Thank you all for the positive comments!
I'm sorry that I will keep the detailed schematics for myself since this will probably be a commercial product. OTOH I find it unlikely that many people would build this variation since it requires substantially more components than Sontec, "only" to avoid some issues most people won't even notice when using the original.
This topology has whole range gain which is not dependent on Q-value. A decision has to made how much is tolerated, and this in turn affects the noise performance and available Q-values. It is a compromise. I settled on 0,27dB whole range gain for one band 8dB boost. This is more than normal Sontec in high Q setting, but substantially less than normal Sontec in low Q settings.
Noise performance also is no more dependent on the Q setting as heavily as in the original. Now one band adds noise only according to its own bandwidth, the peak value does not change along Q. The worst case is high shelf, from 4th bands lowest frequency, where the impedance level is highest. In my circuit this is around -80dBu. Typical bell curve produces only -96dBu noise.
Worst case distortion performance: Cutting max level, 20kHz, maximum amount. Picture attached. In any real world settings the distortion is not measurable with AP portable one.
