How to Succeed in Optimization Without Really Trying

How to Succeed in Optimization Without Really Trying

Here are three things I hear at least once a week:

  • “I can’t afford Smaart.”

  • “I only have speakers on sticks, not big fancy systems.”

  • “I don’t have any system DSP so I can’t do anything.”

Do you know the most common change I make to systems? Pointing the loudspeakers in the right direction. This requires nothing but eyes and ears.

Got a call today from a local theater. They have no house audio system, and for an upcoming student dance production, they were bringing in simple speaker-on-a-stick L/R mains and two sidefills for the dancers on stage. (They called because the mains were buzzing. 200 foot unbalanced TS snake returns back to the PA. I fixed that, but that’s not the point of this post.)

I never miss an opportunity to measure a system, and so I brought along my computer to take some measurements. Here’s the thing: regardless of what I discovered, there’s nothing to be done. There’s no DSP, no crossover filters, no EQ, no nothing.

The only thing I have control over in this situation is how the loudspeakers are aimed. But thousands of us deal with this exact situation every night, so let’s see if we can take any steps toward minimum variance using nothing other than the most basic of tools: speaker placement.

With a deck-stack placement of a single loudspeaker, it’s going to be louder in front. Sorry, game over. Placing them up on stands rather than right on the deck makes life a bit easier for the front-row folks, at the price of some HF roll off from being off axis. But tonal considerations take the back seat to not tearing people’s heads off, so I’ll take it.

Now all we can do is aim the thing in the horizontal plane. For even coverage in a L/R system, each speaker gets custody of half the house. Split each half down the middle, and aim the loudspeaker through it. In the vast majority of cases, this balances the over- and under-coverage better than any other aiming solution, and it’s easy to eyeball.

If you squint you can see cute little JBL three ways on the sides of the stage in front of the proscenium, which I aimed through the center-center of each half of the section, as indicated by my excellent work in mspaint.

(Ignore the underbalc delays. They’re not delayed, and so are better off unplugged.)

I took a few measurements on the on-axis line of each loudspeaker at various depths. Call it rows 3, 10, 15, and the last row. Here’s the left side:

The currently selected (bold) trace is row 3. Note the phase response not matching the others at HF since we’re so far off axis from the horn so we see some funkiness through the crossover range. Note the coherence, though (red line along the top), and then compare it to what we see in the back:

Yikes. A very poor direct to reverberant ratio indeed. (Note: these are the same four traces, I just have a different one selected so you can see the coherence for that trace.) Shame about the underbalcs. Also note that we’re at 3 dB / division here so the back seats are 12 dB lower than the front. To be expected. But actually it’s not….right?

In terms of distance, we’re at about 8 feet to the first trace and 75 feet to the last trace, so we’d expect that to be closer to 20 dB, or a little bit less because walls. The main reason it’s far less is that, in the vertical plane, the loudspeaker is aimed towards the back of the room (speaker stand raised up until HF driver is level with last row). As we move forward along the seating area, we get further off-axis to the loudspeaker, so we can play the axial level loss against the proximity level gain. It’s too far to nullify the distance drop completely, but it’s a help. The bigger advantage of this technique is revealed in spectral variance. If we offset the Smaart traces so they’re overlaid, we can ignore the level drop and focus on front to back tonal consistency:

This is something to feel relatively good about. Some folks like it loud, and they’ll sit in the front, but we can be pretty darn confident that the overall tonal character of the mix (a CD, in this case) is preserved for most of the audience. HF absorption is starting to take its toll towards the back, but it’s negligible below 8k. I’ve been in many, many venues that had a worse situation than this.

I averaged the four house left mic positions, then I mirrored the same four measurement positions on the other side of the room, averaged them, and compared the two sides.

Left and Right match within about 2 dB through the entire spectrum (extra surprising considering the speaker levels are set by somewhat imprecise pots on the back), and we have a good deal of consistency from front to back, just by placing and aiming the speakers with a bit of thought. This takes about a minute, and requires no tools other than eyes and ears. The tilt is an appropriate choice for a music playback system, and even if I had EQ I probably wouldn’t do much to it. Regardless of system scale, it all starts here. No amount of DSP can fix a loudspeaker that’s aimed the wrong way.

Smaart does not a system tech make. If you take my laptop away, I can still use my skills and experience to improve a sound system. If you take my skills and experience away, and leave me the laptop, that’s a different story.

Leave a Reply

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