...so I poked around some more and finally realized it was the Zener that was leaking. Without a Zener at D3, everything is perfect.

One thing to keep in mind for future reference is that low-voltage Zeners will often show a "soft knee", wherein they begin conducting significantly at voltages well below their nominal Zener voltage. Higher-voltage Zeners tend to have a more sharply-defined onset of conduction.

I got bit by this once years ago, when I tried to use a 5.6V Zener as an overvoltage clamp on the 0-5V input of a circuit I had designed; the circuit showed unacceptable nonlinearity near the high end of that range, and I finally tracked it down to the Zener. Changing to a 7.5V part cured the problem.

 

Can the MOS transistor be removed with the following changes?

 

Can the MOS transistor be removed with the following changes?

View attachment 129979

Interesting thought. I believe you're right. I had briefly considered trying to do something like this, but never got as far as working out the details because this is a one-off project that I may expand upon, and I like the idea of it being flexible. If I leave it as drawn I can add more switched outputs in the future, all from the one current limiting circuit.

If this were a fixed design, or if I wanted each switched output to have its own current limiting circuit, your change looks like a great way to reduce the parts count.

I'll keep this in mind for future use. I do have one thought on it. If I'm not mistaken, to maintain the same set point for the current limit, I'd need to reduce R4 to account for the voltage drop through Q1 and keep the voltages and currents the same through R5 and Q3. If you, or anyone else here, can confirm or deny that, I'd love to know for sure. I'm pretty sure I understand the basic mechanics of the current limiting circuit, but just barely. I'm hanging by a thread!