All of our gear is full processed to provide a very level and consistent sound across the frequency spectrum there are designed to cover. Because of this, our gear illuminates the quality or lack of track quality in a very apparent way. Considering the HIFI quality of output and tuning involved with our gear, we advise no processing be applied to the channels supplying signal to the BASSBOSS gear. Processing can affect reliability and subsequently may void the warranty. Additionally, I am confident user applied tuning applied to our gear would degrade the sound quality.
Variation between different tracks is typically a clear indication of a quality issue of the tracks themselves. 320Kbps MP3s are the minimum quality that I will suggest using. AIFF and WAV even higher quality (lossless) options as well. Some record pools and bootleg media can even have doubly compressed MP3s. I have also seen folks edit or remix an MP3 and export the mix as an MP3, resulting in double compression. Even doubly compress 320s are going to greatly suffer in the sound quality and dynamics department. Best practice is to export to WAV to avoid a second layer of compression.
As for the best practice on gear setup:
Generally speaking, when mixing brands of speakers you could be encountering phase and time alignment issues that may be impacting the quality of sound. Also room reflections, nulling, comb filtering could also results in output diminishing cancellations.
I would check and adjust the LF roll off of the tops and the HF roll off the subs (EXO knob) to see if the output improves. You may have a gap between the subs and tops in the 80 to 100Hz area, which can be more apparent on certain tracks.
Additionally, moving the sub itself can improve issues caused by reflections and cancellations. Cancellations are more prevalent in left/right positioning, which lends itself to comb filtering. Sometimes a little bit of repositioning can make a dramatic improvement.
Subs clustered together in the center typically yields the best result, most output, and little to no comb filtering.