Smaart Beta: Will the new filter control in the delay finder help with your main+sub alignment?

The overhauled delay finder in Smaart v8 beta looks like it’s set up to make your main+sub alignments a synch. But will it really save you time?

(Yes, but you can’t throw out the phase graph, yet.)

Key Takeaways

The updated delay finder includes an optional bandpass filter for tracking arrival time by frequency.It may save you some time, but you’ll still need to verify alignment with the phase graph.Results are highly variable based on SNR.Let’s look at how it works.

The new delay finder reminds me of a process I learned at previous Rational Acoustics trainings for observing spectral crossover alignment using the Impulse Response module. It goes like this:

Measure an IR of your main.Observe the ETC graph filtered at the crossover frequency.Set delay at peak.Measure an IR of your sub.Observe the ETC graph filtered at the crossover frequency.Find peak.Time offset = main peak – sub peak

I rarely used it in the field because it seemed slower than other methods, but now these features have been incorporated into the overhauled delay finder making them more accessible.

Your new alignment process might go like this:

Find delay of main.Filter to crossover frequency.Insert delay.Store trace.Find delta delay of sub.Adjust delay line or physical position.Verify phase alignment.Verify summation.Let’s look at this on a real project.

I have a recording of an array of dB Technologies DVA T8 and an S30N. I measured them each solo to find the crossover frequency.

I chose 96Hz because it is in the center of the crossover region, has matched magnitude, and high coherence.

[image error]

In the delay finder I measured main solo again, this time applying a 1/3-octave bandpass filter at 96Hz. At first, I was getting a different result every time I clicked find. This is the kind of behavior I would normally expect from the delay finder trying to measure a sub. For more, read this article from Merlijn van Veen and this one from Bob McCarthy.

I was about to give up when I remembered that you can customize the delay locator FFT size and averages. In the Smaart manual I read:

The default is a 64K FFT with no averaging, which works out to a time constant of 1365 ms at 48k sampling rate. This is sufficient for finding delay times at distances up to a about 450 hundred feet (140 meters) from a source – a good rule of thumb is that the FFT time constant should be least 3x greater than the expected delay time.

I’m only measuring delays of less than 100ms. If I follow the rule of thumb, I should set my delay time to 300ms, which would be an FFT size of about 16k at 48kHz. I tried the delay finder again with the new FFT size. Still squirrely. I tried increasing the number of averages up to 8. Still no.

But then, when I set the FFT size back to 64k, all of the sudden the results were more consistent. Then I started reducing the averages until the results got squirrely again, finally settling on 5.

[image error]

Unfortunately, with an FFT size of 64k and 5 averages, it takes about 16 seconds to complete, which is an eternity in production time. Here’s my measurement.

[image error]

Now I’ll measure the sub solo with the same delay locator settings.

[image error]

Looks like I am not within 60º through the crossover region, so let’s see if the new delay locator can help me out.

[image error]

The delta delay says that the sub is arriving 8.39ms early so I’ll pop 8.39ms into the delay line and store a trace.

[image error]

The phase traces are definitely closer, but I wonder if I can do even better?

At 96Hz the phase measurements are 109º apart. 109º is close to 120º and I know that 120º is ⅓ of 360º. The period of 96Hz is 10.4ms (1/F). 1/3 of 10.4ms is 3.12ms. I’ll subtract 3.12ms from my delay line for a total of 5.28ms for this alignment:

[image error]

How close am I to perfect summation? I’ll load the measurements into Phase Invaders to find out.