In the intended application the actual values of the main rails, V+ and V-, can range from +/-5V to +/-18V.
Fixed voltage supplies referenced to ground cannot be used since they do not track the variable rail voltages.
I developed a shunt regulator that emulates a programmable Zener using the LM317L and LM337L.
The "Zener" voltage is programmed by the supply rails to maintain a tracking voltage that is fixed approximately 1.4V below the supply rail.
Note that the positive supply uses an LM337L "negative" polarity regulator; the negative supply uses an LM317 "positive" regulator.
The regulators are configured as shunts with the inputs to the LM317L and LM337L each connected to ground.
The inputs to the LM317/337L are analogous to a Zener's anode, the LM317/337L outputs are the cathode connection.
The adjust pins program the Zener voltages.
Thus, due to the unusual shunt configuration, the polarity of the regulators are reversed.
Load and Zener current is provided by the programming resistors and an optional pre-load resistor.
A pre-load resistor may be required to provide additional load current for the tracking output and to maintain the minimum load current of the LM317L/337L. (1.5 mA typ. for Vin-Vout between 3 and 15V.)
In many cases the ~ 10 mA current provided by the programming resistors (using the values shown in the example) will be enough to satisfy both the regulator and load current requirements.
If the voltage below the rail is high, the higher values of programming resistors alone may not be able to supply all the load current and a pre-load would also be required.
With the values shown, +/- Vtrack is about 1.4V less than V+ or V-.
Vtrack+ is approximately equal to "V+" minus 1.25[1+(15R/120R)].
(The regulator equation is actually Vout = 1.25[1+(15R/120R)]+ Iadj*15R). Iadj however is quite small (50 uA typical) and this term can usually be ignored.
It should be noted that whatever ripple, noise and load variations on the main V+/V- rails will also be present on the tracking outputs.
A tracking shunt sub-regulator using the LM317L/LM337L.
This EDN article http://www.edn.com/design/analog/442212 ... -the-LM317
made me think it might be a good idea to post the Shunt sub-regulator I did some time ago.