# Pros and cons of low-power integrated series voltage regulators vs. shunt regulators

TL431 (and equivalent) shunt regulators are jellybean parts nowadays and quite cheap. The same can be said of TL317 (low-power version of LM317) adjustable series regulator.

When applied to supplying power to light loads (<100mA) they are both quite effective and simple to use. Despite their different internal topology, it seems that they can be used in the same role quite interchangeably.

I was wondering what are the pros and cons of using one or the other, or, more generally, what are the pros and cons of using a low-power shunt vs. series linear regulator. What are the design decision that can make a designer choose one over the other?

I'm aware that, for example, series regulators may have considerable drop-out voltage (2.5V for TL317), something that shunt regulators don't have (but may have higher minimum bias/quiescent current, instead). But taking into account LDOs regulators, this doesn't seem to be so compelling nowadays.

In the end, are there conspicuous advantages of one topology over the other (and in which case), especially when powering low-power loads?

Note: direct comparison of the two jellybean parts I've mentioned is appreciated as well as a more general comparison between topologies. You choose. The important thing is I'd like practical considerations based on actual design experience or knowledge on reasonably current products or systems. In other words "why and when a design engineer would use one over the other".

• Apparently, the designers that wrote the examples in the TL431 datasheet would never use it as a regulator (but as a reference). – CL. Jan 7 '17 at 8:48

To me the critical difference is how they handle loads with varying power requirements.

A shunt regulator will always take the same power from the supply, even if the current drawn by the load is reduced. In some scenarios this could be a positive feature.

A series regulator will take less power from the supply if the load current drops. In some scenarios this could be a positive feature. This also makes the series regulator more tolerant of sloppy design --- you don't have to know exactly the maximum and minimum power that might be needed by the load in order to design with the series regulator, and your design is not penalized (with extra heat generation) if you overestimate the load's current requirement.

A shunt-regulator, fed from a constant-current source, can produce some of the quietest voltages on earth, with excellent rejection of ripple at all harmonics of 120Hertz FullWaveRectification.

Example is the diyAudio forum/thread "Simplistic NJFET RIAA" phonograph PreAmp. Their goal, with the ShuntReg, is not precise VDD but very quiet VDD.

I've read many of the early posts; there are 16,000+ posts over the last 11 years, and I've concluded the noise floor of their ShuntReg is approximately 1 nanoVolt for 60/120Hz hum.

The very first post of that thread has a PDF with schematics for the RIAA Preamp and for the ShuntReg. Lots of design effort goes into raw_DC design, transformer selection, and Grounding. Some of the DIY folk write about the absolute silence coming from their speakers, with the Gain at maximum and the phone cartridge lifted from the vinyl.

Here is one of the ShuntRegs, used to provide power to the 2-stage NJFET RIAA; the 24v DC output would be for a MOVING MAGNET preamp needed lowgain.

Here is an example of the RIAA preamp, for 250uV MOVING COIL, needing highgain thus high VDD (large Rdrain values)

In the ShuntReg, planning the Ground is big-deal, so the rectifier-diode current surges do not cause I*R drops in copper foil shared by the "clean" output voltage. The "diyAudio" experimenters have converged on remote-boxes for the power-cord, transformer, rectifiers, and first raw-DC filtering; often they'll spend  on inductors for raw-DC filtering, to further slow down the current surges.

Then a 2 meter DC_power cable brings the unregulated DC into the PreAmp box, where ShuntRegulators (one for Left channel, one for Right channel) reduce the DC down to extremely quite (1nanovolt) noise floors with excellent stability so the PreAmp NJFET transconductances remain constant and the LeftRight gain balance remains constant (within 0.1 dB, to maintain sound stage).