Chaining buck converters

Question

I am using these step-down adjustable buck regulators for various purposes (hobby projects and experiments) and they work more or less fine. But I have this little concern - when they convert relatively high voltage (~20V) to relatively low (5V or 3.3V) they tend to get pretty hot (almost impossible to touch them for more than a few seconds).
They still work, and I add some small heat sinks when I actually put them into something more "final". I don't know if they have thermal protection, but they haven't failed me so far.
Still, I've been wondering whether it is a good idea to use them in "series", I mean connecting the output of one to the input of another. For example, the 1st could convert 20V to 12V and the other would further step down from 12V to 5V.
This way they would share the "work" and eventually run cooler.

Will that degrade the overall efficiency?
Will it be twice the losses ... or less ... or more?
There are some cases I would care about efficiency (e.g. if the source is a solar panel or a battery) and others where it is not an issue (mains powered bench top power supply, sorry green people).

Would it actually improve reliability?
Apart from eventually making them run cooler, is there any benefit?
I don't know what happens if one of these fails - but I guess it will either stop conducting (no harm done) or just pass the input voltage to the output (probably frying the electronics connected as a load).
I guess if it is the latter, there is still some (minor) improvement - if the 1st one fails, 20V will come at the 2nd input, which will then regulate them to 5V and the load will survive ... until the 2nd one dies from the same cause the 1st died. At least the load will live a bit longer.
If the second one fails first, the load will see 12V (instead of 20V) and will be a cooked at a lower temperature, so it will taste better :)

Is there any other reason why it makes no sense to do that?

We're talking hobby grade stuff for personal use, mass production cost is not a factor.

Answer 1

Converters get hot due to the non-ideal efficiency of the devices; without knowing the load you are using I can only generalise. Note that the junction temperature of these devices can be quite high and not degrade performance (just what temperature is device dependent).

Will that degrade the overall efficiency?

Yes. The overall efficiency will be \$\eta_1 \cdot \eta_2\$ where \$\eta\$ is the efficiency of each converter: Assuming them to be 85% efficient, you will have a total efficiency of 72.25% for the pair in series.

Apart from eventually making them run cooler, is there any benefit?

Generally, running things cooler extends the life of the component(s)

Is there any other reason why it makes no sense to do that?

Unless efficiency is the overriding concern, no, as you do not appear to have regulators that can handle the load and remain cool.

Note that it is not that unusual to generate an intermediate voltage rail at the expense of some efficiency; this can make power distribution more effective and spreads heat around so it does not concentrate at a single converter.

Update: comment on power consumption of the regulators.

At 20mA for 20V input and 5V out, then the power dissipation is 15 x 0.02 = 300mW. A temperature rise of 30C (which you would notice) implies a thermal resistance of almost 100C / watt.

That said, this is apparently the static dissipation and any extra dissipation will be due only to ineffiencies, so when fully loaded the junction temperature is still safe (probably when operated at room temperature).