Hello folks,
I have a project that needs a high current supply. 24VDC input, 5V @ 20+ amps output.
I think there are some disadvantages to the traditional multi-phase SMPS controller, and external FETs, and all the passives that go along with that implementation. Clearly they exist, and are probably quite good.
However, is it feasible to gang 4 of the TPS54540's, with an external master clock, and just tie the outputs together? I'd like to avoid using an OR'ing controller and FETs. I really like the integrated-FET approach. It just makes the whole subsystem easier to lay out, probably easier to "tune". I like this IC package - it has actual LEGS! Easier to assemble & rework.
The master clock might be a single clock phase, or possibly I could generate a 4-phase clock, this spreading out the peak switching current spikes.
Is it reasonable to use a single Vfb resistor divider (at the final output), and tie all the Vfb pins together?
If this could work, the advantages I see would be:
a) Circuitry is distributed (4x 5V@5A subsystems), so parts can be smaller & cheaper. Heat dissipation is also spread out. I have a good sized PCB, so space isn't too much of an issue for me.
b) If I do need a system configuration with less power output, I could certainly de-populate and build a 5A, 10A or 15A module too. The master-clock would need to be 2-phase or 3-phase of course. I can do that.
What I don't have a feel for... is whether there will be some instability or unpleasant interaction between the subsystems by ganging multiple SMPS chips this way.
Feasible? Crazy? Any other ideas?
