I did the LTSpice simulations of the rev3, rev4 and Stephen's MI LFO.
Here are the three schems in comparison:

The calculated frequency ranges of the three LFOs are pretty similar.
Rev3: 40 mHz to 18 Hz
Rev4: 35 mHz to 13.7 Hz
Stephen: 45 mHz to 9.5 Hz
Specs of the original: 30 mHz to 10 Hz

(V(rev32) is for rev3 with double integrating capacity)
As predicted by oldschoolanalog doubling the integrating cap halves the speed (just checked for ver 3: 20 mHz). Thanks for the formula btw!
Interestingly, the slow sweep frequencies are all exactly at multiples of 5 mHz. I suppose this somehow correlates with the 200s ( Took ages neverthless

) I used for simulation time and the method of calculating the FFT. Maybe someone can shed some light on it ?

Markus