There are as many techniques for precisely controlling a voltage as there are possible needs for such a technique. Here is a technique I wish I could take credit for, but you may not have seen before. The basic idea is kind of hackish: start with a normal adjustable voltage regulator (can be switching or linear, doesn’t matter) and inject a voltage onto the feedback node. The thing to keep in mind is that a regulator will do whatever it can (ie, raise or lower the output voltage) so that the voltage on the feedback node is always equal to a certain value - usually 1.25 volts. If we inject a voltage onto the feedback node, we can trick the regulator into raising or lowering its output. In reality, the idea isn’t as hackish as it sounds, because regulators are “basically” an op amp (with a high current output) and a voltage reference, which is exactly the key components that most fancy bench top power supplies are based on. The reasons why this technique is so appealing (to me at least):

• Works with any regulator type: switching, linear, buck, boost, inverting, etc.
• If the regulator is capable of it, you can produce negative voltages or a voltage which is above the input voltage by monkeying with the resistor values.
• Can generate an output voltage below Vref (usually 1.25 volts) - you cannot do this with an unmodified regulator!
• Possible to use a DAC or filtered PWM to digitally control the regulator.

Check out the attached Excel file. It will let you mess with resistor values and graph the output voltage. In the file, I have called the voltage we inject on the feedback node Vp. I’ve set up Vp to run from 0 to 5 volts with 10 bit resolution. This is only a brief intro to this idea and doesn’t represent a usable design. Maybe sometime in the future I can churn something out. Enjoy…

digitalreg.xls