# Sources on designing a precise feedback loop for designing a linear power supply [closed]

I am a student, and studying Electrical Engineering now. I was tasked with building a 24VDC to 12V dc linear power supply. One of the requirements was big precision of line/load regulation (up to the fourth decimal). From what I read, this is achievable by using a feedback loop, however, I cannot find any reliable sources on how to do that.

• Studying op-amps is a good start. Then study precision voltage references. Then study precision resistors. For all of them study effects of temperature. Commented Nov 28, 2016 at 14:34
• @Andyaka: Maybe studying electrical engineering would work too and is easier... Commented Nov 28, 2016 at 14:37
• Google control theory Commented Nov 29, 2016 at 0:28

Precision of a control loop under static conditions is mostly to do with the open-loop gain of the feedback controller and DC errors due to offset and bias currents. All can be made relatively small by using precision components.

That means that your voltage will match the reference within a component of error that is inversely proportional to the open-loop gain of the feedback controller. There are also errors arising from the offset voltage and changes in the offset voltage of the feedback amplifier, and from the amplifier bias currents.

Of course errors in your reference voltage are going to be directly reflected in the output voltage. If there is a voltage divider from the output voltage to your reference then any error in the ratio will be reflected in changes in the output voltage. Offset voltage errors in the amplifier will be multiplied by the inverse of the division ratio.

You can read more about the effect of loop gain here in Analog Devices Mini Tutorial Op Amp Open-Loop Gain and Open- Loop Gain Nonlinearity. The below schematic shows some sources of error other than loop gain.

simulate this circuit – Schematic created using CircuitLab

Control loop stability and performance under dynamic conditions is not a small subject- you could easily spend a few semesters studying it and still only be scratching the surface. When the input changes or the load changes quickly you will generally see a transient error that is larger than the steady state error as the loop corrects. It may overshoot or undershoot, depending on how the control loop is tuned. If tuned to prevent overshoot or undershoot it will generally be more sluggish in response.

It depends whether you want static accuracy, or dynamic accuracy.

In order to get static accuracy, you only need very large gain in your feedback loop (in addition to stability of course). This basically means including an integrator in the feedback path, so that at DC, you have the full open loop gain of your feedback amplifier available. After a while, when everything has settled down after a change, you ought to be able to get 4 digits of repeatability at your output terminal with no problems. To turn that repeatability into accuracy, you need a stable reference, and to take care about where you sense the output voltage with respect to voltage drops due to output currents (including the ground sense!)

If you need dynamic accuracy, so that under a line or load transient the deviation does not exceed 1:10000, then good luck, that will be a very interesting problem indeed. You will need to understand loop gain and stability interactions very well. Some topologies are easier to stabilise than others. You will need to define the range of impedances your load can present. The wider range of loads it must work with, the more 'interesting' the design challenge.

A lot of work has been done and is available in the application notes of the big manufacturers (TI, Linear, Analog) on voltage regulators. Read them.