I'm building my first switching DC/DC converter as part of a larger circuit. I need to convert a input voltage range to an overlapping output voltage (e.g. 3-5V in, 4V out). This would typically be done with a buck-boost or SEPIC converter. I thought of a way to use a simple boost converter IC instead. If it works, it allows me to go monolithic, while an equivalent buck-boost would require a controller + MOSFETs considering the ICs that manufacturers offer (high currents involved).
Because I don't need the output to be referenced to ground, I was planning to reference the load to the input voltage and change the switching IC feedback network to subtract the input voltage from the output.
For example, let's take a 4V output. Load is connected between input and output.
- Input = 3V, the converter generates 7V (from ground), potential across load is 4V.
- Input = 5V, the converter generates 9V (from ground), potential across load is 4V.
It seems to me like it should work and that I can calculate the inductor/switch current based on the actual voltage across the load. I searched a lot, but couldn't come up with much. The best I could find is the "Negative-to-Positive Buck-Boost Converter" circuit from page 24 of LT's AN19, which looks very similiar to what I'm talking about (with input voltage as ground and ground as negative voltage).
Can this be done? If yes, can inductor and switch currents be calculated on the voltage across the load instead of the full ground-referenced one? Seems logical since the inductor is in series with the load, but if it's not true then I don't get any benefit out of this and I'll have to go with a controller.