We have noticed that using a 100n capacitor to couple an external 500kHz clock signal into the RT pin of the LM5116 is associated with anomalous startup behaviour in a recent batch of parts. In particular, affected parts started switching even though the EN pin was held low. All anomalous behaviour is eliminated if either (1) the 100n capacitor is removed, allowing the LM5116 to run using its internal oscillator, set for 460kHz, or (2) the 100n coupling capacitor is replaced with a 100p capacitor, as suggested in the LM5116 datasheet.
I would like to understand in more detail why the use of the larger coupling capacitor can disrupt the normal function of the LM5116. An earlier production run using the 100n part did not show any abnormal behaviour and I would like to assess the risk posed to those assemblies by not replacing the capacitor with a 100p unit.
