In a very simplified way, you have 3 PT2399 with very short delays, set by the resistor from pin 6 to ground, so 4.99k (longest), 3.24k(intermediate) and 1.82k(shortest) , I *guess* to simulate a 3 spring reverb, the most complex they make.
Fine but itīs way too "accurate" and does not sound that much like a mechanical spring but more some Digital thingie.
So they modulate the middle one with an oscillator.
I donīt know the internal PT2399 timing, although I guess its clock must run quite high.
I guess the external oscillator modulates the internal one (or beats against it) and it introduces some inaccuracy ... which is good and more like mechanical springs do.
beware: lots of caveats and "what if" involved here.
In any casel "they" do not explain it.
They somewhat show "how" (in the patent) but information on "why" is scarce.
I guessed more based on what is being tried to accomplish than on Belton supplied data.
Others may explain it better, of course.