Interesting resonant low-pass filter

[amageddian]

I love the sound of early Lowrey organs.



These work in much the same way as most electronic organs (including combo organs; not including Hammond tone-wheel organs):

• tone generation employs twelve tuned LC master oscillators, and successive divide-by-two circuits to create square waves,
• octavely-related square waves are mixed ("staircased") together into richer tones that approximate sawtooth waves,
• those rich tones are separated into ranks, corresponding to organ pipe lengths (16, 5 2/3, 4, 2 1/3, 2, 1)
• the voicing controls mix one or multiple ranks, or groups of notes within ranks, through various fixed filters, to derive voices of different tonal qualities:
  • diapason/ principals
  • reeds
  • strings
  • flutes

The flutes are the simplest voicing path: the objective is simply to attenuate all harmonic content above the fundamental. Lowrey used an economic approach to this: rather than filtering each note separately, groups of 7 or 8 successive notes are routed through a common filter, whose fixed resonant peak is somewhere in the middle of that note range. Thus, the organ only requires a handful of filters, depending on the compass of the instrument. My own Lowrey (an early 60s "Heritage" model, which uses all tube circuitry) uses only 6 of these.

The circuit for this flute filter is interesting. Here is the 1200 Hz filter from the early tube models (caps are 750pF and 0.001uF; R1 is "selected" probably 39K or 42K; the output is at the leftmost junction):




Here is the same filter from the later transistor versions (redrawn by me for quick simulation in EasyEDA):



simulation response :



I've seen this described as "similar to a phase-shift oscillator, but without enough gain to self-oscillate". However, to me, the passive section in the feedback path looks less like a phase-shift than a twin-T filter -- although the component values chosen do not correspond to the usual R, C, R/2, 2C required to tune a band-reject filter.

I have a pretty elementary grasp of electronics -- acquired mainly through trying to maintain my musical instruments, and from browsing this community, so I'm trying to understand: why does the filter respond this way?
Is the output a mixture of the non-filtered, amplified (phase-reversed) input, with a band-passed signal achieved through the feedback circuit? If so, why does it roll off so much above the resonant frequency, while the response below the peak is relatively flat?
Why is the input signal coupled to the transistor base, rather than passing through the twin-T section?

My suspicion is that there are simpler -- cleaner? perhaps less phase-destructive? -- implementations of flute filters for organs, but that this particular topology contributes to the unique fizzy sound that Lowreys produce, especially when flute voices are combined with reeds or strings. Anyway, I welcome any input!

 

[amageddian]

I discovered a pretty similar circuit (with controls for Q and frequency) in a PAiA band-pass module kit:


https://audiodiwhy.blogspot.com/2020/03/piece-of-paia-2720-bandpass-vcf-lives.html

 

[clintrubber]

Why is the input signal coupled to the transistor base, rather than passing through the twin-T section?

The Twin-T section is in the feedback-path of the active element (tube/transistor/opamp). So what it would filter out on its own is now in fact amplified (due to the combination of this frequency selective feedback and the amplification stage).