Apparently, switched-mode power supplies are quite hard to get right.
How many of these problems (such as oscillations/stability concerns) generally remain when using off-the-shelf DC/DC converter ICs?

What I'm aiming to do is to use two regulators, one DC/DC switched-mode and one linear, to convert a ~22-24 VDC input to an adjustable (0 - ~18 V, perhaps) PSU. The purpose of the switcher would then be to increase efficiency for the lower output voltages (for e.g. 5 V output, (22-5) = 17 W/A of heat for a linear regulator isn't very nice), and to improve the noise figures.

As far as my knowledge/experience is concerned, I've learned a ton from a proper intro course the past 3 months, so I'm a bit embarrassed by my previous postings here at this point.
Still, despite basic coverage of a wide range of topics, I'm obviously not a pro from one course.
As for specific knowledge on this topic, I understand the basics of how switched-mode supplies operate, but not a whole lot more.
I'm willing to learn more, of course - if it's a manageable amount of information to take in, i.e. a lot less than the full design of an proper SMPS from scratch.

 

The vast majority of problems are related to PCB layout issues.

 

So I've heard, but do the same concerns remain for these ICs? I would assume they mostly do, unfortunately, since there'd still be a lot of current being switched in a square-ish wave pattern.

 

You say your input range is 22 to 24 volts,
You say your output range is 0 to 18 volts.
(Your load current is unstated.)

An adjustable buck regulator would work just fine as the voltage ranges don't overlap, so you may not need the output linear regulator. ti.com (or national.com, same thing now) have some good online tools to create complete schematics and parts lists based on their chips.

Other vendors have such too, I'm just more familiar with that one.

As far as the PCB goes, just identify the high current traces and make them wide enough for the max current.

 

So I've heard, but do the same concerns remain for these ICs? I would assume they mostly do, unfortunately, since there'd still be a lot of current being switched in a square-ish wave pattern.

The controller apps are the hardest to lay out correctly as you have to optimize the high current paths and keep things short as well as manage ground currents. Smart thing to do is buy an "demo" board from the manufacturer.

 

..........................

As far as the PCB goes, just identify the high current traces and make them wide enough for the max current.

It's not quite that simple. You need to layout the board carefully so that the high current paths for the switching currents are short and do not introduce interference in the analog control loop part of the circuit, and the PCB should have a ground plane for good decoupling of the supply voltage and a low impedance path for the high frequency ground currents. Generally the spec sheet for the PWM IC will show a recommend IC layout that should be followed as closely as possible.

The ICs themselves are usually well designed and trouble-free if used well within their recommended spec limits.

 

It's not quite that simple.

Amen. I could write a book about the layout screw ups people have done, and some of them were engineers working with me who should have known better.

Laying out switchers is an art form, and the higher the switching frequency, the more critical the layout is.