I have a circuit that uses the LTM4600HV to create a 1.2V supply from 24V. Somehow this design made it into production, despite the fact that this Vin-Vout ratio is outside the range of the part. Not surprisingly, the circuit often oscillates. I've found that if I add a 499K resistor from Fadj to Vin (24V) I can stabilize the circuit. This results in a Ton time of 93ns, which is less than the 100ns minimum. I realize I'm off the map for this part, but how much further can I go without falling over the edge, and what dragons might I encounter on the way?
Update: I've gone with a 402K resistor, and therefore an 89ns pulse width without issue on a half-dozen units. I I also had to remove the recommended 100pF cap between Vout and Voset, and instead connect a 270pF cap from Voset to GND. I'm sure this slightly compromises transient response but it keeps the output stable.
Problem solved. The circuit I'm dealing with had the feed forward cap connected outside of an output filter. This put a 0.22uH inductor in series with the feed forward cap. With the feed forward path connected directly to the output I can now get the circuit to be stable with a 400kHz frequency, as suggested. Thanks.