I’ve built two of these as well and have run into similar issues. Maybe we can try and troubleshoot this one together.
First off, the module just sounds really thin for a 4-pole LPF. In addition, the Cutoff & Resonance are very much affected by the Input level attenuation control. As I increase the signal level into the module (AKA decreasing the amount of attenuation), the slope of the filter seems to decrease. Like at below 50% on the Input knob, I get that very obvious carving 4-pole filter sound (even if it’s just carving noise). As I increase the Input knob setting, the filter loses its sharp slope & sounds more like a 2-pole LPF. As a result, I’ve removed both filters from my rack as they were unpredictable to patch and I have 2-pole filters that sound more like a Moog VCF.
On my builds, I used polyester film caps and did match the capacitors to <1% on a cheapo TC-1 Multifunction Tester (although I doubt the tester is really accurate down that far):
For transistors, I used BC547B instead of BC547C (it’s what I had on hand). Maybe this could be part of my issue, does anyone have experience with how hfe affects ladder filter performance? I also used Ian Fritz’s transistor matcher on ~75 transistors from the same batch, and found 3 pairs that were matched in VBE to 0.01mV (although I don’t trust my meter going down that far.)
I also tried a bunch of different CA3086s from several vendors (including a bunch of NOS ones sitting in storage at work). There may have been some difference between their performance, but not enough to make one set solve the issue.
I would love to get this thing figured out as I enjoy the sound at a narrow band of settings, but it’s just kinda useless as-is.