Shine On, You Crazy Diamond

Shine On, You Crazy Diamond

A colleague’s upcoming gig posed an interesting set of constraints for the subwoofer subsystem design. It’s a diamond-shaped stage and due to the nature of the event, all the boxes must be located against the stage, not spaced out in a more typical broadside or L/R configuration. (Seeing as there’s a drape/video line about midstage, it would be a nice additional goal to steer energy away from the front wall of the venue if we can.) This presents some problems.

Stage close-up, plan view

All predictions here are MAPP XT, 1/3 oct, at 30, 60 and 100 Hz, instead of my usual 125 Hz.

Any basic configuration (read: no beamsteering) will result more LF energy on stage than any other point in the room. That’s because the subs will all arrive at center stage about the same time and create a huge pileup. It’s Power Alley on steroids, and it’s 12 dB above the rest of the room. Yikes.

2x Broadside arrays

Wait, I know, let’s invert polarity on one side, so we can turn the alley into a cancellation.

Broadside array w/ inversion one side

You may have heard of the Aardvark sub technique. Well I’m calling this the Butterfly. Arrival times are too staggered to do much in the way of cancellation. And the length of the lines means very narrow beam lobes with a huge center gap, with or without the polarity inversion. And it’s still really loud on stage and nowhere else, which is the opposite of what we want.

Let’s try to do some steering. We’ll isolate one line and add 1ms of delay per box as we approach the audience in an effort to bend the beam out into the house.

Broadside with delay taper, 1ms / box

That sort of works. We are steering a little bit but it’s still too narrow due to the length, and as it’s axisymmetric, we’re still putting just as much energy on stage as we are out the other side.

So let’s look instead at making each element a 2-box cardioid array.

Cardioid Broadside

Well, it does keep the stage clean as it is designed to do. But we still have a center gap and narrow beams, which translates into really terrible uniformity in the house. How about combining that with the delay taper?

Cardioid Broadside with delay taper

That’s not a rig I’d want to have to tune. It works great at 30 Hz and stinks above that.

The theme here is that we have a long line perpendicular to the throw direction, which is creating a not-very-useful beam narrowing effect. So let’s change the parameters. Let’s try to use each sub array to cover the opposite side of the room.

A single endfire:

Four element endfire

Now our array is only one box wide in the direction of interest, which allows us a nice wide cardioid shape. However, the two sides don’t play nice together.

“La Verga” Dual Endfire

Aside from being overtly phallic at 63 Hz, this isn’t…terrible. It’s not good either. It’s still loud on stage at 30 Hz, and we still have some significant variance in the diagonals of the audience. What about if we try inverting one side again?

Dual Endfire w/ Inversion

This would honestly be awesome, if we could just get rid of the middle third of the audience and move FOH to the side wall. Instead of “obstructed view” tickets, you could sell “obstructed sub” tickets.

As we know from vector summation math, cancellations are only deep when the levels are very close. If either source has a significant level advantage, the null is not nearly as deep. So if I use EQ to make an alternating series of boosts and cuts throughout the sub range, and do the opposite on the other side, this should give one side or the other a level advantage at a given frequency, and thus reduce the depth of the cancellation. How sneaky!

You’re going to have to ride the train of thought all the way to the station, I’m afraid. The Meyer DSP gives you ten PEQ filters per channel, so we need to figure out how to evenly (on a log scale) space 10 filters between 30 and 100 Hz. That’s a frequency ratio of 3.33:1. So the question becomes “what number, multiplied by itself 9 times, will give an answer of 3.33?” What we want to know is the 9th root of 3.33, which is equivalent to 3.33^(1/9). Put another way, 30 x N9 = 100 should give the spacing factor for the filters. If you’re good with your logarithm properties, you will end up with about N = 1.14, which gives filter center frequencies of 30, 34, 39, 44, 51, 58, 65, 75, 86, and 99 Hz.

One array looks like this, and the other is the complement (brown trace = filter phase response):

And the crazy thing is, it sort of actually works:

It’s definitely better than it was. But it still stinks. The Dual Endfire array just isn’t to be. Back to the drawing board.

Well, wait a second. We’re so programmed to think about symmetry in our rigs but that’s what’s causing the trouble here. What if we break that mental constraint? That single endfire array was really nice and wide. Let’s take another look at that.

Single 5 element endfire

This produces seriously good coverage for most of the audience and avoids painting the upstage wall. It’s still hot on stage but it’s definitely a step in the right direction. Now if we can just fill the gap on house right. I want something relatively narrow so as not to interfere too much with the main array coverage. How about a three-box broadside array.

5 element endfire with 3 element broadside fill

This is the closest we’ve seen to uniformity so far. The entire main endfire array is timed two ms later than the fill array so as not to disrupt the steering too much (the fill array in a sense acts as the first element in the endfire). It’s starting to come apart at 100 Hz but we’re getting warmer, and most of the audience is covered quite well in the bottom octaves. But we still have a firestorm on stage. Swapping the fill array to a gradient cardioid array didn’t do much better.

But let’s discard one more assumption: that all our loudspeakers must be in front of the drape / video line. It’s a sub. It doesn’t care. And so this monstrosity is born:

5 element endfire with 3 element polarity inverted endfire fill

The 3-element endfire fill array is polarity inverted and timed a net 4 ms behind the main array to steer a bit of cancellation onto the stage. With the exception of a single notch shooting into the corner, the audience area is actually covered pretty well.

And just to put a cherry on top, since the fill array is doing more harm than good at 30 Hz, let’s high-pass it at 50 Hz. 

Five element endfire with 3 element polarity inverted delayed highpassed endfire fill

This puts a little bit more 30 Hz energy on stage but that’s a trade I might be willing to take, especially if it’s a playback / DJ show without a ton of live inputs going on. This is very even coverage for such a goofy setup.

For comparison, in my opinion this is way less scary than the lobing from a regular old cardioid left-right array (taken from this post):

And there you have it! I thought the IR would be an absolute disaster but MAPP says otherwise. If you want to take the time to cable, verify, and tune this bad boy, you’ll be in great shape. And I’m relatively certain nothing like this has been done before, so that’s cool too.


Leave a Reply

Your email address will not be published. Required fields are marked *