In the top Idea for Design of the year, “An Easy Way To Roll Your Own Programmable Power Supply,” Donald Schelle and Marc Davis-Marsh discuss using a digital-to-analog converter (DAC) as a variable voltage control for a switching dc-dc regulator in a closed loop. But there’s a challenge.
Looking at the schematic, at the control pin on the regulator, you have a high-Z input. A resistive voltage divider connected to this node feeds back part of the regulator’s output voltage. In addition, the DAC output is applied to the same node again with a voltage divider, but this voltage divider uses the same lower resistor as the first voltage divider. The task is to calculate the three resistor values, accounting for the maximum output voltage of the regulator and the available voltage swing at the DAC’s output.
The Ideas for Design are among Electronic Design’s most popular features, not because hundreds of readers are plagued with the identical problem, but because they like to test themselves against other engineers. “Is this as elegant a solution as possible?” they ask. Or it may be: “Am I smarter than these guys?” IFDs are Electronic Design’s equivalent of the Sunday crossword for engineers. This IFD is proof.
Schelle and Davis-March applied Kirchhoff’s current law to the node at the regulator control pin, assumed an inverse linear relationship between the DAC output voltage and regulator output voltage, and wrote the node equations. Then, as we used to say back when we were taking EE201, it’s a simply matter of “turning the crank.”
If it weren’t for a tiny typographical error in the original publication of the IFD, I would have accepted the crank-turning as a given. As it happened, in the equation that sums the currents into and out of the Kirchhoff node, there was an “=” where there should have been a “+.” (They’re the shifted and unshifted symbols on the same key, but we should have caught it.)
That made it necessary to check the whole derivation. What if there were a more embarrassing typo further down? It turns out there was: an extra term had crept into Equation 8. These are corrected below, and we’ve fixed the version on the Electronic Design Web site. And that left one final question, which Editor-in-Chief Joe Desposito pointed out. In the derivation of the equations, the authors illustrate the point that one must assume a value for one of the resistors to calculate the other two values by postulating 10 kΩ for R1 and then deriving expressions for R2 and R3.
Then, in the IFD’s real-world real design example with its bill of materials, R1 is not 10 kΩ, although the calculated values for R2 and R3 are correct, given the value of R1 they used (assuming you use the correct version of Equation 8, which Joe didn’t have).
If you’re hardware-oriented, you’ve probably figured out the answer to that one already. It’s necessary to set up the equations and plug in standard resistor values for R1 until R2 and R3 come out to standard values too (within tolerance).
The Idea for Design readers voted most popular, then, is not about something that is not obvious “by inspection” as our professors used to say. You don’t “get it” until you think about it. That supports my hunch about what makes the feature so popular.