I'm concerned that because they use totally different drivers and horns and because the 'high' input on the 3020HT still uses part of the internal
crossover in the 3020HT (, then the MF1-X in parallel with the 3020HT high input along with acoustic interactions between the horns is likely at least part of the problem (and also likely presenting a low load to the amp).
The 3020HT is actually a four-way box that can be run full range or tri-amped. When run full range the is a 6dB/octave low pass filter at 400Hz for the lows (the dual 15"). Then a matching 6db/octave, 400Hz high pass filter and a 12dB/octave, 2.2kHz low pass filter for the mids (dual 10" drivers). The 'high' section actually has two components, a
compression driver on a
horn and a 'super
tweeter'. The
compression driver an
horn gets an 18dB/octave high pass at 2.2kHz, however there is also an 18dB/octave low pass at 8kHz that is part of the internal
crossover and is in the loop even when tri-amping. There is also a 24dB/octave high pass at 8kHz for the 'super
tweeter' that is part of the internal
crossover and always in the signal path.
So the 8kHz 18dB/octave low pass and 24dB/octave high pass internal
crossover is always in the signal path even when tri-amping. When you
wire the MF1-X in parallel, which should have a 600Hz, 12db/octave high pass, you have added another complex device in parallel, which probably messes with the internal
crossover. That could certainly be part of problem. Adding to this, with the MF1-X in parallel you have two devices with different ranges, different impedances and different sensitivities in parallel, so the resulting combination cause variations in the output of the different devices over frequency.
Finally, with the MF1-X overlapping in frequency with two devices in the 3020HT and being separated from them by some phsyical distance, you are also likely getting some relative
phase interactions between the sources. The
effect of summation and
combfiltering might audibly affect the response as you change the
crossover point.