Another thought. Maybe we have two issues at work here.
The BA4560 (I had to look up the RC4560 opamp to get this) has a high drive capability, twice that of the LM4562. I wonder if the impedance of the filter network is just loading the opamp output stage too much. That distortion looks like an output stage loaded to heavily. So you may have to use the 4560 where originally fitted.
The other issue is the PSU loading with all those LM4562's. They do seem to be loading the rails and causing a lot of high frequency hash on the supplies.
mooly,
as of the moment the only lm4562 remaining are the ones @ the output stage. i have another 2xs here, stock. i opened it up and checked the opamps, the highpass section actually had a ba4558 and a ba4560 @ the hipass output. i mentioned earlier it had 2 ba4558. i was wrong.
however, it made sense. i removed the 4562 @ the highpass output and replaced the ba4560. the waveform came clean again.
unfortunately i noticed that at the lowpass section 1khz is clean but as i approach 4khz, the distortion starts to show up again. i was using 3.5khz as the cutoff that time so i thought that may be the cause. however it still is the same even if i changed the cutoff frequency beyond 4kHz.
now, on to the filter section. when i swapped the 4558 to 4562, distortion shows up at the outputs even @ 1khz on the lowpass section that's why i reinstalled the 4558.
here's a higher resolution (somewhat) of the entire board. maybe you can make something out of the layout.
http://i123.photobucket.com/albums/o283/bzhei/2xsboard.jpg
btw, the function of the product is actually an active crossover AND line-driver. that may explain the use of ba4560.
my PSU filter caps are stock. do you think it could help if i swap them out to higher quality ones? i have Panasonics here but didn't want to install them until i get the op-amps working properly. i also have silmic IIs for the signal outputs but didn't install them for the same reason.