Is it always better to have drivers in a two-way as physically close together on the baffle as possible? If I'm working from an existing design, should I maintain absolutely the spacing of the original design, or could I go so far as to trim a tweeter flange to bring its centre toward the woofer's centre?
In general, the closer the better, but just moving drivers together doesn't necessarily make a design better. You should simulate with actual measurements and see if you get an improvement by doing it.
Depends on the spacing of the original design, versus the crossover frequency wavelength and what you might gain by 1/2 - 3/4" closer.
Until the driver centre-to-centre is less than a quarter wavelength at teh XO the XO has probably been optimized for the non-ideal driver spacing in most cone+dome boxes. Even coax (usually the tweeter is physically behind the woofer in time).
dave
dave
Part of my point was that a crossover based on a response plot pair has had no such consideration since this this needs a polar optimised crossover. In such a case a person is free to take on the procedure themselves.. nothing to lose.
an allpass LCLC network I think can be used to better align coax with rear mounted compression drivers. Frazier's CAT50 has its compression driver in front of the 8 inch woofer and claims time alignment. How might the filter solution for each of these coaxial arrangements be described and estimated?
An alternative view... 1.2x CTC spacing at the XO frequency can actually help:
https://www.diyaudio.com/community/threads/new-project-tower-3-way-with-twin-8s.378223/post-7079820
and
https://www.audiosciencereview.com/forum/index.php?threads/some-help-with-lobing.22661/#post-753404
https://www.diyaudio.com/community/threads/new-project-tower-3-way-with-twin-8s.378223/post-7079820
and
https://www.audiosciencereview.com/forum/index.php?threads/some-help-with-lobing.22661/#post-753404
Here's my attempt to show the different CTC impact on Directivity index which is also impacted by vertical off axis. Please critique
The first graph takes an ideal flat frequency response driver pair that are co-incident (that is radiating from the same point in space). To make it simple, I'm using active filters. We have chosen a LR4 topology at 2,000 Hz.
The drivers are 1" diameter (i.e. tweeters). Yes I understand you wouldn't use 2 tweeters like this, but I wanted to push the natural horizontal off axis drop in SPL to above the XO point (which I would have contended with if a larger piston driver), so we only factor the lobing behaviour from CTC spacing changes.
Next we'll move Driver #2 resulting in a 125mm CTC spacing. We now have a CTC ratio of 0.73 (being target 2,000Hz / 2,752 Hz representing 1 wavelength between drivers). Because I have moved the 2nd driver away from the 1st, I have added delay on the 1st to re-align the relative phase so we maintain a perfect 6dB LR sum at the crossover.
Note though the directivty index. We get a -2.3dB drop in the power response just under the XO point (~1,868 Hz).
Also keep an eye on the vertical directivity - see we get a peak null around -25dB from reference..... remember this when looking at the next screenshot
Finally - let's increase the CtC spacing of our drivers so that our XO is 1.4x the wavelength of the CTC spacing (2,000 / 1,410 Hz).
We increase the delay to maintain LR in phase response. As you can see the DI is smoother and vertical directivity has also improved:
Now this is an extreme example, I may have got something wrong and real driver on and off-axis behaviour needs to be factored, along with real world filters.
The first graph takes an ideal flat frequency response driver pair that are co-incident (that is radiating from the same point in space). To make it simple, I'm using active filters. We have chosen a LR4 topology at 2,000 Hz.
The drivers are 1" diameter (i.e. tweeters). Yes I understand you wouldn't use 2 tweeters like this, but I wanted to push the natural horizontal off axis drop in SPL to above the XO point (which I would have contended with if a larger piston driver), so we only factor the lobing behaviour from CTC spacing changes.
Next we'll move Driver #2 resulting in a 125mm CTC spacing. We now have a CTC ratio of 0.73 (being target 2,000Hz / 2,752 Hz representing 1 wavelength between drivers). Because I have moved the 2nd driver away from the 1st, I have added delay on the 1st to re-align the relative phase so we maintain a perfect 6dB LR sum at the crossover.
Note though the directivty index. We get a -2.3dB drop in the power response just under the XO point (~1,868 Hz).
Also keep an eye on the vertical directivity - see we get a peak null around -25dB from reference..... remember this when looking at the next screenshot
Finally - let's increase the CtC spacing of our drivers so that our XO is 1.4x the wavelength of the CTC spacing (2,000 / 1,410 Hz).
We increase the delay to maintain LR in phase response. As you can see the DI is smoother and vertical directivity has also improved:
Now this is an extreme example, I may have got something wrong and real driver on and off-axis behaviour needs to be factored, along with real world filters.
Out of interest what happens if the XO moves down to 300-400 Hz (assume teh tweeter goes that low)?
dave
A good demonstration of what you set out to show.
I like another method. It works with different and even closer spacings and in this example, using your worst case 125mm spacing, it shows perfect DI.
Yeah thats right.
I see you have used some data as the tweeter beams, but what about the woofer? You can take this bit closer to reality by making ideal spinorama data with the diffraction tool to make the woofer beam as well, and include effects of baffle edge Now you'd see that also the crossover point is important to ulitize woofer beaming, which the baffle also affects some by increasing directivity around wl of baffle width.
Peak in DI is due to lobing, strong side nulls and a lot of sound cancelling out. Intensity of the lobing is maximal when both sounds are equally loud to a direction. In general, when two sounds interfere the resulting comb filter has maximal peaks and dips when they are equally loud, and less when either is louder than the other.
So, utilizing baffle size and shape and driver sizes so that tweeter directivity collapses while woofer has it high also affects the DI. Together with c-c you can tailor the DI. Just remember the main lobe reduces some in size so now listening height is more critical. Also sound toward ceiling and floor specular first reflection change, so you might want to weight all these effects with your situation and make sure the drivers are fine doing what ever the xo needs be and so on 🙂
I have no idea which sounds better. I would like to think that since listening height rarely changes smoothest DI wins, especially if you listen far away like other side of the room like many people do. But if you sit very close, then the listening height easily changes if you lean back or forward, and closer c-c would possibly win having less variation in sound. Also audible effect of early reflections are somewhat less with small stereo triangle, so strong side lobes/nulls have less impact.
What is close and what is far listening then, do you hear any of it, and do you care?
Have fun listening! please report if you do find either setup better sounding (small/big c-c), don't forget to include information about the positioning and room 🙂
Last edited:
Firstly, I said
Finally - let's increase the CtC spacing of our drivers so that our XO is 1.4x the wavelength of the CTC spacing (2,000 / 1,410 Hz).
It should have said
Finally - let's increase the CtC spacing of our drivers so that our XO is 1.27x the wavelength of the CTC spacing (2,000 / 1,571 Hz) - being a 219 CtC spacing.
Moving the XO point alone - same spacing, yields a negligible difference. Larger spacing shows a slight bump in DI
I suppose I should have applied the 1.2* wavelength driver separation rule to make it more applicable. I'll do that tonight
CTC spacing is far more critical in the near field and with MTM designs. Some of the designs are specifically intended to steer the main lobe and adjust directivity in attempt to keep power response even.
Allen, why did you centre the mic between the drivers? Wouldn't it be better to put all the Y offset on a single driver being the usual case that one or other is the measuring or design axis?
For teaching and demonstration purposes I think it's only proper to isolate the effect in question.
But surely in that case as the drivers are constantly equidistant throughout the horizontal plane there will be no distance relative phase changes like there would be when centered on one driver?
I showed power relations and DI adjustment technique. They can be adapted to real world situations. I saw no need to complicate it at this stage.