Modelling driver pointed in multiple directions in passive crossover designer

[Rijaak]

How do I use horizontal panning? I'm designing a system in which the tweeters are angled 45 degrees from the woofers so that the tweeters are pointing directly at the listener when the woofers are aimed 45 degrees outward. How would I model this in passive crossover designer?

 

[DavidL]

If you have a mic and measurement software, just aim the mic at the tweeter on axis, leave it there and measure both tweeter and woofer response one at a time and model your system from that.

 

[AllenB]

This would be a fairly involved design. I'm sure you could use PCD for some of the hard parts but it doesn't look like it could do the design for you, you'd have to already know what you wanted the crossover to be doing.

 

[Rijaak]

I think I figured it out. I input the original .frd file for a single woofer then set horizontal panning to 45 degrees. I then printscreened and used spl trace on the graph to create a new .frd file compensated for the roll off in db due to the 45 degree angle. Does that sound like it would work?

 

[AllenB]

I don't think so. Did you use the 'Driver horizontal offset (x)' box?

 

[Rijaak]

Not when I made the new .frd file. The horizontal panning simulates listening from different parts of the room. I used horizontal and vertical offsets on the woofers after I input the .frd file compensated for the angle.

 

[AllenB]

It should work. I wouldn't trust it inti the breakup region though. Up here the response you began with would probably not be consistent with the total ouput.

 

[Rijaak]

Good thing the break up is above the crossover point then huh?

 

[aczern]

LspCAD

 

[AllenB]

If you judge the breakup based on the on-axis measurement, you may miss a few things. Like in this example. From the on-axis plot it appears that the breakup is at 2kHz, but if you look at the bigger picture breakup appears to begin below 500Hz. It doesn't become a problem until after 1kHz, but the on-axis frequency response is inaccurate between 500-1k.