I have been using an LM3150 to convert well-regulated 24VDC down to 12V. The supply was designed on Web Bench for a max current of 8A, but in reality it rarely exceeds 5A. I used the circuit mostly as designed by Web Bench, but substituted lower RDSon transistors.
Anyway, the circuit becomes noisy and fails to regulate under light loads (0.1A - 1A). After reading posts here I decided it must be the ripple injection circuit, read the App note, and guessed that Web Bench had transposed the capacitor values (App Note says AC Coupling cap should be 10x the integrating cap). Another hint was that the regulator was going into Discontinuous mode, i guess due to too much apparent feedback. With the capacitor values swapped it works much better, but I still see 1V noise under load. The frequency of this noise is much lower than the 260KHz switching frequency, so it's not inductor ripple.
Can TI provide any further guidance on tuning the ripple-injection circuit while still maintaining stability?
original circuit (C62 and C65 later swapped):
