Active Speaker - EQ full band or individual drivers? (or both?)

[micfre]

I've got a three-way active setup, driven by a MiniDSP 4x10HD. It has 6-band parametric EQ for each driver channel, plus a system-wide/full band EQ stage.

I have tried to individually measure and EQ each driver channel, with the assumption that if each range is EQ'd to flat (or my desired house curve) and summed with the other two drivers, the result will be flat.

I have also tried, and ultimately preferred, leaving the driver channels untouched and EQ'ing at the system level. To me I got a better result in terms of overall match to the target curve.

Anyone have experience or advice on this front? What do you do?

Thanks!

 

[boris81]

I guess the general rule is to EQ driver issues on the channel level and system issues on the global level. Both approaches work but EQing drivers individually runs the risk of misaligning the system phase. I usually follow along these steps:

1. Measure the individual drivers on the baffle (use HP to protect tweeters)
2. Decide crossover points and draw your ideal target curves
3. EQ individual drivers to match the target curves
4. Measure and fiddle around with each crossover point until you arrive at an acceptable frequency response summation and phase tracking.
5. EQ the system response to iron out inconsistencies or to set a preferred tonal balance.

 

[CharlieLaub]

You may not realize it, but when you apply EQ to change the amplitude you also change the phase. When you apply EQ to a driver, you can change its phase relative to other drivers around the crossover point. This depends both on the proximity of the EQ center frequency to the crossover frequency and the Q and amount of boost/cut. For these reasons, applying EQ to the input is better. The only possible exception is when the EQ band is not located near the crossover point and is high in Q and low in boost/cut - but it's difficult to put an absolute limit on these in a general sense.

I typically do a couple of revisions of the crossover filters and then go back and flatten the response using EQ at the input. Done in this way, the EQ will not effect the crossover filter responses that you have already optimized.