By accident I stumbled upon this Linear Technology application note (LT1074/LT1076 Design Manual):
Just for fun I simulated the positive to negative converter in Fig. 16. exactly as it is.
The authors state on page 26:
Positive to negative converters have a “right half plane zero” in the transfer function which makes them particularly hard to frequency stabilize, especially with low input voltage.
Why is this the case especially for low input voltage? I see that the compensation network R1, R2, C4 is very important for input voltage Uin=5V and that there is ringing oscillation at the output when it is omitted.
I would like to understand from a theoretical point of view, what this network does: Why is it especially required for low input voltages and not for higher? Most interestingly, at higher input voltages it can be removed and the circuit is stable. I wonder also, why it is connected to the switch side of the inductor and not at the output?
At least in LTSpice it is very critical to use the stated values; changing the capacitor by factor 2 or 0.5 leads again to ringing in the output.
What is the key to understand this sensitive design?