The reference you are dividing down the output voltage by is usually 1/2 the supply, IME anywhere from 1.25V-2.5V for your typical controller.
Also WRT to getting sag out of an SMPS, there are better ways to skin that cat. The way you propose is wasteful and IMO sort of defeats the purpose of using a lighter weight, high efficiency power supply. Plus using a resistor will cause things to sag linearly which is not how tube rectifiers respond. Like all thermionic devices that we use tube rectifiers follow a 3/2nd square law. Since we are using much more advanced circuits than a single transformer and a few diodes (silicon or tube) we should use clever circuit design to at least emulate the type of sag that we are accustomed to. I'd look to adjusting the response of the control loop to modify the behavior of the power supply itself, this way everything stays efficient, cooler (literally and figuratively
) and less stressed in the end.
My dad had some great insight into this:
When one wants to introduce "droop" into a switching supply's output voltage, a very simple technique exists for incorporating a controlled output impedance as the means to accomplish this. Normally, this is the last thing you'd want to do to a regulated voltage supply, since it amounts to a sanctioning of sloppy regulation. However, Intel used to actually specify droop for use with some of its processors as a means of minimizing the number of bulk power supply caps required, thereby reducing cost. Audio circuits that sonically benefit from the incorporation of rail sag can use the same technique. To implement voltage droop, all that needs to be done is to measure the output current of the supply and feed that information back as a means of modulating the reference voltage for the control IC. This action has the effect of establishing a load line for the output of the power supply, in effect establishing an output resistance that can be adjusted as desired. A non-regulated power supply has a fixed output resistance but this cannot be changed and can be quite variable over the operating current range of the supply. In a tube-type rectifier circuit, the space charge at the cathode more-or-less sets the output resistance of a supply (in combination with the output caps, the transformer and the output choke if one is present). The law of the thermionic diode is a 3/2 power law, relating a-k potential to conducted current. So, dynamic output resistance of a thermionic diode is quite non-linear. This sort of resistive non-linearity could be incorporated into a "droopy" switching supply with a bit of clever circuit design.
Cliff Schecht - Circuit P.I.