]But how will you know that it's working correctly; outputting zero volts instead of blowing itself up and outputting nothing?
That's not as important as it might sound at first. For example,
turn the pot a tiny bit and you get 0.1 volts output, where with
a blown circuit you still get 0.0 volts.
Usually someone who wants a power supply that goes down to
0.0 volts really needs it to go down very low, like to 0.4 volts or
maybe even 0.2 volts, but rarely 0.0 volts. Without the mods
it only goes down to 1.2 volts, which isnt low enough for some
While we are on the subject, it's also interesting that when this
IC is used to supply a voltage that has to be current limited and
that current limiting is via the control signal at the bottom of R2,
sometimes the current stays high if the load still draws current
below 1.2 volts. This kind of situation also needs a negative
bias or the current might not actually get limited as intended.
As a side note, i build up a general purpose bench supply with the
LM317 a long time ago. I use it for many things including charging
Li-ion cells (or at least i did back then) so i built in a simple transistor
current feedback circuit (very common) that pulls the bottom of R2
low when the current gets above a preset level (base resistor sets
current limit). I didnt have a problem with the output only going down
to 1.2 though, as most of my loads only draw significant current
above that, like say 2.5 volts or so. This means i didnt need the
negative bias for this circuit to fit my purposes.
If an app does need the output to go to zero in order to properly
limit current however, the current feedback transistors emitter goes
to the -1.25v supply instead of ground...not too hard to do as long
as you have the -1.25v supply scheme worked out already. In this
setup however the -1.25v supply would have to handle the main
current flow. Better might be to use an op amp for sensing and
drive the bottom of R2 with the op amp output, using the negative
supply to power the negative power supply of the op amp, and use
an op amp whos output goes all the way down to the minus supply
rail. Another idea might be to bias the bottom of R2 in such a way
that the NPN transistor can still control it all the way down to -1.25v
even though its emitter is tied to ground. IF this works it would
require a few more resistors.