8
\$\begingroup\$

My system contains several switching and linear regulators:

  • Main DCDC regulator taking the incoming 50v power down to 6v.
  • Several small regulators (DCDC and linear) creating 3.3v, 5v, etc. from the 6v.

Power distribution

If the small regulators are very close to the main regulator (and thus very close to its output capacitor), do they all need their own input capacitors?

Could they share the main regulators capacitor?

Could they all share one input capacitor?

\$\endgroup\$
3
  • 2
    \$\begingroup\$ Why the down-vote? Are these bulk capacitors or bypass capacitors? This question would be interesting with only "bypass" capacitors, too. \$\endgroup\$ Commented Feb 11, 2013 at 13:07
  • 4
    \$\begingroup\$ @LeonHeller - I thought this question would be more useful to the community if it could be asked in a general way. If the answer is that it totally depends on the particular regulators, then say that in an answer rather than getting all upset. \$\endgroup\$ Commented Feb 11, 2013 at 13:17
  • 3
    \$\begingroup\$ What do the data sheets say? Probably nothing about that specific arrangement. \$\endgroup\$
    – Kaz
    Commented Feb 11, 2013 at 14:08

3 Answers 3

6
\$\begingroup\$

If the ICs are close enough the input capacitors for the second stage converters COULD be combined with the output capacitor of the 1st stage converter.

However, I'd be cautious about using a minimum value capacitor here as there is the potential for dynamic interaction between the various converters - especially so with 3 combined if all are smps. I assume that the 6V/5V regulator is linear as efficiency is then 5/6 = 83%. You can get somewhat more than that with a smps with care but to get much over 90% you'd want a synchronous converter and a limited range of loads.

At a minimum I'd suggest that, if a common capacitor was used, the value should be a minimum of the sum of the values required by the 3 individual designs - and then increase as much as possible above that.

Some regulators may require ESRs in a specified range - with lower and upper limits. You may have to look at the conceptual pole zero diagram which drives this specification and see if you can discern any likely effect from inter regulator interactions. OR, more easily, use a largish cap that does not violate any of the spec sheet values and see how it goes :-).

\$\endgroup\$
6
\$\begingroup\$

If they are linear LDO regulators, within an inch or so of the main, they probably can share safely.

If they are switching regulators, a low value (say 10nf, maybe 47nf) capacitor directly across their input pins will reduce the loop area for high frequency transients and reduce RF emissions. However they should be OK with sharing any larger reservoir capacitor (10uf or larger).

This is just guideline advice : any specific advice in the datasheets may override it.

\$\endgroup\$
1
  • \$\begingroup\$ The situation with two switching regulators might couple with each other on switching harmonics if they only have one cap, which 2 caps with ferrites or the inductance of distance would possibly help filter the interactions. \$\endgroup\$
    – Kortuk
    Commented Feb 11, 2013 at 18:47
3
\$\begingroup\$

For the big ones, caps in the uF range, who act as current reservoirs, they can be trustfully lumped if the regulators are not spread more than a few cm. For the small ones in the nF range, acting more as high frequency shunts, I suggest to keep them by the respective IC as the impedance of a piece of trace can be important at high frequencies. And as they are SMD and cheap, should not be a problem to keep them duplicated for the sake of "safety".

\$\endgroup\$

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.