Investigating floor and ceiling bounce

[augerpro]

Looking for everyone's vote on which plot they prefer below. I'm specifically investigating floor and ceiling bounce, and their impact on in-room response, and DI. The models use a 6" woofer, crossed to a 4.5" midrange at 2khz. The reason for the 4.5" is that I needed something to mimic my waveguide at the crossover frequency. It rolls off fast >4khz, so just ignore that, it's not relevant to what I'm looking at anyway.



On-axis

In-room

ER Floor

ER Vertical (thick line)

ER Ceiling (thin line)

DI

[chrisn]

um..the last one? It looks maybe a bit smoother. Followed by the first one I think.

[augerpro]

Here are the horizontal polars for C,D, and E, which seemed to be agreed as being better vertically than A and B:

On-Axis
In-room
ER Horizontal
ER Side
DI
ERDI Horizontal

[augerpro]

Since the horizontal polars are pretty close, I'll move to the next phase of analysis. Would you be comfortable with a ctc of 7.68" (195mm) for C, 8.46" (215mm) for D, and 9.25" (235mm) for E?

[theSven]

Are you going to test 3 of the CTC's? I don't have any experience with this part of the hobby. Painting and assembly are all I have done thus far, but enjoy getting deeper into this hobby!

[augerpro]

Test all 3? No. But it would be nice to understand the tradeoffs and go with the best balance.

[Reet]

Since the horizontal polars are pretty close, I'll move to the next phase of analysis. Would you be comfortable with a ctc of 7.68" (195mm) for C, 8.46" (215mm) for D, and 9.25" (235mm) for E?

All 3 are fine given a typical listening distance of 2-3m+. This really has nothing to do with floor or ceiling reflections though, but rather with aligning the directivity pattern of the drivers and crossover frequency and slope for the most optimal result. Many DIYers have followed a design rule of thumb that the closest driver separation is best, unfortunately it is based on missing information from the old days where single axis design and very basic polar charts were the only tools available. The idea was that closest driver separation makes for the widest polar lobe, therefore that's the best result. Reality is that at 2-3m+ the closest driver separation may not be ideal for in-room and power response, possibly the worst, and the wide vs narrow polar lobe is mostly irrelevant.

Kimmo provided some very good insight into this topic at the DIY Audio forum, unfortunately the VituixCAD discussion there is also a single thread so past topics of discussion are easily lost and hard to locate later on. I will try to reiterate, and hopefully he will chime in if I have some incorrect information here.

For LR crossover of standard cone midrange and dome tweeter with flat face plate (non-waveguide or horn), driver separation for best in-room response and power response is 1.0-1.4x the wavelength of the crossover frequency, ideally 1.2x separation. In VituixCAD, investigation of driver separation is easy. Simply load in measured data for each driver, apply crossover and response shaping using active blocks, then adjust y axis distance for the midrange driver. I assume you've done something similar to this above, but using simulated data from the diffraction tool rather than real driver measurements.

Given your crossover frequency of 2kHz, 1.2x distance would be 206mm, pretty close to your "D" result. With drivers closer together, increase the crossover frequency for better in-room and power response result.

In your horizontal data above, I don't understand why the power & DI chart has changed from the first post? In any case, since you are investigating waveguide response, best result would be to use real measured data of the waveguide in cabinet on the intended baffle, or at least in a box of typical width.

[augerpro]

My question concerning what CTC you would be comfortable with comes from my experience with some speakers, where (usually) vocals wander between drivers, and I find it distracting. So those who have tried these higher 1.2x ctc haven't had issue with that?

If not, that was my last concern about the 1.2x idea. My other concerns have been largely removed, and, serendipitously, have dovetailed with two other concepts I've been thinking about in the last couple years.

My first concern was that as you move drivers apart, the main lobe becomes much smaller, and new lobes at the top and bottom appear and start to move in. Now, the narrowed main lobe is obviously happening, just look at the plots. No doubt the sweet spot is much narrower vertically. However, I don't mind designing around that, and at the end of the day, would always be willing to build a stand that puts the speaker at the appropriate height for the listener. But I think this requirement should be appreciated, especially if one is designing for retail.

What I did not expect, was the upper and lower lobes move inward quite quickly, and the nulls between lobes would be lessened. I like that. I also like that the ceiling and floor reflection are also smoother as a result and vertical DI higher. That's the first finding that dovetails with some of my thinking lately. I've become interested in wide and smooth horizontal response, but also constrained and smooth vertical response. My experience is that hard floors and the typical 8' ceiling really muck up the sound, and I wanted to remove it by constraining it to some degree, but also make it smooth so at least it blends with the horizontal sound well. The 1.2x ctc helps do that. A speaker that really showed this behavior is the Perlisten bookshelf speaker tested by Erin, and which is extremely well received critically.

The second finding that meshed with some recent thinking was that, if I were concerned about vocals wandering around, I could keep a close(r) ctc, but raise the crossover frequency and get the same vertical responses. Now this is often a problem, due to DI mismatch, but with a waveguide I realized it could work well. The ability to raise a crossover, or at least use a formerly common 2.5-3.0khz, is actually easier with a waveguide. Because of the "path dependence" of the use of waveguides in the DIY forums, I think we kind of missed this. As I recall the evolution of this thinking, oh 15 years ago now, was that people were wanting to push tweeters lower, so that the crossover could be placed before the woofer starts beaming. This of course puts a lot stress on the tweeter and it was recognized that a horn could relieve this stress. And ever since the focus has been to use a waveguide to push the crossover lower. All well and good and true - as far as it goes. But at the same time, the use of the waveguide alleviated the need to cross to the woofer below where it started to beam in the first place. Because the waveguide was also "beaming" at that frequency, and as long as the match was smooth, and in-room response a nice smooth slope - mission accomplished.

I think perhaps minimizing ctc was one reason thinking stayed in the particular lane it did for so long. But now that there is some proven benefit to wider ctc, I no longer have a reason to not use waveguide to match a woofer at a higher than expected frequency. The tweeter will like it much better, but more will be required of the woofer. Still, I don't think there is any longer an a priori prohibition against, say, a 2.7khz crossover on a TM, from a sound quality perspective. Thoughts?

[Reet]

My question concerning what CTC you would be comfortable with comes from my experience with some speakers, where (usually) vocals wander between drivers, and I find it distracting.

And what did the power & DI chart look like for those speakers? Easily audible separation of midrange and trebel is easily described as "poor driver integration" and should be easily shown in the power & DI chart, and possibly there is some dramatic difference in distortion profiles between drivers as well, or even a basic level matching problem.

My first concern was that as you move drivers apart, the main lobe becomes much smaller, and new lobes at the top and bottom appear and start to move in. Now, the narrowed main lobe is obviously happening, just look at the plots. No doubt the sweet spot is much narrower vertically. However, I don't mind designing around that, and at the end of the day, would always be willing to build a stand that puts the speaker at the appropriate height for the listener. But I think this requirement should be appreciated, especially if one is designing for retail.

Keep in mind that at 3m listening distance, +/- 10 degree listening window is +/- 0.5 meter, hardly a "head in a vice" situation IMO. As well, the "lobe" is a narrow band problem only at the crossover frequency, again at listening distance it gets somewhat masked by room interactions, so I still put more weight into overall in-room and power response than the polar lobe, other than to make sure the lobe is forward facing.

What I did not expect, was the upper and lower lobes move inward quite quickly, and the nulls between lobes would be lessened. I like that. I also like that the ceiling and floor reflection are also smoother as a result and vertical DI higher. That's the first finding that dovetails with some of my thinking lately. I've become interested in wide and smooth horizontal response, but also constrained and smooth vertical response. My experience is that hard floors and the typical 8' ceiling really muck up the sound, and I wanted to remove it by constraining it to some degree, but also make it smooth so at least it blends with the horizontal sound well.

Perhaps you'd enjoy a ribbon or AMT, narrow width and long vertical dimension driver will have wide horizontal and narrow vertical by design. But seriously, buy a rug for the floor, tapestry on the wall, heavy blinds, etc. will help more than trying to fix a bad room with clever speaker design. We recently replaced the carpet in our basement with vinyl plank and the difference in room response was astounding, but addition of a big heavy rug really helped, and I'm still shopping for some nice tapestry to hang above the couch.

I think perhaps minimizing ctc was one reason thinking stayed in the particular lane it did for so long. But now that there is some proven benefit to wider ctc, I no longer have a reason to not use waveguide to match a woofer at a higher than expected frequency. The tweeter will like it much better, but more will be required of the woofer. Still, I don't think there is any longer an a priori prohibition against, say, a 2.7khz crossover on a TM, from a sound quality perspective. Thoughts?

Still need to balance the upper limits of the midrange as it works into breakup and distortion at higher frequencies. 3 way speaker still offers a lot of benefit here. 7" 2-way is like an all season tire, it tries to do everything but excels at nothing.

[JonMarsh]

My question concerning what CTC you would be comfortable with comes from my experience with some speakers, where (usually) vocals wander between drivers, and I find it distracting. So those who have tried these higher 1.2x ctc haven't had issue with that?

If not, that was my last concern about the 1.2x idea. My other concerns have been largely removed, and, serendipitously, have dovetailed with two other concepts I've been thinking about in the last couple years.

My first concern was that as you move drivers apart, the main lobe becomes much smaller, and new lobes at the top and bottom appear and start to move in. Now, the narrowed main lobe is obviously happening, just look at the plots. No doubt the sweet spot is much narrower vertically. However, I don't mind designing around that, and at the end of the day, would always be willing to build a stand that puts the speaker at the appropriate height for the listener. But I think this requirement should be appreciated, especially if one is designing for retail.

What I did not expect, was the upper and lower lobes move inward quite quickly, and the nulls between lobes would be lessened. I like that. I also like that the ceiling and floor reflection are also smoother as a result and vertical DI higher. That's the first finding that dovetails with some of my thinking lately. I've become interested in wide and smooth horizontal response, but also constrained and smooth vertical response. My experience is that hard floors and the typical 8' ceiling really muck up the sound, and I wanted to remove it by constraining it to some degree, but also make it smooth so at least it blends with the horizontal sound well. The 1.2x ctc helps do that. A speaker that really showed this behavior is the Perlisten bookshelf speaker tested by Erin, and which is extremely well received critically.

The second finding that meshed with some recent thinking was that, if I were concerned about vocals wandering around, I could keep a close(r) ctc, but raise the crossover frequency and get the same vertical responses. Now this is often a problem, due to DI mismatch, but with a waveguide I realized it could work well. The ability to raise a crossover, or at least use a formerly common 2.5-3.0khz, is actually easier with a waveguide. Because of the "path dependence" of the use of waveguides in the DIY forums, I think we kind of missed this. As I recall the evolution of this thinking, oh 15 years ago now, was that people were wanting to push tweeters lower, so that the crossover could be placed before the woofer starts beaming. This of course puts a lot stress on the tweeter and it was recognized that a horn could relieve this stress. And ever since the focus has been to use a waveguide to push the crossover lower. All well and good and true - as far as it goes. But at the same time, the use of the waveguide alleviated the need to cross to the woofer below where it started to beam in the first place. Because the waveguide was also "beaming" at that frequency, and as long as the match was smooth, and in-room response a nice smooth slope - mission accomplished.

I think perhaps minimizing ctc was one reason thinking stayed in the particular lane it did for so long. But now that there is some proven benefit to wider ctc, I no longer have a reason to not use waveguide to match a woofer at a higher than expected frequency. The tweeter will like it much better, but more will be required of the woofer. Still, I don't think there is any longer an a priori prohibition against, say, a 2.7khz crossover on a TM, from a sound quality perspective. Thoughts?

The main thing that popped in my mind was how well would the polar of the mid woofer match up with that of the tweeter/waveguide at 2700Hz? For a cone mid, of the right size, probably no worries; for something like a 7" mid woofer, it might be another matter. Even Accuton 6.5" midrange drivers really fall off off axis above 2 kHz, one reason most of my two way designs like the Natalie P cross at 1800 Hz.

I'm sort of a member of an old school of thought, which is to use the waveguide to extend the bottom end of the tweeter's capabilities enough that it's very comfortable at that crossover frequency.

Also, I was somewhat confused by the way you phrased your initial premise- I've spent a fair amount of time in the past looking at floor and ceiling bounce issues for the woofer/midwoofer and how to optimize the driver configuration in the range below 1kHz to get the smoothest overall in room results. But this doesn't seem to be something you're looking at here. Of course, I wasn't doing it with simulation, I was measuring configurations, which is why the early Modula Supreme got modified a lot, and then eventually morphed to the Isis style approach after testing that.

[augerpro]

Jon> yes it puts certain dispersion requirements on the woofer, so 2700hz with most 18cm is probably not going to happen. But now it just becomes a variable to juggle, not an automatic "never cross a woofer to a tweeter at such high frequencies" sort of rule.

Actually I am interested in having some vertical directivity <1khz too. I'm modeling some 3-ways in various configurations now to sort out some candidates for testing. One thing I have noticed, LR2 really helps things here. I would like to see how a Dueland would model, I've posted in a VCad thread to include it in a future update.

[augerpro]

I'm sort of a member of an old school of thought, which is to use the waveguide to extend the bottom end of the tweeter's capabilities enough that it's very comfortable at that crossover frequency.

And you certainly can, it's just not the primary focus in my mind anymore. It's more nuanced. Here are plots of D again (8.5" ctc, 2000hz), and new one crossing at 2550hz with 7.68" ctc. Obviously they need a bit more EQ for a perfect comparison, but notice how the ceiling reflection 1.5-3khz really gets knocked down and the in-room response is smoother because of it.

[Reet]

I honestly have never cared too much about ER plots and put much more weight into overall in-room and power response. The problem with ER estimation is it is very generalized, lacks information about your room and placement, and calling it "reflection" data is a bit misleading, since it doesn't include the reflection, but rather is a spatial average of the response that would be directed at the floor or ceiling, etc. In any case, for good measure here's the description of what these plots are from the CTA-2034 standard.

Early Reflections
The early reflections curve is an estimate of all single-bounce, first-reflections, in a typical
listening room.
• Floor Bounce: 20º, 30º, 40º down
• Ceiling Bounce: 40º, 50º, 60º up
• Front Wall Bounce: 0º, ± 10º, ± 20º, ± 30º horizontal
• Side Wall Bounces: ± 40º, ± 50º, ± 60º, ± 70º, ± 80º horizontal
• Rear Wall Bounces: 180º, ± 90º horizontal

Vertical Reflections
The “floor reflection” is defined as the spatial average of three measurements at 30 degrees
below the main-axis ± 10°. The “ceiling reflection” is defined as the spatial average of three
measurements at 50° above the main-axis ± 10°.
• Floor Reflection: - 20°, - 30°, - 40° vertical
12
ANSI/CEA-2034-A
• Ceiling Reflection: + 40°, + 50°, + 60° vertical

Horizontal Reflections
• Front: 0°, ± 10º, ± 20º, ± 30º horizontal
• Side: ± 40°, ± 50°, ± 60°, ± 70°, ± 80° horizontal
• Rear: ± 90°, ± 100°, ± 110°, ± 120°, ± 130°, ± 140°, ± 150°, ± 160°, ± 170°, 180°
horizontal, (i.e.: the horizontal part of the rear hemisphere).

[augerpro]

In-room response (per CEA2034 definition) is the orange trace. It tracks strongly with vertical responses. Horizontal responses are not affected as much as I was moving the ctc distance. Which is expected.

[augerpro]

So I had a 2-way TM proof-of-concept put on a Klippel to verify both my measurement setup, and the ABEC diffraction model, and the vertical response. Below are the ceiling and floor bounce as defined by the CEA2034 standard. If you compare it against other top speakers that have been Klippeled, I'm very happy with the results. Still not sure if this is the way to go though, the vertical sweet spot is tightish. On a side note, I was prepared for disappointment, the Klippel is heartbreaker. Kind of like running your car at Bandimere. But I couldn't be happier with the results, especially how well the waveguide integrated with the baffle.

[augerpro]

BTW this is with a 1.4x crossover wavelength ctc. Here is comparison vs the Kef LS50 META measured on the same Klippel:


: