The off-state path gives me a particular place to follow power loss. In the XL4015 example, current passes through the freewheeling diode. In the MP2307DN example, a controlled low-side MOSFET conducts. The useful comparison is between a forward-voltage-drop term and a resistance term, with the conducting fraction kept beside both.
For the simplified diode model, that means current times forward drop times the off fraction. For the low-side transistor, it means current squared times on-resistance times the off fraction. The square changes how the loss depends on current. That gives me something I can vary and explain, instead of simply repeating that a synchronous converter should be efficient.
For a diode, the useful starting idea is the forward voltage drop. If current is flowing through a component while there is a voltage across it, there is power associated with that drop. A rough diode conduction-loss estimate uses forward voltage multiplied by current, with the conducting fraction of the cycle included. The exact current waveform and the way the diode behaves matter when I want a more accurate result.
The transistor starts with a different expression. When the MOSFET is on, its conducting path has resistance. The simplified conduction loss is related to current squared times that on-resistance, again over the interval when the path conducts. The square is important. Increasing current changes that term more quickly than a first glance at the circuit diagram suggests.
I find it easier to compare the expressions with completely values first. I kept the current and conducting fraction the same for the comparison. In this case, a path carries a constant 1 A for half of each cycle. If a diode has a forward drop of 0.4 V, its simple average conduction-loss estimate is 0.4 × 1 × 0.5, or 0.2 W. If a transistor has 0.1 ohm on-resistance, its corresponding estimate is 1² × 0.1 × 0.5, or 0.05 W.
That example shows why the transistor path can look attractive. It also lets me change the current and see which term moves. At a 3 A with all those simplified assumptions held fixed, the diode estimate becomes 0.6 W and the transistor estimate becomes 0.45 W. The gap hasn't simply scaled in the same proportion because one expression contains current squared.
I wouldn't use that exercise to choose a real part. Actual forward voltage and on-resistance aren't constants that remain exactly the same in every condition. Temperature matters, the current has ripple, and switching brings other losses. The example is useful because it exposes the different dependencies, not because it replaces the datasheet.
The conducting interval also deserves more attention than I gave it initially. In a simplified buck circuit, the off fraction is related to one minus the duty ratio. A different input-to-output ratio changes that fraction. So even if I left the load current alone, the time spent using the freewheeling path could change. The power supply's voltages belong in the discussion, not just in a separate heading above it.
This is where I need to keep a whole-converter efficiency claim separate from one loss term. Reducing the loss in the freewheeling path doesn't remove the main switch's losses, the inductor's losses or the controller's own requirements. A smaller term in one calculation can be meaningful without proving that every part of the completed converter will be better.
There’s also more work involved in controlling a second switch. I don't want to turn the explanation into a long list of disadvantages, but the timing isn't free just because it is handled inside the chip. The two paths have to be managed so they do the intended jobs. The simple drawing leaves out a lot of what the controller is doing.
I'm writing the origin beside each number I use. A value copied from a datasheet belongs with its stated conditions. An assumed value belongs with the reason I chose it for the exercise. A measured value would need the actual setup and record. Once they are all multiplied together, it is surprisingly easy to forget that they did not begin with the same kind of certainty.
The part I want my mentor to check is my explanation of the intervals and the current used in each term. If I accidentally use average current where the calculation needs an RMS value, or ignore ripple without saying so, I could get a tidy answer that doesn't mean what I think it means. I want to understand that before adding more loss terms.
What I like about this calculation is that it gives the comparison somewhere to go. I can ask what happens when the current increases, or when the duty ratio changes, and follow a reason for the answer. That's much more satisfying than repeating that synchronous converters are efficient.
The report's numerical examples put about 0.876 W in the diode path and about 0.652 W in the low-side transistor path. These are model calculations, and they use different off fractions: 0.584 and 0.725. That detail matters before I compare the two answers. It also makes the whole-converter boundary clearer. The high-side switch, switching transitions and inductor copper have their own terms; reducing a freewheeling term does not make the rest of that power balance disappear.
There isn't a winning module at the bottom of this page. There’s a clearer view of one place where power is lost, and a set of conditions that explain why. For the next part, I want to see what happens when those individual losses are added back into the whole converter.