I had a really
long post all typed out and then accidentally deleted it. The gods of brevity sending me a message
My goal here is to use my current midwoofers (TD15m) from 80-1000ish and have controlled directivity from 300hz up. In my room I'd like to have a 60° beamwidth and I'd make elliptical os waveguides to match, but I won't settle for anything more than 90°.
There's really three ways that I see to do this: dipole, cardioid(or variations thereof), or some big horns. I don't like dipole because that's a 120° pattern and I don't like the rear energy bouncing off the front wall....been there and done that. Horns will be another project. So that leaves cardioid. Face and RyanCo have passive resistance boxes for their midbass. The rear wave of the driver is down in level and delayed to narrow the pattern and create a null at the rear. I suspect that there isn't a perfect null because the damping needed varies with frequency but that's a close enough approximation. IIRC Face said his box had issues over 4-500hz, and this is something I've seen with other passive implementations online.
Being the pc dsp dork that I am I thought I would try an active cardioid using smaller drivers on the rear or sides of the cab. Linkwitz says a cardioid can be made with 2 monopoles back to back, with the rear driver out of phase and delayed the time it takes to reach the front baffle plane. I'm using some junk 2cu ft sealed cabs, old (early 80's...older than me!) 15" Eminence woofers and 8" Vifa hifi drivers. The first implementation was a single 8" on the back. I tried different delay settings, levels, as well as hi and low pass on the 8" but could only get good results up to 400hz. Above that things get goofy. Sorry didn't save the data.
I wasn't surprised considering the cab is 15" deep which is more than a wl at 1khz where I want to xo.....so I put an 8" on each side of the cab about 3.5" from the front baffle plane. I ended up delaying the main 15" by .25ms (which I thought was odd). The 8's are out of phase, -9dB in my dsp and have a first order low pass at 700hz. Here's 0-90°:
And a polar plot:
Sonogram? Why not:
Not bad! Right around 90° or less from 300 up. As you can see it's really only cardioid at 1khz with a good bit of rearward energy as you go down in freq. I'd like to see some better rear attenuation, and that 90° measurement is funny. These cabs already had a 12"x12" hole in the back that I covered with mdf, so I opened it up and stuffed them with bonded logic:
Here you can see that it didn't really affect the forward response, but attenuated the rear a good bit and tends toward a hyper-cardioid. Nice.
I did a lot of playing around with this. I could get a 60° beamwidth by shaping the side drivers with hi and low shelving filters, but it would create some serious lobing at 90° and beyond. Ideally I would have a driver on the back as well. That would need a channel of it's own for a different delay and that's getting too complex. We're talking 5 channels at that point for a 2.5 way speaker
I've got enough data at this point to feel comfortable moving forward with this. Any ideas on suitable 8" drivers for the directivity control? They're down quite a ways in level, and since they aren't really contributing to the direct sound they don't need to be anything special. I'm thinking the Dayton PA's. Also, any thoughts on the closed vs. open back? I'd like more attenuation to the rear but the open back does have more lf roll-off.