There are several ways to implement the current/voltage control for an SMPS. The easy way I see is to use an ASIC/PMIC to manage the control loop. One of the other methods would be implement a PID control algorithm using a microcontroller and control the duty cycle of an SMPS.

I need some help in understanding the pros and cons of the above two methods in order to arrive at a justification of selecting one of the two methods.

Besides, if in case, there are control/communication/dimming requirements, then a PMIC with an MCU could also be used.

Please support me with some comparison.

  • 1
    \$\begingroup\$ I'm not sure I see any real difference between the two, other than the level of integration... \$\endgroup\$ Commented Sep 22, 2015 at 5:59
  • 3
    \$\begingroup\$ Advantages of Analog: speed, reliability, accuracy, fault tolerance (cannot get stuck on software bug). Advantages of digital: more flexibility after building hardware. In high-power SMPSes I dislike using an MCU, what if the software crashes and keeps the power FET on all the time. In an analog implementation the current limiting would prevent such disasters. \$\endgroup\$ Commented Sep 22, 2015 at 6:59
  • \$\begingroup\$ ASICs are very expensive unless you need thousends of them. There is a 3rd possibiliy: FPGAs. But I think a ready to use PMIC is cheap and much less error prone. \$\endgroup\$
    – Paebbels
    Commented Sep 22, 2015 at 7:01
  • 1
    \$\begingroup\$ A dedicated SMP chip is always my preferred solution. \$\endgroup\$
    – Andy aka
    Commented Sep 22, 2015 at 7:33
  • 3
    \$\begingroup\$ Although full digital control has flexibility, if the PID algorithm has a bug, the output may even oscillate, another type of disaster to be avoided. Dedicated controllers are plentiful and cheap. \$\endgroup\$ Commented Sep 22, 2015 at 7:41

2 Answers 2


These are my general thoughts: -

  • Easiest - power management IC (PMIC)
  • Smallest footprint - PMIC usually
  • Cheapest - usually PMIC (for a certain minimum performance level)
  • Highest performance - PMIC will outshine all others
  • Best reliability - PMIC
  • Best stability - PMIC
  • Quickest reaction time to load changes - PMIC
  • Quickest reaction time to incoming voltage changes - PMIC

And now the bad things: -

  • Unreliability - microcontroller
  • Instability - microcontroller
  • Noisiest - microcontroller
  • Poor reaction time to load changes - microcontroller
  • Poor reaction time to input voltage changes - microcontroller
  • 1
    \$\begingroup\$ Software maintenance - Microcontroller \$\endgroup\$
    – Lior Bilia
    Commented Oct 22, 2015 at 11:36
  • \$\begingroup\$ Where will the microcontroller get its power from? Does the power supply need a power supply? \$\endgroup\$
    – Oskar Skog
    Commented Feb 17, 2017 at 13:50
  • \$\begingroup\$ @OskarSkog I feel I don't need to answer that! \$\endgroup\$
    – Andy aka
    Commented Feb 17, 2017 at 13:51
  • 3
    \$\begingroup\$ @Andy Aka: It's a rhetorical question. \$\endgroup\$
    – Oskar Skog
    Commented Feb 17, 2017 at 14:12
  • \$\begingroup\$ Everything you write is true but I feel like your answer lists reasons to "never build something you can buy an optimized off the shelf version of". If you compared a PMIC with a manually designed analog controller circuit (I know nobody does that but bear with me) you would basically get the same set of answers. \$\endgroup\$
    – jpc
    Commented Jan 30, 2021 at 12:31

Imagine: your SMPS has to operate in arbitrary load conditions, most often in pulse loads; you need almost impossibly low ripple(<5mV), transient response almost in 1 cycle or a few, least amount of reflected input ripple, stable under all variations/tolerances, no electrolytic capacitors allowed, smallest size, almost flat efficiency across line/load. Meanwhile, please report all diagnostics to the host.

Get me a "dedicated" analog chip for this app please. I know, I know, you think'n, heck I can throw in this opamp, & that circuit you did for Julie…...while adding up hundreds of components. Now, you have an unproducible unpredictable SMPS - all for a few bucks.

You get the picture.

If your uncle Harry still needs an old fashioned 13.6V @ 100 amp, and will not mind fanning it, chilling it, throwing it away when not happy, I will get you a "dedicated" 50 cent chip for sure.

Using analog approach, none of the advanced control schemes can ever be implemented nor a system built with the demands I mentioned above, only to make you feel jittery if not annoyed.

Please don't.

That is what I have to deliver.

All shortcomings mentioned as to "microcontroller" based system are null & void off the bat.


That is the catch.

  • 4
    \$\begingroup\$ Welcome to EE.SE, Robin, but I've read this twice and I don't know what your message is. Please write in plain English. e.g., What does "null & void off the bat" mean in a technical context? (I'm a native English speaker but you are on an international site.) \$\endgroup\$
    – Transistor
    Commented Sep 13, 2018 at 21:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.