We use an LM5116 Wide Range Synchronous Buck Controller to generate 12VDC. The DC load is typically only 220mA. The capacitive load is 200uF. The soft-start capacitor is 0.1uF which provides a 12msec ramp-up to 12V on the output. The design is extremely similar to the LM5116-12 Evaluation Board. Like the LM5116-12 Evaluation Board, the regulator output voltage is fed back into the VCCX input, to allow the gates of the external Si7852DP FETs to be driven at 12V instead of the internally generated 7.4V.
When the VCCX input reaches 4.5V on its way up to 12V, the LM5116 begins to use this voltage (minus the external diode voltage drop) to drive the gates of the external FETs. Unfortunately on some of the boards, the resulting voltage at this point appears to be too low to reliably turn on the external FETs. The LM5116 enters hiccup mode, as if an over-current were occurring.
More than a single date code of the LM5116 exhibit this behavior. The LM5116 was replaced on at least one problematic board, correcting the symptom. However, when that LM5116 was checked at TI, the chip tested good.
Is this a known problem?