Tant. cap ESR

Which kind of goes with (1) earlier -- TL431 has pretty high gain. It's really an op-amp, which happens to be unity gain, dominant pole compensated (so beyond just an op-amp, it's even more specifically an inverting integrator), and like many op-amps, doesn't always tolerate capacitive loading.

Incidentally, of the manufacturers that even provide "forbidden regions" of bypass, all the best values are out -- roughly 0.01 to 1uF.

Perhaps unsurprisingly, the TL431 has an open collector output, putting it in the same class as LDOs. The difference is simply which rail the error amp is referenced to -- TL431 is referenced to ground and pulls down with an NPN; LDOs pull up with a PNP, but the amp is still referenced to ground. As with LDOs, ESR stabilizes it nicely; about 50 ohms is enough most times. You can do the little compensation cap from output (i.e., "anode") to -in ("ref") and run your resistor divider to the actual output (after the ESR resistor) to get exact DC response still.

Tim

It is the marketing people that don't like the external cap. You get dinged for every external component, even if the pin can be NCd. I'm with you here.

But overall, the shunt is easier to compensate. Nobody likes them due to efficiency or lack thereof!

Then it's high time to reign in those nonsensical "incentivation rules" and marketing types. Although I have to tread carefully there because I married a marketeer.

An unconditionally stable LDO with 1-2 external parts that does not cost more than 30c in quantities would become a real juggernaut. Because people like me would start using LDOs instead of shunning them and that means lots of sales volume. Of course, this does require longterm and out-of-the-box thinking on the part of management and that's what's often lacking.