If you lay awake at night thinking about improving your crossover alignments and winning big at Phase Invaders, then this guide is for you.
What follows are three proven battle tactics from beginner to advanced.
#1 – Beginner
Press all the buttons until you win.
a monkey hitting keys at random on a typewriter keyboard for an infinite amount of time will almost surely type any given text, such as the complete works of William Shakespeare.Wikipedia
I have to thank my wife for this one. She is often my first beta tester and while not a professional audio engineer, she is driven by a competitive desire to win unlike most people I’ve met. While making random combinations of the delay slider and polarity switch, she keeps track of her score and returns to the combination that returned the highest score.
#2 – Intermediate
Make the pictures match.
The battle tactic I think most audio engineers will start with is the visual one. Naturally, we are all visual learners. One good picture is worth a thousand words. It’s one thing to talk about speaker coverage, but the first time you see its prediction in your modeling software or do your first speaker autopsy, a new level of understanding is reached.
In Phase Invaders, you’ll want to use coherence blanking to remove any noise, zoom into the crossover region, then adjust delay and polarity until the Sum matches the Target. Use the score to fine-tune.
Pro battle tactic: Click and drag to zoom in on the graph. Double click to zoom out.
#3 – Advanced
Find optimum alignment through the crossover region using the phase graph and phase delay formula. (scary!)
Finding alignment on the phase graph may be as easy as sliding the delay around a bit until the pictures match, but many times the data is so hard to read that it can be difficult to tell if you are a half or full rotation away from a better result.
Here are the steps:
- Pick a frequency (f) that is near the center of the crossover region, has near matching amplitude on main and sub, and relatively high coherence.
- Use the phase delay formula to calculate the delay needed to align main and sub at one frequency. (((Main Phase/360)(1000/f))*-1)-(((Sub Phase/360)(1000/f))*-1). Pop in the four variables and you can paste it directly into Google. As far as I can tell, you can simplify this formula by removing the parenthesis and it still works, but I gave you both just in case. (Main Phase/360*-1000/f)-(Sub Phase/360*-1000/f). For more on this formula read this article.
- Use the result as your delay in Phase Invaders. If the phase traces are aligned, Sum is on top of Target, and you have a big fat score, you’re done. If not, try something else. Add or subtract delay to rotate the phase at f by 180º+pol. inv. or 360º. Use 500/f and 1000/f, respectively. Keep going until you find the combination with best alignment, summation, and score.
Pro battle tactic: The phase delay and time period are calculated for you in the cursor read out at the top of the screen.
Let’s work through it together. Here’s a four-mic average of a measurement I took of a Martin CDD-Live 12 (main) and an SXP118 (sub). The main is 18ft directly above the sub.
The first thing I’ll do is adjust the coherence blanking to get rid of some of the noise.
Now I’ll zoom into the area of interaction and pick a frequency. I’m going to pick 105Hz because it’s near the center of the crossover region, has matching magnitude, and high relative coherence.
If you are already familiar with the phase graph then this probably looks like a polarity inversion. If you’re not, you might look at the cursor readout and see that it says 184.72 °Δ.
(I know the font is hard to read. It seemed clever at first.)
I’ll insert a polarity inversion and we get a near perfect score of 9947396.
This is a pretty clear cut case, but let’s try a couple of other options.
Earlier, with the cursor on 105Hz, we saw that the delta delay was 184.72º and the phase delay was -4.87ms. I’ll take out the polarity inversion and try -4.87ms in the delay.
Right away I can see that while we are aligned at 105Hz, the phase slopes do not match, the sum trace is not on top of the target trace through the entire crossover region, and the score of 9719481 is lower than before.
Our alignment was not improved by going in this direction, but now that we are aligned at one frequency we can easily test other possibilities.
In the cursor readout I can see that the period of 105Hz is 9.48ms and half of that is 4.74. Let’s try 9.48ms.
Our current delay setting is -4.89ms. To get a 360º rotation: -4.89+9.48=4.59ms.
Now we have better alignment of slopes, better summation, and a score of 9913456. That’s still not better than our first score.
Let’s test one more. This time half a rotation at 105Hz with a polarity inversion. 4.59+4.74=9.33ms
This gives us worse alignment and a score of 9619582.
Through efficiently testing a handful of options we have discovered the option with maximum summation and improved our relationship with the phase graph.
Pro battle tactic: Have an empty text document open to keep track of your scores and settings.
Have you tried Phase Invaders? What’s your favorite battle tactic?