Is it possible to "stack" the outputs of 2 or more buck converters?

Question

Is it possible to "stack" the outputs of 2 or more buck converters?

The buck converter circuit in question I am considering is the 48v to 5v typical example circuit from page 20 of the datasheet PDF for the "LT1339ISW#PBF" part found on mouser here.

LT1339ISW datasheet

I am posting from mobile, but if I can figure out how to directly post the datasheet circuit I will edit it in here later.

I wondered if it was possible to get the outputs of 2 of these buck converters to "stack" their output voltages by connecting the 48v inputs in series to a 96v voltage source and floating the grounds on the second buck converter.

I am thinking that if done correctly the 2 buck converters would share the 96v input source just like 2 equal valued resistors in series will equally share a voltage across them. And as long as you connect the load correctly, the outputs of the 2 buck converters would then act as if they were 5v voltage sources adding to an output of 10V at the up to 50A max rated output current.

For the sake of simplifying the question I am assuming it is possible to balance the inductance and capacitance of the the "stacked" buck converter circuits such that they function the same as the values in the single example circuit mentioned above.

I do realize that even if it is possible, this would not be the most practical circuit since real world components have varying tolerances and any significant imbalance or failure would likely cause cascading failures several other places in the combined circuit. At these voltages and currents it would make for a great risk of fire and injury to nearby people and objects.

Answer 1

Usually: no.

Buck converters are usually not isolated, i.e. the output shares a potential with the input. Most often, that's ground. Therefore, "stacking" would just short the output of the lower one to ground. "Stacking" on the input simply makes no sense, these are not resistors that share the voltage drop. So that can't work out at all.

You could also build an isolated converter using the same or a different IC, but you'll need to come up with a rather complex method of feedback from the secondary side if you want competing, stacked regulation to still be good. Can't think of a case where that complexity would make sense more than you building a regulator meant for 96 V.

Answer 2

... just like 2 equal valued resistors in series ...

No. Buck converters have switching transistors.

Figure 1. Image source: Toshiba Semicon Storage.

Since your two buck converters will switch asynchronously you can be guaranteed that at various times one transistor will be on and the other off. You will not get even division of voltage across the two devices. The one that is off will have full line voltage and the one that is on will have zero current (because the other is off).

That's the first problem.

Figure 2. All the pink lines are connected together forming a nice short-circuit across the input terminals of the lower switcher. (Image by author using Toshiba artwork above.)

The second is that usually there is no isolation between input and output. Connecting the outputs in parallel will short-circuit the power inputs on the lower of the two units.

---

Series connected inputs + series connected outputs

Figure 3. An attempt at series connecting the inputs and series connecting the outputs won't work either. The lower switcher is completely bypassed.

Answer 3

See the topology image of the buck converter:

You are planning something like this:

It's still one converter except the control electronics will not get any more power from the input and the diode path of the upper converter is blocked, both are blocked by the lower converter. Or the diode path is not blocked, if the switching transistor has a drain-source reverse diode. But without proper simulation the function of the control circuit is a puzzle which cannot be reasoned beforehand.

When the system is started the output capacitor of the lower converter is empty, the upper one gets all of the input voltage. It's maybe not designed for +100% overvoltage and probably puffs the smoke out.

Does not work, forget it!

Answer 4

I believe now that I misunderstood the question, but as I was very insisting in several comments, here is what I had in mind, namely stacking the output voltages of several regulators (which was not asked).

Barring internal details, both an LDO and a Buck look like this:

The internal details really do not matter for this fundamental analysis. However, the switching nature of the Buck will demand more care in a real implementation.

Output voltage stacking of such regulators is possible:

But of course, the output transistors of the top regulator have to tolerate the voltage to Ground. Also the regulation will become worse as more regulators are stacked, as each of their misregulations add up. Also, problems with oscillations can be harder to fix. So I agree, it is not a straightforward or usual thing to do, but is not fundamentally impossible.