Active system time alignment

I've been experimenting with time alignment with an active 3 way speaker and DCX. Dome tweeter + 6.5" mid in a sealed box with an angled front baffle sitting on top of a 12" Rythmik sub each side.

Using DCX auto align I get interesting delay settings! Run the routine twice and it does things slightly differently. One time it might invert polarity on a particular channel, the next time it might not, but compensate with delay differently. This is without any other changes.

Here is a typical setting:

Mid and top both delayed by 20 ms (7.8m), both with inverted polarity, then the mid has 718mm and the top 672mm delay.

Another try:

No inverted polarity but the same delay to the mid + top, 300mm to the mid and 338 to the tweeter.

Just curious what kind of settings others are finding.

As I don't fully trust the auto settings, I'm going to measure for myself when I get the time.

Without any delay settings in place, the tweeter needs polarity reversal.


2009-11-12 12:56 am
Haven't used auto-align for several years, but I do remember erratic behavior if the acoustic levels were not high enough. Perhaps crank it up a bit and see if results become more consistent.

I think that just might be it. I've only had slight differences in outcome and always used the auto alignment with enough volume.

I think it's best to run auto once with just polarity, once with only delay, and once with both and then decide which one to use.
In the manual, Behringer himself says that Auto-Align works in most of the cases...only ! This is targeting the PA world for fast settings only.

So, the best is to forget this and get familiar with a good (free...) soft, as Arta, that displays clearly impulse and step. Learning this tool is a long curve but you gonna get paid back.

feel free to mail me if you want more details, as how to get this...:scratch:


  • step.jpg
    7.7 KB · Views: 181
Were I to do it (never did before), I would do a bit of basic trig and get the hypotenuses of the listening triangles between listening position and each driver, pick a reference, and then subtract horizontal distances from the hypotenuses to yield traveling distance difference and then just divide by c.

Tweak after that, but that should get you pretty close.

My 2 cents