The low dropout regulator is a valuable tool in the power-supply roster; don’t assume its "drawbacks" should automatically rule it out of your design.
I sometimes pity the low dropout regulator (LDO). This wonderful power-supply component has been around almost "forever" in IC form, and new ones are released every year.
Yet some designers scoff at them as being too inefficient, unable to provide more than about an amp or two, and just not sophisticated enough for today's advanced supply needs. Like the late comic Rodney Dangerfield, too often "they can't get any respect around here." This is despite the fact that the LDO is easy to use, does one thing, does it well if you exercise modest care, and doesn’t try to step beyond its basic function.
Let's take the last presumed weakness first. Although the LDO looks simple from the outside̫after all, it's a three-terminal device, and usually requires just an external capacitor or two in use–internally they are quite complex, with temperature compensation, thermal and load protection, stability compensation, and all sorts of other goodies. (Some of the LDOs released in the past year add additional clever features such as the ability to parallel them for higher output current, without trivial additional circuitry.)
As for efficiency: yes, they are less efficient than a well-designed switching DC/DC supply. But sometimes, that extra efficiency is not worth the cost in components, PCB real estate, and other operating factors.
And yes, if you need tens of amps, an LDO is not the way to go. But many applications need far less current, especially in multirail application where there is a primary supply while a few other special supply-rail voltages are needed.
The LDO's other well-know virtue is its low noise. Again, many application can tolerate tens of millivolts of supply noise, but there are quite a few–especially related to transducer and I/O–-that cannot. An LDO may be the nest choice in terms of total system performance, EMI/RFI, and other factors.
I'm not saying that the LDO is the right choice, period, end of story. But rather than assume it is not, check the specs, look at the real costs of its inefficiency versus noise versus filtering versus footprint versus all the other factors that go into supply-subsystem execution. Maybe the LDO is the better choice within certain limits, maybe not. I do know that vendors sell a lot of them each year, and are announcing new ones all the time, so there must be something virtuous about them, right? ♦