Crossover ideas for processing

I have heard from a reliable source who was instrumental in dub reggae at the time that certain reggae engineers used to take mono signals and divide them into treble and bass using crossovers or low pass/high pass filters - and then feed them into different amps/preamps - meaning they could take treble sections and drive them hard through tube gear to achieve a certain degree of gnarliness but process the bass at not such high gains to maintain clarity and punchiness - and then they would mix them back together after processing.

Can anyone see any underlying problems to this. Would there be phase problems from inserting 2 entirely different preamps for example one tube one transisor????

Kind of like multiband processing. I have often wondered about this. Like with a drum sound that sounds awesome when the treble/hi mid has a lot of 2nd harmonic distortion but the low end sounds awful. If you could process them differently and then combine them that would have interesting results I believe.

This has been done in broadcast for decades.

Early work used crude loud speaker crossovers.

I would suggest using a variant on frequency dividing networks called "derived" (I think), where either the HP or LP function is generated by subtracting the other from unity. You don't get as steep of a slope on the derived output, but they will recombine better, assuming you don't introduce phase shift in your processing.

JR

FWIW:

http://www.amptone.com/quadrafuzz.htm

http://www.dlldll.com/quadrafuzz.dll_download.html

You can get some interesting effects using a couple of console channels with EQs. Run the same signal through both channels and use the filters to pass the low band on one channel and the high on the other. Depending on how capable your console is, you can get some very elaborate combinations. Post-eq inserts would allow other effects such as compressors and reverbs to add to the fun.

That capability was designed into some DJ mixers so they could seamlessly combine the bass line from one record, with the top from another.

I wouldn't expect console EQ to very cleanly separate and recombine.

JR

[quote author="JohnRoberts"]...

I would suggest using a variant on frequency dividing networks called "derived" (I think), where either the HP or LP function is generated by subtracting the other from unity. You don't get as steep of a slope on the derived output, but they will recombine better, assuming you don't introduce phase shift in your processing.

JR[/quote]

Came across this one this weekend:

http://sound.westhost.com/articles/derived-xovers.htm