I hadn't heard of this but have been looking into it and reading
@kimmosto posts about it. Always thought you just wanted them as close as possible. Indeed, seems 1.2x XO wavelength is good for DI. But that's like ~48cm for an 800Hz XO, which seems huge to me. I'm curious what the drawbacks of 1.0 vs. 1.2 vs other multiples are for a system like this with big horn and relatively low XO. JBL M2 ctc appears to be ~17" and they say the XO is 800Hz (16.9" wavelength).
And regarding the vertical nulls being close to 0-axis - don't love this but could be tolerated. I like it when you can stand up and walk around a bit without significant changes in sound.
Hi, yeah, 48cm is huge, but so is your woofer and waveguide, everything is relative
Alternatively you could just make the xo higher and keep c-c shorter. Important point here is, it's design freedom, and you can use it to your advantage as long as you understand your context (your preferences, your room, your expectations, your listening skill, your budget and so on). It's just not possible to make it point source completely, so now it's your duty to figure out what compromise suits your application better. You could also build adjustable system and try it out, because no amount of thinking and readibg replaces AB listening test in your own context to really connect the dots between theory and perceived sound. Make one speaker for one c-c, and one for another so you can AB test. You could also just decide this is too deep stuff for you and just go with what ever option you like most right now and call it a day
It's all about your context, what you value.
For example, if you like to have same sound sitting and standing there is one easy option: have bad acoustics and then stay far enough from speakers so that the early reflections completely dominate the perception. Now your sound is roughly the power response no matter whether you sit or stand. This is easy, but let's try some other example and simple math.
When sitting your ears might be at 90cm height from floor. Let's assume your ears are at 180cm when you stand for easy math, so 90cm difference between sitting and standing. If you mostly stay further than 2m from speakers, like a sofa usually is when there is big carpet between television/speakers and the sofa, then let's see how big of vertical coverage angle is needed to get same sound at 90cm and at 180cm height (anechoic):
- assuming speaker height is set so that 0-axis it's above ear level when sitting and below ear level when standing, at 135cm height: right angle triangle calc with sides of 2m and 45cm gives roughly 13deg off-axis, sitting or standing.
- Since dips are not that audible, depending on your listening skill, you might consider perhaps 1dB dip due to lobing is acceptable, or even up to 6dB, perhaps even more! Let's take the 1dB for golden ears.
Now, let's do simple experiment in VituixCAD using ideal drivers with flat response, LR4 at 800Hz to find out how big of a c-c we can have before dip we have at ~13deg is about 1dB.
Setup looks like this.
Then, since the polar line chart doesn't have 13deg line available, I use polar map here and move mouse cursor to ~13deg and see how much attenuation there is. Then I adjust Y-coordinates of the sound sources to find c-c where the attenuation is the arbitrary chosen -1dB at xo. Here I arrived to about 24cm c-c where this happens. Now, you likely cannot do this low c-c with 15" 9 and ~15" waveguide on top, you'd be better of with ~24cm driver and max 24cm waveguide, set touching each other, to get the 24cm spacing we arrived with this tesr.
Anyway, this test wasn't ideal though, it assumed your speaker is quite high in air, above sitting level, and you'd have no optimal sound anywhere, because optimization was "the same for sitting and standing", -1db dip in both cases at the xo.
What you could do instead is to question more about yourself and about your context there, and take some tests. For example find out what kind of a dip is too much to you at 800Hz? You could test this with some EQ app/tool using headphones. Then, what is your room acoustics like? Would you still hear this dip if power response was ok anyway? I mean, when you are standing then are you actually hearing (dip in) the direct sound or does the room sound dominate instead? You could also experiment how close the sources would actually need to be to have 0-axisbat sitting height, whilenstanding gives only 1dB dip at 26deg off-axis. Or ehatever your listening distance actually is.
After questioning yourself and your application you'd get real data what to aim for with your system.
Perhaps you'd like smooth power response because when you are standing you are not at the listening spot. And, when you actually are at listening spot, then direct sound there should be flat. Not let's see how to optimize this goal:
- sitting it would be always flat with following assumptions: you can freely set speaker height to your design height and you have ability to EQ this axis flat. In this case no matter how narrow the vertical "lobe" is, like +-5deg or what ever, you'd still get flat direct sound because you set it so.*
- Now, standing, only power response matters so you'd optimize power response as well. Here optimal would be a point source, but since you cannot get the horn that close physically, you could test the as close as possible, and compare that to some wider c-c setup of itnmakes difference (kimmosto 1.2wl stuff).
- Assume 15" driver right on top edge of a box, so 7.5" to top edge. Then 4" compression driver on top, so about ~9.5" c-c is about the absolute best case scenario you could actually build, ~24cm.
Now let's build simple test setup.
Here is the power response using ideal drivers, 800Hz LR4 filter, and you could use anything you like to test this stuff out. Anyway, about 2.5dB dip in power here, while on-axis is flat. Can we get the power better if c-c was increased?
Just use mouse wheel to adjust y-coordinate of woofer for example and here in this case increasing the c-c all the way up to 46cm has halved the dip in power response. On-axis is still flat so now we have optimized this particular system for this particular goal by increasing c-c.
I could have got to same result keeping the 24cm c-c by adjusting xo frequency up instead:
Home exersize using VituixCAD ideal drivers, and do some actual listening tests your current system what ever you have there atm.:
- Can you make ~flat or smooth power response in any way other than having coaxial setup? with some other filters perhaps?
- If not, what's next best in this respect? What is actually doable with passive XO, or DSP you have?
-* why some examples above don't actually have exactly flat response, while some do? What this hint's about good design axis and what trade-offs it comes with?
- Is this even necessary, how big of a power dip is benign to you perceptually in your context? Is it the same "audibility" for ~few db dip at 800Hz or 1600Hz?
- Is smooth more important than flat on axis or can you just boost at the xo
- How wide is your high frequency coverage, does that alone limit your good sound regardless of xo?
- and so on.
Also, read kimmosto messages about c-c, and what he writes about context with it, like directivity of drivers.
Well, getting out of hand, out of topic
Point is, there are million things in a speaker system, preferences and room acoustics, listening skill varies. 1/4wl xo is possible only in few special scenarios where you specifically design to achieve that. A woofer and separate tweeter on top is not this special situation.
Main motivation for this post is for people to think and test this stuff, understand their own application, the context, and whether some rules heard on internet are valid in your context, or not, perhaps they don't even matter. What I've said in this thread might be relevant to you, to your context, or not. Understanding your own context is more important than anything you read about audio, because even if something is best for somebody else it might be exactly wrong thing for you. Minimal c-c could be good in your context, or 1.2wl might be better, perhaps you are not able to notice difference most of the time so it's not even relevant question and purely visual or practical thing. Perhaps you neglected edge diffraction and there is no flat / smooth sound available regardless of this stuff.
Well, remember to have fun with your projects!
