For my own regulators, I use the forward drop across an LED to reference a bipolar current source into a fixed resistance. Obviously, fixed
current into a fixed resistance results in a fixed voltage, which is hard referenced and stabilized by the forward drop across the LED.  Not incredibly stable, but it is a "hard" enough reference for bias supplies and the like. Very little noise generated compared to a zener diode.
  You mentioned using two resistors and a bipolar to make your virtual zener. I assume this is a simple voltage divider, referencing base of an emitter follower--which would be a "soft" device that floats up and down with the main supply variations--am I correct here?
  Perhaps you could elaborate for me if you are using a different technique that I am heretofore unaware of

  I have found that not all zeners are created equal. The lower voltage and higher wattage types are less noisy than most. (Some zeners come packaged in blue plastic, which signifies tighter tolerances and lower noise.) One test is to attach a series resistor that connects to a much higher voltage power supply and measure the noise at the zener's anode. Interestingly enough, the noise is very stubborn and requires a huge amount of capacitance to shunt it away.  A high voltage zener can be made by using a low voltage zener, a large valued (but low voltage) electrolytic capacitor, two resistors and a high voltage transistor. This circuit is a simple shunt regulator. The advantage it holds over a high voltage zener or string of lower voltage zeners is a easily tuned break voltage, much lower noise, much higher wattage, and a slow turn on. The last attribute is vital to many circuits. Because the capacitor is very large, say 10 kF, it slowly charges, which in turn slowly

ramps up to the break voltage. Of course this will put some extra demand on the series resistor, but this resistor's rating should be high enough to allow shorting it to ground to be perfectly safe.

   The zener ca be replaced by a precision IC voltage reference. This part upgrade will certainly make for a more exact reference voltage, but care must exercised with the selection of the capacitor value, as many IC voltage reference cannot withstand the high reverse current flow from a large valued capacitor at shut down.
    I prefer using a PNP transistor, if the required break voltage is not too high (over 250 volts), as the collector is directly connected to the case or tab, which means that these elements will be at ground potential and will not present a shock hazard or require a isolation washer between the heat sink. Of course, if the virtual zener is going to be used in the negative leg of a power supply circuit, then the NPN transistor is a better choice, as its case or tab will be at ground potential. If two transistors are cascoded to extend the break voltage, then at least one transistor body or tab will be charged.
  While on the subject of zener diodes, it would not hurt to mention that most zeners are not in fact true zener diodes, as true zener effect (not just reverse avalanche breakdown) occurs only below 6 volts. True zeners have a voltage drift with an increase temperature that moves in the opposite direction of the transistor.

pg. 17

<PREVIOUS   Copyright © 1999 GlassWare. All Rights Reserved