Reverse-engineered op amp crossover questions

Hi. I volunteered to modify the crossover points of a crossover. Existing crossover points were 250Hz and 2500Hz.

The 250Hz low pass sections look OK but in the mid-range and HF filters the response rises with decreasing frequency below about 60Hz. There would be significant drive at 20Hz and below to the MF drivers and tweeters.

Then I reverse-engineered the schematic shown below. It’s like nothing I’ve ever seen before.

You can see buffers and standard Sallen and Key 24dB Linkwitz-Riley 250Hz and 2500Hz low pass filters in there.

There are no standard Sallen and Key high-pass sections.

But what are IC2b, IC4a and IC6a and their associated resistors and capacitors? Simulating them in isolation shows they have a kind of peaked band-pass response.

IC4a doesn’t appear to be doing anything in that position and maybe should be in the high-pass section.

The output of some stages appears to be added/subtracted at various points.

The PCB is an old, custom single-sided affair with no resist or component markings. Maybe whoever layed it out made lots of mistakes.

Maybe I’ve made mistakes reverse-engineering the schematic but I’ve checked it a number of times.

Can anyone tell me what the designer might have been trying to achieve and what might be the advantage over a standard Sallen and Key topology?

Ultimately, I need to be able to change component values which is easy for the 2 low pass sections but not easy for the high pass sections.


  • bass.png
    84.7 KB · Views: 94
  • mid.png
    92.1 KB · Views: 92
  • high.png
    89.7 KB · Views: 92
  • schematic.png
    215.4 KB · Views: 97
I gave the schematic a good look-over. Didn't notice any schematic typos, so to speak. It looks like the high pass function is created by subtraction of an all-pass (not sure on this) from the low pass via R27. The trouble with subtraction is to get good stop-band rejection the gains of the two paths need to be well matched - chances are the re-emergence in the bass is due to insufficiently close tolerance resistors? This is just a guess as I've not simulated it, too lazy to type it all in :)