Having settled on crossovers, listening position, position and angle of the speakers in the room, I traced out and measured the early reflections, to compare with the Bech et al papers mentioned by Fluid earlier in the thread (via a link in #82). The detail of horizontal reflections is presented in a subsequent post. I’m not sure how to interpret these papers for music – I guess that’s closer to the “noise” results.
I wondered better control of horizontal early reflections that might make it easier to hear effects from floor or ceiling “vertical” reflections.
Delay times are inferred from distance and – in most cases – measured; helped by disabling crossovers to allow higher frequencies through to obtain a clearer peak. Most delays are rounded to 1ms and are relative to the direct sound: i.e., subtracting 8ms from the distance calculation or measured relative to the direct peak.
Floor
The first arrival is the bass-floor reflection: it should be around 1.5ms after the direct path, but smeared out due to the geometry. It should be a strong reflection of the half of the total amplitude that comes through the bass drive with the 400Hz LR4 crossover, but I tried modifying it using obstacles and couldn’t hear any change what-so-ever. Perhaps because of low sensitivity to such effects around and below 400Hz.
Where I might hear a change in imaging and possibly timbre is when modifying the midrange-floor reflection. The delay is 4ms. In the middle of its band (~2kHz) the floor reflection is something like 8 dB down, and I can reduce that by at least several dB using a heap of polyester wadding about 25cm thick, and loose enough not to reflect much from its top surface as far as I can find. I’m not sure if I’d score well on an ABX of this, but there might be a subtle effect. I will think about further investigation of this.
Ceiling
The bass-ceiling reflection is 11ms delayed, should be relatively low due to the longer path, and either from Bech or in comparison with the floor reflection, I doubt it matters. I couldn’t find it in measurements, even after disabling the low pass filter.
The midrange-ceiling reflection comes at a steeper angle than that from the floor and with 8ms delay relative to direct sound. This means it starts off around –12dB in the mid-band due to reduced off-axis radiation and has farther to travel. I suspect isn’t problematic – as Tmuikku suggested upthread.
Conclusion
With the possible exception of the midrange-floor reflection, I’m not hearing effects of vertical early reflections.
Since long before I started with these speakers, I had in mind the problem of controlling early reflections. Experience led me to think that side wall and front wall reflections have an effect on imaging clarity. I’ve been puzzled for years about the significance of floor and ceiling reflections, and until now unable to find correspondence between measurements or models of reflections and timbral or imaging differences.
Having settled on crossovers, listening position, position and angle of the speakers in the room, I traced out and measured the early reflections, to compare with the Bech et al papers mentioned by Fluid earlier in the thread (via a link in #82). The detail of horizontal reflections is presented in a subsequent post. I’m not sure how to interpret these papers for music – I guess that’s closer to the “noise” results.
I wondered better control of horizontal early reflections that might make it easier to hear effects from floor or ceiling “vertical” reflections.
Delay times are inferred from distance and – in most cases – measured; helped by disabling crossovers to allow higher frequencies through to obtain a clearer peak. Most delays are rounded to 1ms and are relative to the direct sound: i.e., subtracting 8ms from the distance calculation or measured relative to the direct peak.
Floor
The first arrival is the bass-floor reflection: it should be around 1.5ms after the direct path, but smeared out due to the geometry. It should be a strong reflection of the half of the total amplitude that comes through the bass drive with the 400Hz LR4 crossover, but I tried modifying it using obstacles and couldn’t hear any change what-so-ever. Perhaps because of low sensitivity to such effects around and below 400Hz.
Where I might hear a change in imaging and possibly timbre is when modifying the midrange-floor reflection. The delay is 4ms. In the middle of its band (~2kHz) the floor reflection is something like 8 dB down, and I can reduce that by at least several dB using a heap of polyester wadding about 25cm thick, and loose enough not to reflect much from its top surface as far as I can find. I’m not sure if I’d score well on an ABX of this, but there might be a subtle effect. I will think about further investigation of this.
Ceiling
The bass-ceiling reflection is 11ms delayed, should be relatively low due to the longer path, and either from Bech or in comparison with the floor reflection, I doubt it matters. I couldn’t find it in measurements, even after disabling the low pass filter.
The midrange-ceiling reflection comes at a steeper angle than that from the floor and with 8ms delay relative to direct sound. This means it starts off around –12dB in the mid-band due to reduced off-axis radiation and has farther to travel. I suspect isn’t problematic – as Tmuikku suggested upthread.
Conclusion
With the possible exception of the midrange-floor reflection, I’m not hearing effects of vertical early reflections.
Early reflections: Part 2: horizontal reflections
The question I had in mind is how dipole speakers can be deployed to help clean up the early reflections.
Listening space
The main area of the listening space is 6.5m (21’) wide by 4.5m (15’) deep. The front wall is hard stone; it’s reflective, but it should diffuse wavelengths below about 10cm as it is rough on those scales, still I expect the critical band for imaging to be reflected.
The side walls and rear wall are drywall, windows, partitions, etc. There’s a large alcove & often-open door to a hallway behind the listening position which reduce the significance of rear reflections.
The speakers are toed-in about 50 to 55 degrees and point a little in front of the main listening position, this creates two other balanced listening positions to left and right. The triangle at the main listening position is approximately 9:12:9 (in feet, roughly).
From the listening position the speakers are in front of corners of the room. This gives the longest delay that I could achieve for the rear dipole lobe.
Measurements and summary of results
The delays noted below are again relative to the direct sound. Relative levels are estimated based on a mixture of the known speaker radiation (averaged over frequency) or observing impulses. Amplitude estimates are rough: allow ±3dB at best. Where the reflection has a different frequency response than the direct sound, the amplitude is estimated at 2-3 kHz. These measurements were done full-range, and it was much easier to pick out delayed impulse responses than for the vertical measurements in which bass and midrange reflections became separated.
The following are the early horizontal reflections, in order of increasing delay:
Conclusions
Some important early reflections are strongly suppressed by the particular deployment of these speakers in the room. The rear dipole radiation is delayed by what seems to be a reasonable amount for a small room. The opposing-wall reflections are the only other strong ones within 20ms.
Given the freedom to locate them in the room, dipoles provide options for control of early reflections that wouldn’t be possible with small monopole speakers. In this regard they are behaving like “narrow directivity” loudspeakers: imaging that could be described as precise or clear, but not extremely wide.
The question I had in mind is how dipole speakers can be deployed to help clean up the early reflections.
Listening space
The main area of the listening space is 6.5m (21’) wide by 4.5m (15’) deep. The front wall is hard stone; it’s reflective, but it should diffuse wavelengths below about 10cm as it is rough on those scales, still I expect the critical band for imaging to be reflected.
The side walls and rear wall are drywall, windows, partitions, etc. There’s a large alcove & often-open door to a hallway behind the listening position which reduce the significance of rear reflections.
The speakers are toed-in about 50 to 55 degrees and point a little in front of the main listening position, this creates two other balanced listening positions to left and right. The triangle at the main listening position is approximately 9:12:9 (in feet, roughly).
From the listening position the speakers are in front of corners of the room. This gives the longest delay that I could achieve for the rear dipole lobe.
Measurements and summary of results
The delays noted below are again relative to the direct sound. Relative levels are estimated based on a mixture of the known speaker radiation (averaged over frequency) or observing impulses. Amplitude estimates are rough: allow ±3dB at best. Where the reflection has a different frequency response than the direct sound, the amplitude is estimated at 2-3 kHz. These measurements were done full-range, and it was much easier to pick out delayed impulse responses than for the vertical measurements in which bass and midrange reflections became separated.
The following are the early horizontal reflections, in order of increasing delay:
- Right speaker sound scattering from corner of alcove and passing seat-back: about 3ms delay and about 20dB down at 2kHz.
- Same-side wall (either speaker): 7ms delay, comes from about 100 degrees off axis therefore strongly suppressed by dipole pattern. This arrival peak was not visible in measurements.
- Left speaker – rear wall – the main reflection has at least 10ms delay with, in addition, scattering from alcove corners that, however, couldn’t be measured.
- Front wall – 12ms delay, again comes from about 100 degrees off axis therefore strongly suppressed by the dipole pattern, roughly 20dB down. The peak was found but hard to measure as it was smeared out over at least 2 ms. Probably due to furniture on the front wall.
- Corner reflections – 13ms delay, comes from rear radiation, at least 6dB down at listener around 2kHz, again the impulse is smeared.
- Opposing wall – 17ms delay, comes from 45 degrees off axis– about 10dB down due. This is the clearest horizontal reflection: it is nearly a copy of the direct sound, with some roll-off above a few kHz.
- Rear-then-front walls – at 24ms delay is the last identifiable peak above the chaotic background in the total impulse response.
Conclusions
Some important early reflections are strongly suppressed by the particular deployment of these speakers in the room. The rear dipole radiation is delayed by what seems to be a reasonable amount for a small room. The opposing-wall reflections are the only other strong ones within 20ms.
Given the freedom to locate them in the room, dipoles provide options for control of early reflections that wouldn’t be possible with small monopole speakers. In this regard they are behaving like “narrow directivity” loudspeakers: imaging that could be described as precise or clear, but not extremely wide.
I re-read a few posts from kimmo today that made me rethink on my dipole and line array experiences.
https://www.audiosciencereview.com/forum/index.php?threads/some-help-with-lobing.22661/post-753404
https://www.audiosciencereview.com/forum/index.php?threads/some-help-with-lobing.22661/post-755427
"Traditionally XO frequencies 2.5-5 kHz were common, and sound of those speakers was typically smoother and more tolerable than (modern) low XO point. So I'm not trying to invent anything new or provoke. Just giving an answer why some sound features were better in the past; no blood from ears while listening 80s' Gary Moore or Iron Maiden".
Iron Maiden was something I had real trouble with on the LX521's. Perhaps I attributed this wrongly to the dipole radiation when in fact it could have be the difference in vertical directivity changing the early reflection patterns. I have to wonder if the narrow vertical directivity (due to the planar drivers) is having a similar effect here and would go somewhat to explain the tolerance of it to "difficult" music.
Higher vertical directivity being suggested in the second link as an alternative way to achieve similar results to the ~1.2 to 1.4 wavelength separation.
https://www.audiosciencereview.com/forum/index.php?threads/some-help-with-lobing.22661/post-753404
https://www.audiosciencereview.com/forum/index.php?threads/some-help-with-lobing.22661/post-755427
"Traditionally XO frequencies 2.5-5 kHz were common, and sound of those speakers was typically smoother and more tolerable than (modern) low XO point. So I'm not trying to invent anything new or provoke. Just giving an answer why some sound features were better in the past; no blood from ears while listening 80s' Gary Moore or Iron Maiden".
Iron Maiden was something I had real trouble with on the LX521's. Perhaps I attributed this wrongly to the dipole radiation when in fact it could have be the difference in vertical directivity changing the early reflection patterns. I have to wonder if the narrow vertical directivity (due to the planar drivers) is having a similar effect here and would go somewhat to explain the tolerance of it to "difficult" music.
Higher vertical directivity being suggested in the second link as an alternative way to achieve similar results to the ~1.2 to 1.4 wavelength separation.
Fluid, good points. Also, timely, as I've been musing on early reflections (ER) affecting timbre and imaging.
Kimmo's comments, read in brief, freed my mind to consider three lobes for both crossovers. For the mid-tweeter, the drivers touch and couldn't cross low enough to get a single lobe, so there's only the frequency and slope to choose. The only "hardware" choice was to apply proportional spacing for bass-mid leading to W-T-M rather than W-M-T order.
Now I that I've moved the bass-mid crossover to 400Hz, that's moot. I guess I could invert the mid-tweeter unit to try W-M-T as well. Moving the mid down (keeping the tweeter in the same place) trades off ceiling reflections for floor - no idea if that's better or worse.
I'm still not very clear on importance of floor and ceiling reflections, as opposed to, say, total ER. I remain unsure where the thresholds of ER audibility fall in the time-frequency plane for timbre and imaging.
On the design-side, I hadn't considered 2nd reflections of one driver adding to the first reflection of the other (as mentioned up-thread).
Now that I've cleaned-up the first 20ms early reflections, I've been trying to hear if I can hear any effect of absorbing the floor reflection. I measure several dB change at 1-2kHz in the floor reflection, but the only thing I hear is a subtle (if not imagined) timbral change.
I'm considering rigging up a "clothesline" from which I can suspend absorbing materials. I won't do it soon as I doubt that I'll learn much nor will I leave such absorption in place.
Extending the topic: perhaps there's something to be learned from the links between direction and frequency range?
I'm getting smooth sound power curves, so don't think there's much to gain in terms of timbre, and I don't expect vertical to affect imaging in the L-R sense, but I wonder if there's something to learn from Blauert's bands - five frequency ranges associated with direction information in 3D.
http://www.sengpielaudio.com/DieBedeutungDerBlauertschenBaender.pdf There are supposed to be English translations but the links I found are dead.
If I understand correctly, and reading the band edges roughly:
I've not seen much discussion of this in the context of ER. The last column is a comment on what I've got in my setup ER means up to 20ms. "Low ER" means well below direct sound (see #91). Perhaps strong ER in one or other band causes confusion.
Floor bounce is again the main suspect, and it's hard to absorb <540Hz. I tried experiments with baffles and absorption that reduce the floor reflection by a few dB around 500Hz, but couldn't say I heard a clear difference.
I'm also wondering what I'd do differently with the experience gained, that's for later.
Ken
I see tables are still broken. I'll leave the input, in case it's fixed later.
Kimmo's comments, read in brief, freed my mind to consider three lobes for both crossovers. For the mid-tweeter, the drivers touch and couldn't cross low enough to get a single lobe, so there's only the frequency and slope to choose. The only "hardware" choice was to apply proportional spacing for bass-mid leading to W-T-M rather than W-M-T order.
Now I that I've moved the bass-mid crossover to 400Hz, that's moot. I guess I could invert the mid-tweeter unit to try W-M-T as well. Moving the mid down (keeping the tweeter in the same place) trades off ceiling reflections for floor - no idea if that's better or worse.
I'm still not very clear on importance of floor and ceiling reflections, as opposed to, say, total ER. I remain unsure where the thresholds of ER audibility fall in the time-frequency plane for timbre and imaging.
On the design-side, I hadn't considered 2nd reflections of one driver adding to the first reflection of the other (as mentioned up-thread).
Now that I've cleaned-up the first 20ms early reflections, I've been trying to hear if I can hear any effect of absorbing the floor reflection. I measure several dB change at 1-2kHz in the floor reflection, but the only thing I hear is a subtle (if not imagined) timbral change.
I'm considering rigging up a "clothesline" from which I can suspend absorbing materials. I won't do it soon as I doubt that I'll learn much nor will I leave such absorption in place.
Extending the topic: perhaps there's something to be learned from the links between direction and frequency range?
I'm getting smooth sound power curves, so don't think there's much to gain in terms of timbre, and I don't expect vertical to affect imaging in the L-R sense, but I wonder if there's something to learn from Blauert's bands - five frequency ranges associated with direction information in 3D.
http://www.sengpielaudio.com/DieBedeutungDerBlauertschenBaender.pdf There are supposed to be English translations but the links I found are dead.
If I understand correctly, and reading the band edges roughly:
| Front (LF) | 270Hz | 540Hz | Substantial, mostly floor refl. |
| Rear (Mid) | 700Hz | 1.4kHz | Low ER |
| Front (Mid) | 2.5kHz | 5.3kHz | Low ER |
| Up | 7kHz | 9.3kHz | Low ER (due to tweeter directivity) |
| Rear (HF) | 9.8kHz | 13kHz | Low ER |
I've not seen much discussion of this in the context of ER. The last column is a comment on what I've got in my setup ER means up to 20ms. "Low ER" means well below direct sound (see #91). Perhaps strong ER in one or other band causes confusion.
Floor bounce is again the main suspect, and it's hard to absorb <540Hz. I tried experiments with baffles and absorption that reduce the floor reflection by a few dB around 500Hz, but couldn't say I heard a clear difference.
I'm also wondering what I'd do differently with the experience gained, that's for later.
Ken
I see tables are still broken. I'll leave the input, in case it's fixed later.
Don't visit much - listening to music!
To cap this thread with brief comments.
GRS tweeter: couldn't get completely happy with 8kHz resonance; in lieu of a dipole tweeter tried some co-ax options:
Added some ceiling absorption and floor diffusion. Raised tweeter close to floor-ceiling midpoint (1.2m) which seems to work well for unknown reason (And not only in normal listening position). Dipoles now sitting on 12" closed box woofers (raised 15"), giving cardiod response over a couple of octaves. Stable for months now.
Done.
Ken
To cap this thread with brief comments.
GRS tweeter: couldn't get completely happy with 8kHz resonance; in lieu of a dipole tweeter tried some co-ax options:
- BMS 10C262 - close but couldn't quite get on and off-axis performance right at 6kHz (tip - 0.15mH parallel with tweeter, 8 ohms series gives good starting point response)
- KEF 1708 - extremely picky about diffraction from baffle, couldn't get it quite right, though came close; without the following this could have been fine.
- Sica 5.5/1" - minimal rectangular baffle works well, amazingly easy to get smooth performance except exactly on-axis. Some trouble at 3.2kHz, but no audible consequences.
Added some ceiling absorption and floor diffusion. Raised tweeter close to floor-ceiling midpoint (1.2m) which seems to work well for unknown reason (And not only in normal listening position). Dipoles now sitting on 12" closed box woofers (raised 15"), giving cardiod response over a couple of octaves. Stable for months now.
Done.
Ken
Hmm, the middle is a deep fundamental null, but the odd harmonics on either sides gets a summed blend of both, so a good place to position horns IME.
Interesting change, also an option that fell within kimmo's choices for vertical 🙂 If you feel like posting a picture or two it might made it easier to visualize the end result. I am interested in how you balanced the crossover for the coax to your liking too, as at same point I will have the same task.
No pictures available for now.
The baffle for the Sica is 6 inch wide by 7 inch tall approximately with the OD of the Sica touching left right and top edges. It is attached directly top-center of the baffle for the 14 inch (as pictured earlier, no changes there), giving a c-2-c of 13 or 14 inches.
The Sica crossover points are ~450Hz and ~3.0kHz. The 3kHz electrical low pass is two poles at 3kHz, two at 5kHz, all 0.6<Q<0.7, the high pass is 2nd order at 5kHz, Q=0.55 plus some other shaping. To reach -6dB at ~400Hz requires only a few dB of boost (<4), so the Sica is lightly loaded even with loud music.
The net effect is to give a slope of about -1dB from 1kHz to 10kHz at about 15 degrees off-axis, as usual the on-axis response is rougher.
For a while I had the KEF and Sica units on their small baffles side by side (with a gap) on top of the bass-mid. Each time I tweaked the crossovers I felt that I was struggling to smooth out problems on the KEF while the Sica kept sounding fine and in some respects measuring better. After a few weeks and several iterations, I went with what was easier.
The KEF has lower midrange distortion, but I don't hear that difference with music. The choice was made based on higher frequencies.
Ken
edit - missing word
The baffle for the Sica is 6 inch wide by 7 inch tall approximately with the OD of the Sica touching left right and top edges. It is attached directly top-center of the baffle for the 14 inch (as pictured earlier, no changes there), giving a c-2-c of 13 or 14 inches.
The Sica crossover points are ~450Hz and ~3.0kHz. The 3kHz electrical low pass is two poles at 3kHz, two at 5kHz, all 0.6<Q<0.7, the high pass is 2nd order at 5kHz, Q=0.55 plus some other shaping. To reach -6dB at ~400Hz requires only a few dB of boost (<4), so the Sica is lightly loaded even with loud music.
The net effect is to give a slope of about -1dB from 1kHz to 10kHz at about 15 degrees off-axis, as usual the on-axis response is rougher.
For a while I had the KEF and Sica units on their small baffles side by side (with a gap) on top of the bass-mid. Each time I tweaked the crossovers I felt that I was struggling to smooth out problems on the KEF while the Sica kept sounding fine and in some respects measuring better. After a few weeks and several iterations, I went with what was easier.
The KEF has lower midrange distortion, but I don't hear that difference with music. The choice was made based on higher frequencies.
Ken
edit - missing word
I see another Sica coax user vineethkumar01 whose results above ~2kHz are similar to what I get despite differences in the baffle design. We have a near-minimal front face in common - see A 3-way ... those off-axis measurements are better/clearer than mine but the trends are similar. Due to space constraints and nearby walls, etc., I did not carry out a full set of "spin" measurements of the final assembly & crossover, only of the basic test baffle with a crude test-bench crossover (DCX2496) and 4ms gate. The 3.2kHz glitch in the tweeter response I mentioned above looks the same - drawing attention close to the crossover frequency but it's not serious.
I didn't observe as pronounced a peak at 15kHz, by 1dB or so. Possibly mic calibration differences.
I've not suppressed the peaks in the midrange response by quite as much (due to the split pole crossover rather than LR24) so the residues of them show up more clearly in my listening-axis measurements. Otherwise the effect of our crossover choices is not very different.
Anyway, that link provides great information for anyone interested in the Sica coax.
Ken
ps. In the previous post, regarding the 15 degree measurement, I should have mentioned that was at 1.2m with 4.5ms gating. On-axis is broadly flat (but rougher, as seen in Vineeth's results). I'm not sure if we've gone for exactly the same balance, but it's very close (above 1-2kHz, not below due to my OB).
I didn't observe as pronounced a peak at 15kHz, by 1dB or so. Possibly mic calibration differences.
I've not suppressed the peaks in the midrange response by quite as much (due to the split pole crossover rather than LR24) so the residues of them show up more clearly in my listening-axis measurements. Otherwise the effect of our crossover choices is not very different.
Anyway, that link provides great information for anyone interested in the Sica coax.
Ken
ps. In the previous post, regarding the 15 degree measurement, I should have mentioned that was at 1.2m with 4.5ms gating. On-axis is broadly flat (but rougher, as seen in Vineeth's results). I'm not sure if we've gone for exactly the same balance, but it's very close (above 1-2kHz, not below due to my OB).
Amazingly interesting thread!
I can't explain how I missed it until now. Will re-read again several parts to let it sink in - lot's of food for thought.
What's your thinking/how are you liking the integration between the 2 sealed + 2 reflex subs with the OB midbass? I have two 12" sealed subs and been thinking of adding another two and apply Multi Sub Optimizator to get a smoother response up to 80-90Hz, and have been wondering whether the next 2 should be sealed or OB subs. Was wondering if you had explored OB subs, maybe had the itch and arrived at some conclusions?
I can't explain how I missed it until now. Will re-read again several parts to let it sink in - lot's of food for thought.
What's your thinking/how are you liking the integration between the 2 sealed + 2 reflex subs with the OB midbass? I have two 12" sealed subs and been thinking of adding another two and apply Multi Sub Optimizator to get a smoother response up to 80-90Hz, and have been wondering whether the next 2 should be sealed or OB subs. Was wondering if you had explored OB subs, maybe had the itch and arrived at some conclusions?
Thanks.
The subs predate OB investigations by almost two decades. If starting now I'd probably use sealed with EQ for all. The big sub is a bandpass and only used below 30Hz. The other reflex sub doesn't work well if sealed, giving high distortion in the frequency range I need it most (it's a ~2004 BK electronics Monolith but I've bypassed the plate amp to use my own EQ and amp).
The sealed subs, sitting under the OBs, naturally form a cardioid response over a couple of octaves. That seems useful, given the distance behind the speakers which "point" into the corner of the room (2m distance each way). Of course, this indeases side-wall reflections below ~200Hz. The arrangement was reached by trial and error, and is room dependent. There's still a lot of the speaker-room interaction that I don't understand.
I'd probably do well to replace the reflex sub with two sealed ones placed to smooth out modes a little better (below 50Hz), but there's not much needing to be fixed, and I'd need more amp channels, so it's not a quick test.
OB lets me avoid side-wall reflections at higher frequencies, which I find helpful, and simplifies construction in the range where box resonances are troublesome. I'm not attracted by the idea of trying OB below the Schroeder frequency - I don't know how the mix of pros and cons would play out. If I had a huge room, I might consider using OB down lower - I've not really thought about that as it's not likely to happen.
Ken