High-voltage feed forward

Article By : John Dunn

Using a simple "feed forward" provision can help you create a stable feedback loop for a high-voltage power supply.

I used to work on high-voltage power supplies whose basic high-voltage generation plan was the following (Figure 1):

Figure 1 High-voltage power supply.

High-voltage multiplier ladders varied from having multiplication factors of x2 to x20 (the one shown is x8) for output voltages that ranged from a few hundred volts up to 100 kV.

The voltage control feedback loops always required the use of a “feed forward” capacitor in the high-voltage section and that capacitor absolutely HAD to be fed from the highest voltage node of the voltage multiplier, the output node of that multiplier, and just ahead of the RC high voltage filter. There was no alternative to that placement.

In the sketch above, two nominal candidates of feed forward capacitor connection are shown. A great deal of effort was put into trying to use the lower voltage node at the bottom of the “quiet” side of the voltage multiplier but trying to do so simply DID NOT work.

That investigation was sweated over for a solid month without the aid of computer simulations but the end result was unequivocally certain. Now, decades later, SPICE modeling demonstrates why the results came out the way they did (Figure 2).

Figure 2 Feed forward analysis.

The sense voltage at the low end of the 20 MΩ resistor must rise in proportion to the high-voltage output voltage and must do so quickly. Slowness of that rise equates to having pole(s) in the high-voltage feedback loop. If C11 is omitted, the feedback loop WILL oscillate. Trying to connect C11 to the top of C2 barely affects the speed of response of the sense voltage but connecting C11 to the top of C8 has a very pronounced speed-up effect. (See the lower red scope traces.)

Placing C11 in the high voltage assembly effectively bypasses the pole(s) of the high voltage RC filter, which then allows the high-voltage feedback loop to operate with stability.

This article was originally published on EDN.

John Dunn is an electronics consultant, and a graduate of The Polytechnic Institute of Brooklyn (BSEE) and of New York University (MSEE).

 

Leave a comment