Question about the LM3150 ripple injection circuit

Hi Go,

Let me try to answers your questions.

1  "・Could someone tell me why LM3150 needs to add the circuit in a little more detail?"

The  3150 is a very similar to a standard "Constant on Time" (COT) Controller.  When the output voltage falls the controller turns on the FET for a fixed amount of time. The controller then waits for the output voltage to fall below the threshold for the next ontime to occur.  This regulates the output voltage.

The drawback of this scheme is that if the output voltage ripple due to the on time is very low, the next on-time pulse can come at erratic times.  This leads to a switching frequency that is not consistent.  To keep the switching frequency more predictable, the LM3150 injects some ripple information from the lowside FET of the buck regulator onto the output voltage comparitor.  This adds a consistent ripple to the comparison between the on-time pulse and the output voltage.  This leads to a more consistent switching frequency.  The circuit works well for Fets with Rdson greater than about 20mohms, but for lower Rdson fets the ripple information from the Fet is not enough by itself to keep the switching frequency consistent.  For these cases, the part relies on ripple from the esr of the output capacitors or ripple from an external injection circuit to keep the switching frequency consistent.  

So to your question.  The reason for the external ripple injection circuit is to add even more ripple to the output voltage comparison for circuits where the ripple is not enough to keep the switching frequency stable.

2)

"・the output voltage setting value is the only point to figure out if we need to add the external circuit or not?"

So it is not really the output voltage setting that causes the issue, the reason to add the external injection circuit is to add more ripple for cases where the ripple is not large enough to keep the frequency constant.

So for large output voltages, the divider on Vout reduces the ripple  from the esr of the output capacitor that the comparitor sees.  So your idea of using a Feed forward capacitor can be used in some cases instead of an external ripple injection circuit.  The times when the Feed forward capacitor would not work well is when you are using a mostly ceramic output capacitor bank.  Becuase of the very low resistance of ceramic capacitors, the ripple information is 90 degrees phase shifted from the switching of the fets. This phase shifted ripple information causes havok with the converter.  So if the feedforward isn't working to stabilize the frequency you would then try the injection circuit.  The injection circuit mimics the esr ripple of the output capacitors.

Things to watch out for.

1) Adding to much ripple using the injection circuit can cause the part to hit OVP on the FB pin, and can cause line regulation issues (as Vin increases the ripple increases effectively raising the threshold of your comparitor)

Ap note-1481 should have more details.

Regards,

Marc