Many of the bad optimization decisions I have seen were made because folks didn’t correctly understand what the analyzer was telling them. Here we have a collection of data from Smaart along with context for each one. Your mission, should you choose to accept it, is to figure out what the analyzer data is indicating, and what you should do in each situation.
1 – System DSP Bench Test
Here’s a brand new, out of the box system processor being bench tested before an install. The unit has been freshly factory reset. What’s the problem?
Answer: (click to expand)
There isn’t one. Note the drop in coherence and unstable magnitude data at HF. The analyzer’s delay has not been set to account for the processor’s latency. The phase trace goes through one wrap from DC to 1 kHz, so the latency = one cycle in 1,000 Hz = 1 ms. Correct delay setting:
2 – “Did No One Check This First?”
This is a loudspeaker processor, factory reset and in bypass. What’s the problem?
Answer (click to expand):
The phase trace is around 180° for all frequencies. No time delay can cause this. This unit left the factory with an internal polarity inversion between the inputs and the outputs.
The phase trace is around 180° for all frequencies. No time delay can cause this. This unit left the factory with an internal polarity inversion between the inputs and the outputs.
3 – Rehearsal Space PA
We are measuring variance over the space in a rehearsal room with a flown PA by taking measurements from multiple mic positions. The second measurement (teal) yields a crazy phase response. What is going on?
Answer: (click to expand)
We moved the mic but forgot to reset the analyzer’s delay. Second question: Based on the change in phase response, has the mic moved loser to, or further from, the PA system?
We moved the mic but forgot to reset the analyzer’s delay. Second question: Based on the change in phase response, has the mic moved loser to, or further from, the PA system?
Answer 2 (click to expand)
Closer. The upward slant from left to right indicates early arrival.
Closer. The upward slant from left to right indicates early arrival.
4 – Gradient Sub Array Alignment
We are aligning an inline gradient sub array (using delay and polarity inversion to cancel the rear-going radiation). Since this array is tuned from the rear, we place our mic behind and measure the response of each subwoofer before any DSP.
Which trace is the front sub? Which one is the rear sub? How much delay do we need, and on which?
Answer: (click to expand)
The blue trace is higher in magnitude (louder, so closer to the mic) and has a less steep phase trace (less time offset, closer to the mic). So that’s the rear sub, and it needs to be delayed to be in time with the arrival from the front sub (black trace). How much? The offset at 100 Hz is 120°. A third of a cycle at 100 Hz is 3.33 ms.
The blue trace is higher in magnitude (louder, so closer to the mic) and has a less steep phase trace (less time offset, closer to the mic). So that’s the rear sub, and it needs to be delayed to be in time with the arrival from the front sub (black trace). How much? The offset at 100 Hz is 120°. A third of a cycle at 100 Hz is 3.33 ms.
That’s perfect to align the arrivals:
Add a polarity inversion and we’re set. Note how phase is now offset 180° at all frequencies:
5 – Corporate Event Quick Tuning
We’ve been given only a few minutes to rough in a quick tuning before a corporate event. We stick our mic out in the room and take a measurement. The curve looks reasonable for a speech system, but we should probably EQ out that 900 Hz bump. Good idea or bad idea?
Answer: (click to expand)
Note the regularly (log) spaced pattern of peaks and dips in the coherence trace, which matches the regularly (log) spaced pattern of peaks and dips in the magnitude response, which matches the regularly spaced wiggles in the phase trace. We’re measuring with both sides of the PA on. This is a comb filter. Don’t touch that EQ! Mute one side and try again.
Note the regularly (log) spaced pattern of peaks and dips in the coherence trace, which matches the regularly (log) spaced pattern of peaks and dips in the magnitude response, which matches the regularly spaced wiggles in the phase trace. We’re measuring with both sides of the PA on. This is a comb filter. Don’t touch that EQ! Mute one side and try again.