Troubleshooting bad mid filter on Shelves

I’ve acquired a used Mk1 Shelves in a module swap. It sounds killer but the low-mid filter doesn’t do anything. All the chips have power and CVs are passing into the circuit. However, my attempts to troubleshoot any further are possibly thwarted by my lack of understanding of how the 2164 VCAs work, particularly because they’re connected directly to the OPA inputs inside the filters as well as to the plus/minus busses. Taking measurements on a current-based circuitry (without breaking it) isn’t something I’ve done.

A saner person would just send this to Emile, but I’m an intrepid hardware hacker on a quest. :slight_smile: Any light that could be shed on the theory of operation of the filter* as well as troubleshooting steps to narrow down the culprit would be greatly appreciated.


*PS …including what makes that filter “Low-Mid” when components seem identical to the “Hi-Mid”? I may be “missing the forest for the trees” here, to use the saying.

It’s a pity because it takes me much less time to troubleshoot the module myself directly rather than playing blindfold chess.

Do the LP/BP/HP outputs of the filter work? At least we can know if it’s the filter core or if it’s the VCA that adds the BP output to the mix.

Nothing, both mid sections are identical, it’s a deliberate design decision to allow the the FREQ knob to sweep the whole spectrum on all sections.

Thanks. Like I said, I’m looking too closely without seeing the big picture. Agreed that you’d find the problem quicker–but there are also signs that a previous owner unqualified to wield a soldering iron took some matters in their own hands around the power header :\

Your suggestion about the HP/BP/LP outs led me to assemble a DIY Shelves Expander to audition those filter points. I fed Shelves a sawtooth and modulated both Mid band frequencies with a Sample/Hold in the spirit of Karn Evil. The problem filter is apparently working, but not as well, which is probably why I didn’t detect much audible influence in the mix output.

With Q controls at minimum, both filters sound identical and look the same in shape and amplitude on my scope in every respect. With Q at maximum, there is a difference. All the outputs work, but the second “teal” filter is slightly quieter with somewhat less “rezziness” than the “magenta” band. I recorded the audio here.

The recording is at maximum Q and has short samples in this order:

  1. magenta HP
  2. teal HP
  3. magenta BP
  4. teal BP
  5. magenta LP
  6. teal LP


I’ve started on some reading of a good resource on voltage-controlled state variable filters. What looked like “magic” is finally starting to make sense to me. A learning moment. :slight_smile: