I'm using a 12v power supply to operate a system of sensors that is very sensitive to fluctuations in the voltage input. Unfortunately, I can't control what goes on upstream of my components and this is causing a few issues:

1) On occasion the voltage will dip or spike to levels that are outside my usable range (I've observed fluctuation from 10.5v to 13.5v). The change doesn't last long, but it's enough to mess with my electronics.

2) Additionally, the power line seems to pick up a not-insignificant amount of low voltage interference (EMI?) that I can't control wholly via software.

To that end, I would like to introduce a regulator/filter to control my input voltage - keeping it as close to 12v as possible and eliminating the outside interference. I don't have a lot of space to build up a whole new circuit or introduce custom components, so I'm trying to use as many off-the-shelf parts as possible.

My thoughts:

A) I've used buck converters in the past, and I'm familiar with the operation of boost converters. Could I use a buck converter (dirty 12v -> 5v) and then a boost converter (5v -> clean 12v) to deal with the large fluctuations? I'm also trying to keep the current draw in mind as the power supply has a 10amp fuse in my circuit. What kind of impact would this have on the available current? On heat dissipation in a closed environment? One of the reasons I'm leaning toward this is because I may have need of a clean 5v power line later, and this takes care of the conversion.

B) Could I ground out some of the noise with a dc power-line filter? I haven't been able to identify the source of the noise, so I'm hoping to go as broad as possible - would a filter to reduce common and differential mode noise be a good start or should I investigate other options?

I know that my points are rather vague - I'm still in the early stages of putting my sensor system requirements together, and I'd like to protect against as much as possible now, while I still have easy access to the raw power line.

Any advice / direction is greatly appreciated!

  • 2
    \$\begingroup\$ Can you provide more information on this? Specifically, when you say the "change doesn't last long", how "long" is it? A few micro? A few mili? Better yet, provide a picture of your oscilloscope measurements. Using switch-mode converters wouldn't really help since they would probably introduce more noise, so you'd need adequate filtering anyways. Also, maybe you don't need any extra hardware if the frequency of this noise is not the same as your signal's, since you could filter this noise with a digital filter (I'm assuming you have some kind of microcontroller since you mentioned "software"). \$\endgroup\$
    – Chi
    Oct 19, 2018 at 22:08
  • \$\begingroup\$ A 1 Henry choke and 1Farad capacitor resonates at 0.16 Hertz and will reduce the size of most dropouts, as well as reduce the high-frequency trash. Perhaps you could use 1H and 10 milliFarad, resonating at 1.6 Hertz; you should plan appropriate dampening. \$\endgroup\$ Oct 20, 2018 at 3:43

3 Answers 3


The 12V → 5V → 12V is going to take up quite a bit of space and may waste power. Might be worth looking into other options first such as:

  • a 12V regulator (78xx)

  • a reverse biased zener diode. These are simple, cheap and fairly effective. If your circuit needs to be very temperature stable you may want to look at placing two zener diodes in series, one forward active and one reverse biased. Generally speaking, the larger the nominal zener voltage, the larger the temperature coefficient

  • if you have a digital circuit or any circuit with lots of switching you’re going to need a capacitor (or a few) to help counteract parasitic inductance (check out dc bus capacitors). But capacitors in general are a good idea for stabilising your input voltage, especially when you turn the circuit on

  • if EMI is an issue try checking out using twisted pairs or coaxial cables

I could be wrong but what I gathered from your question is that the 10A fuse is in the power supply, and not in your sensor circuits. Presumably, your sensor circuit requirements would be quite low.

If they are below 1A, and the power supply is variable, I'd suggest 7812 Voltage Regulator IC and running it at slightly higher voltage, say 17V. I think it'll give you your required stability in input. If the power supply is fixed and you cannot change that, maybe a boost converter and then 7812?

P.S. Remember to attach a heat sink with the IC.

  • \$\begingroup\$ 78xx regulator series is old and has pretty bad PSRR. I would suggest a more modern regulator. \$\endgroup\$
    – Mike
    Oct 20, 2018 at 1:10

How about powering it from 24V supply? And what about power - how much do you use? I mean, you could have even a linear regulator down to 12V if it's just hundred(s) of milliamps, waste of power will be (relatively) tolerable. And with some really small caps to filter out 24V's switching supply noise in addition to some regular 1uf or maybe even more, you could probably get good results. And bypass everything. Anyone correct/add?


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.