78L15 with Latch-Up Error

Look at this simple and stupid power supply. It should work pretty good, or not?

All tested and used fixed-voltage regulators are marked with an ST logo, no third-party production.

The rectified input voltage is at about ±18 V. So no stress for the regulators.

Error pattern⚓

Often (>50 %) on start-up, the positive power supply 15P sticks on –0.7 V (really negative). No current is drawn, C15 charges to the peak value of rectified input AC voltage.
After a short (<1 s) power cycle, the supply works as expected.
A replacement 7815 (note without „L“) did not show this effect.
Another 78L15 did show the same effect but less frequently (20 %).
Changing values of C14 or C15 did not visibly change this behaviour.

Explanation⚓

The input AC is rectified by two half-wave rectifiers. In case of on-switching at the negative half sine wave, C14 will be charged via load resistance in negative direction. When reaching –0.7 volts, the 78L15's internal loopback amplifier seems to invert its output value, and the regulator will not pass current to the output to reach the nominal +15 V. This is a typical latch-up.

Interestingly, the negative regulator (79L15) did not make problems. Also the big 7815 „L“ works as expected.

Solution⚓

A Schottky diode with cathode on 15P and anode on ground solves the trouble!

A notice in the data sheet would be highly recommended. Or better a new series of latch-up-free 78L15. The error would surely affect another output voltage types like 78L05, 78L10, 78L12, 78L24 etc.