LDO input/output capacitor - is bigger better?

[Saimoun]

Hi

Quick question: is it generally a good idea to have bigger caps around an LDO?

Taking the NCP160 as an example (https://www.onsemi.com/pub/Collateral/NCP160-D.PDF), in the datasheet they do all their measurements with 1uF MLCC for the input and output cap.
If on my board I happen to use a 4.7uF or even a 10uF MLCC, will that improve the noise/PSRR performances, compared to the ones shown in the datasheet?

And are there any "negative" effects of having bigger caps?

PS: I know MLCC caps with large capacitance are affected by dc bias - but a 1206 10uF cap with a 5V dc bias still has about 8uF capacitance.

[Siwastaja]

Excess on the input doesn't matter (as long as you are aware of the issues a large MLCC on the input itself generates, regardless of the regulator; namely inrush voltage peak and mechanical).

On the output, however, carefully follow all advice given by the regulator datasheet. Depending on regulator, they tend to give ranges of acceptable capacitance and ESR. Note that whenever they suggest an "MLCC", they already assume some kind of "average" MLCC DC bias effect; if they say 2.2uF, they have designed and tested it with a 2.2uF nominal MLCC, and do not require actual 2.2uF. You need to work with some assumptions here, though. They likely won't mean the crappiest Y5V part.

[exe]

The positive side is clear: better transient response.

Negative sides I can think of: longer startup time, higher ESL meaning possibly less rejection in higher-frequency region (1MHz+), high initial inrush current. Also current regulation becomes worse, but it's probably not a problem unless you are using it as a lab supply. Sometimes there is a limit on how much capacitance you can put due to (in)stability issues. Beware that physically big ceramic caps are fragile. I think I once seen in a datasheet that placing too big output cap can cause reverse current issue and requires external protection (diode). Check datasheet for this.

Practically I don't think it will hurt except some specific cases.

[Saimoun]

Alright thanks guys that helps 

The NCP datasheet actually mentions the minimum capacitance for stable operation (0.7uF). It also says "Larger output capacitors and lower ESR could improve the load transient response or high frequency PSRR", without mentionning a max value.

In my case I will be using 4.7uF 0803 caps, unless I misunderstood you I think will work perfectly fine

[magic]

There is only one answer
CTR+F and look for things like "stable", "stability", "capacitor", "capacitive" through the whole datasheet.

Sometimes in "typical characteristics" there is a plot with C and ESR axes and stable/unstable regions shown.
Sometimes they say it's stable with any MLCC of at least X capacitance and that's it.

[David Hess]

Quick question: is it generally a good idea to have bigger caps around an LDO?

Transient response is improved with more output capacitance.