Electronic – Learning about voltage references

In normal operation, you would be feeding reverse current through the device, so the Absolute Maximum Rating would be 15mA; the line on the reverse-voltage graph stops at 10mA, but it probably extends pretty straight up to 15mA. Not that one should operate the part there, but one probably shouldn't worry if current gets up to 11mA.

On the other hand, in most cases there's probably not much point in driving the thing with more than 1mA. The graph suggests that the voltage drop will be pretty constant provided there's at least ~0.4mA flowing through it, but the voltage drop falls off very rapidly below that threshold, so one needs to ensure one is operating safely above it. To allow for part variations, 1mA seems like a good safe operating level. The only time I can see usefulness to going above that would be if one has a load whose demand might vary e.g. from 0-8mA. In that case, one might drive (part plus load) with 10mA and figure that the device would have 2-10mA passing through it. Voltage should remain constant provided the load current stays below about 9.5mA.

A silicon diode has a negative temperature coefficient of about -2mV/°C, so a 10°C change in temperature will result in about a 3% shift in your "reference". The forward voltage is also not all that well specified and is fairly dependent on current (which also means that loading the reference will change the voltage). If none of those things bothers you, have at it. It's been done.

A better idea is to use a three terminal regulator perhaps with a voltage divider and buffer. It will draw a few mA \$I_q\$, but will be much more stable than a series string of diodes.

Low voltage (like less than 5V) Zener diodes are not very good (very soft "knee"), so I would avoid them.

Incidentally, should you have a defunct PC power supply kicking around, they typically have an LM431 shunt voltage reference in them (on the side isolated from the mains), and it's usually not what fails.