Take the 2-minute tour ×
Electrical Engineering Stack Exchange is a question and answer site for electronics and electrical engineering professionals, students, and enthusiasts. It's 100% free, no registration required.

I hear the ripple current rating of a cap is important because caps have internal resistence which can cause over heating of capacitors. I was wondering is there any general rules a professional engineer would follow when designing a linear supply?

share|improve this question
    
This question may be of interest: electronics.stackexchange.com/q/27881/1609 –  DarenW Jan 15 '13 at 4:49
add comment

2 Answers 2

up vote 2 down vote accepted

The short answer is you calculate the RMS current the capacitor is going to see, and then choose a capacitor with an RMS current rating that is higher. In some cases this is easy, in most others it is not exactly easy.

Here is easy:

  • Output capacitor for a buck regulator. You can use V = \$\frac{L\text{ di} }{\text{dt}}\$ to determine the peak to peak current the capacitor will see. With a little experience you know that the waveform is just a ramp-up and ramp-down sawtooth, from which you can estimate pretty accurately the RMS value.

  • Low Dropout (LDO) linear regulator. These parts have pretty strict specifications on filter capacitors, otherwise they aren't stable. So, the data sheet will define requirements for capacitor value, and often ESR (not too much, not too little, just right).

That about sums it up for easy.

The process to choose proper capacitors is just part of the design process:

  • Start with a design topology (a really top level or broad approach).
  • Make a simplified analytical model. Actually, make several for different parts of the supply. Analytical models are good to work on concept and think things through. A really well done analytical model should get you about 80% of the way. The form of the model will likely be Laplace equations that have been idealized just enough to still be meaningful and yet useful. If switches are involved you will have two or three representations of the same circuit to account for the states of the switches, but these multiple representations will really help you think about how the circuit works and what kind of ripple currents the capacitors are likely to see. If things don't look promising, throw the design away or make whatever changes needed to improve things and start over.
  • Make a numerical model. Use something like SPICE or something with Modelica extensions (or SABER if you can afford it). Make sure you know the basis of the models (you know where they are valid). By the time you get this far you should know basically what to expect (and you have probably also started some circuit lab work), so you should be able to look at the numerical results and know if they are BS. If the results look reasonable you can have the simulator calculate the RMS currents to guide your choice of capacitor.
  • Always verify with your first bread board or prototype. The earlier the better.
  • Some lean much more heavily on lab work to get things worked out. Just remember that there are a bazillion ways to get things wrong compared to only a few to get them right.

This is an iterative process. Hopefully this will give you some idea of what lengths people go through to get good electronic designs (and good at designing).

Try a design. Realize you are going to make lots of mistakes. Get some specific results, either theoretical or in the lab, then you will have focused questions that can be given specific answers.

share|improve this answer
add comment

Yes, read and follow the datasheet, of course. Capacitors intended for such applications wil have a ripple current spec. Follow it.

share|improve this answer
    
Yes but how would you know that the 35ma of ripple the capacitor is rated at, is what's on the circuit you're trying to filter? If it's 100ma then that won't be good for the cap right? –  Ageis Jan 15 '13 at 17:29
    
@Aegis It's called "engineering". A proper design will include accurate ripple current calculation. –  Madmanguruman Jan 15 '13 at 18:13
    
@Madmanguruman but the question is how to work that number out! –  exscape Jan 15 '13 at 18:34
    
@Ageis: That's a separate question. To explain that right requires more room than a comment allows. –  Olin Lathrop Jan 15 '13 at 18:40
    
Ok shall I start new question or can you refer ,me to some lititure I can read up? –  Ageis Jan 15 '13 at 20:52
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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