[SOLVED] Battery power circuits and bypass/decoupling capacitors?

[David_]

Hello.

I'm unsure about battery power circuits and how much capacitance/capacitors I need, I am often fiddling with analog circuits and I as standard use a 0.1µF capacitor next to each IC. 10µF/0.1µF if its a precision IC to be sure.
And when I have a regulator I usual put 10-100µF at its output and perhaps another 10 or 22µF further down the rail if it spans a few centimeters. But this is from ciruits assumed to be powered by a transformer-rectifier.

In trying to ask this question I find that I don't have any clue about how much capacitance I need where, but I have to think that when battery power is applied I can ease up on bypass capacitors?

I know that capacitors is often used as a fast-reacting power sources to supply ICs as the rail can't deliver the power in the time frame needed, but I would like to get a firm grip of how much/high capacitance I need to use, I have two situations.
1, Analog precision circuits.
2, Digital circuits.

Is it possible to offer some more detailed guidelines regarding this subject?

 

[betwixt]

Usually when a power supply is used it will have regulation and low output impedance by design. A battery is far more unpredicatable, especially as it discharges and ages, the impedance across it's terminals tends to rise and the capacitors are actually more important.

There is no exact science about which values to use but in general, using more does no harm. Think of all the supply wiring as being little inductors and therefore dropping voltage as the curent through them changes. Then consider that almost all ICs draw a current which changes as the signal they carry changes. Keeping a local reservoir of energy close to hand becomes essential. There is also the case of IC with built in oscillators where the circuit relies on the supply and ground being at the same AC potential. Any resistance or inductance in the wiring becomes part of the oscillation path and can distort or alter the signal. Obviously as the frequency increases so does the need to use good capacitors.

With most regulators, it is more important to place the capacitance close to the input pin than the output side, although some on the output is always a good idea. If you put too much capacitance immediately after the regulator it can slow down it's response time. The values you are using seem quite reasonable though.

Brian.

 

[dick_freebird]

As a simple charge reservoir, you would like the supply
voltage to remain in spec (+/-5%) without demanding
that the charge taken by switching events be put back
instantaneously by the remote source (it can't).

If you took the dynamic Idd (less static) at the clock
rate the spec is tied to, you could calculate a charge
slug per clock (remembering to account for outputs with
capacitive loads, which dynamic Idd may spec at low or
zero value) and then dQ=CdV and dV is <= 5% of supply.
So you could figure a capacitor which, with a constant
current supply equal to average current drawn, keeps
your supply perturbation within recommended range for
the component.

Then you'd worry about how ESL/ESR degrades that,
and maybe go bigger and better quality.