1
\$\begingroup\$

I am designing simple monitoring application circuit, used 3analog inputs. Im confused that how to choose decoupling capacitor for that dc voltage stability and analog inputs. What kind of capacitor need to add, how to choose?

\$\endgroup\$

closed as unclear what you're asking by Leon Heller, PeterJ, Joe Hass, Andy aka, Chetan Bhargava Apr 12 '14 at 17:39

Please clarify your specific problem or add additional details to highlight exactly what you need. As it's currently written, it’s hard to tell exactly what you're asking. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

  • \$\begingroup\$ I dont know exactly but i can tell you in my designs and i know used a lot everywhere, in a switch you would use a 10nF capacitor, non polarized. \$\endgroup\$ – Electropepper Apr 12 '14 at 11:25
  • 1
    \$\begingroup\$ There can be a lot of considerations depending on the application and circuit. Did you have a schematic and/or further details to share? Otherwise it's probably too broad for a reasonable answer. \$\endgroup\$ – PeterJ Apr 12 '14 at 11:36
  • \$\begingroup\$ I cannot tell you a concrete answer. I can only share my experience. If we use IC with push pull output, and when output changes states, there comes a moment, when the current requirement increases dramatically. We put a capacitor across the power lines, to shield other ICs from this spike. I generally use 100uF with 100nF capacitor in parallel. Sometimes, when I am not sure, where to put a decoupling cap, I touch the power lines with bare fingers and see if the result improves, if so, I put a 10 - 100nF cap there \$\endgroup\$ – Arjob Mukherjee Apr 12 '14 at 11:37
  • \$\begingroup\$ possible duplicate of Decoupling capacitors: what size and how many? \$\endgroup\$ – Chetan Bhargava Apr 12 '14 at 17:39
6
\$\begingroup\$

Without any particulars, a 1 µF ceramic capacitor placed physically close between the power and ground pins of the device you are trying to decouple is a good generic answer.

Back in the pleistoscene, 100 nF capacitors were commonly used. 100 nF would work for most decoupling situations too, but the choice of capacitance was dictated more by what was possible to make cheaply. Today with multi-layer surface mount ceramic caps being cheap and plentiful, there is little harm in increasing the capacitance a bit. With proper layout, these will have less series inductance than the 100 nF thru hole caps of ancient legend.

Take a look at the impedance curves of a family of modern capacitors. You will see there is little penalty in using higher capacitance. 1 µF is enough to to the job, and are jellybean commondity items at normal logic voltages.

In one case I used a specific model of 100 pF cap for decoupling because it was a RF application, and that cap had a lower impedance at the RF frequency than any other I could find.

\$\endgroup\$
  • \$\begingroup\$ Pleistocene epoch was so exciting with unprotected reversed electrolytic decoupling caps, firing up on smoke tests. It was like a Chinese New Years with firecrackers if there were a bunch all reversed. \$\endgroup\$ – user38637 Apr 12 '14 at 14:58
  • \$\begingroup\$ @Joe: Electrolytic caps were ever used for decoupling. They didn't then and still don't today have low enough impedance at high frequencies to be useful in that role. \$\endgroup\$ – Olin Lathrop Apr 12 '14 at 18:09
  • \$\begingroup\$ But they were good for decoupling below 1MHz. Did u thunk decouple V+ from input was RF only? \$\endgroup\$ – user38637 Apr 13 '14 at 3:17
  • \$\begingroup\$ @Joe: Below 1 MHz it's not really "decoupling" anymore. Now you're getting more into bulk power supply smoothing, having little to do with any individual power consumer. Put another way, you can just as well lump all the capacitance for 1 MHz or less together to a single point at the power supply feed. \$\endgroup\$ – Olin Lathrop Apr 13 '14 at 14:30
  • \$\begingroup\$ Coupling is defined by capacitive or inductive effects. Decoupling is similar but inverse. In my mind, decoupling can be at any frequency if there is crosstalk from supply to signals or signals to supply which can interfere with other signals. Others think of it as either suppressing conducted noise going in, others going out, when in fact it is for both reasons. \$\endgroup\$ – user38637 Apr 14 '14 at 6:25
1
\$\begingroup\$

Capacitors used for decoupling means the supply noise source is suppressed by the lower impedance at high frequency so that does not couple to the amplifier signal. Generally low series resistance, low inductance, bulk multilayer ceramic and film types are best in SMT. You don't care about tolerance as much impedance vs f, which for audio and low power is easy to accomplish because the current is low. Cost is a major factor, when are used in many places, so decoupling types are cheap, but low impedance over range of interest. This used to be occupied by two cap types in older voltage regulators with more ac noise. One electrolytic large for low frequency noise and one small for RF noise. Now it depends on your supply noise what you need, as there is no one solution. Too little and too much uF, can be bad for some regulators and don't care for others.

It is wise to follow recommended practice in datasheets.

\$\endgroup\$

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