Hi all. I own A3PX speakers and a pair of Rythmic F15 subs. My CD collection is ripped to a laptop and played via JRiver whose DSP functionality is used for EQ and to play a FIR (Finite Impulse Response) convolution file generated by Audiolense XO software to correct frequency response, time alignment, and crossover duties. Setup steps to help the sub newbies (e.g. DrBond) were as follows:
> My Prerequisites: multi-channel DAC (e.g. 8ch.), acoustically treated room and acoustical measurement capabilities for which I use the OmniMic for general freq response and decay time measurements and Audiolense for FIR filter generation
> identify all possible subwoofer candidate locations in my room and put a piece of painters tape with a letter by each location
> place sub at a candidate location and take about 15 frequency sweep average measurement and save the file with the candidate Letter in the file name
> repeat for all possible sub candidate locations.
> compare each candidate location to identify the best spots for my two subs where “best” is based on the following criteria: (1) loudest SPL at lowest frequency, (2) smoothest frequency response that is basically a straight horizontal line or tilted downward as frequency rises (i.e. 20hz is louder than 100hz), and (3) if there are any nulls that they occur higher up in frequency so that they can be bypassed via the crossover or filled in by the mains
> with subs in their new spots for best frequency response, I turn my focus on time alignment. Changing software to Audiolense I run Impulse Response sweeps and note the relative delay times that it generates. I want the sound from both mains and both subs to hit my ears simultaneously to avoid “slow” bass that is half-a-step or more behind the beat of the music. Take the longest delay time and subtract from it each of the speaker’s and sub’s delay time and enter them in JRiver’s DSP area where each channel can be optimized because of the multichannel DAC. For example, say sub #1 is way behind me and off to one side and is 8ms delayed while the front mains are delayed 2ms and sub #2 is 3ms delayed. Delay values for JRiver would be Mains 6ms (8-2=6), Sub #2 would be 5ms (8-3=5) and sub 1 that is farthest away is 0ms (8-8=0).
> now with subs in the best spots for frequency response and time aligned with the mains, I can focus on the crossover. I’ve learned not to assume that crossovers should be the standard 80Hz/24dB filter slope, in fact sometimes staggered or overlapping crossover frequencies is best based on room and speaker/sub locations within it. I was driving the SL speakers with a tube amp so wanted to relieve the tube amp of the low bass notes, so typically my high-pass crossover was between 40-50Hz for the mains. Sub’s low-pass crossover frequencies were where the measured frequency response with the closest main was the smoothest and I’d have to play with various frequency and slope combinations which in real-time with OmniMic to view the changes is a piece of cake and fun to do. I found that 1st and 2nd order slopes (6dB/12dB per octave) on the subs blended best generally speaking especially when the subs were farther rather than closer to the SL mains. Having a sub far from a main speaker and using a 24dB slope can create a ping-pong effect where you hear a bass note from the main then the sub as a bass plays a scale or run down (or up) the notes. Feathering in the sub’s bass with the mains to avoid this ping-pong effect is a combination of things: 1st or 2nd order slopes which require a lower crossover frequency so the subs aren’t localized by playing higher frequencies, and volume adjustment of the sub.
> after identifying the crossover settings for subs and mains I enter the info into Audiolense to take another impulse response measurement that’s converted via FFT into a frequency response. I create a target curve and Audiolense then does its magic to fit the measured frequency response to the target curve using the time delay values and crossover settings from above.
The above is not for the feint of heart or newbies as it involves a long learning curve about several factors, but it is the best way I know of for optimizing sound quality and can be done in other tools like DIRAC or Anthem’s ARC software. I believe that my approach is more flexible.
For those that don’t have DIRAC/Anthem, or Audiolense, then I’d suggest at a minimum using JRiver or ROON with a 2channel DAC and an acoustical measurement toolset (OmniMic or REW for example) so you can time align things and use parametric EQ to cut peaks for smoother frequency response. All of the other tools and steps above just take this to the next level of performance…..