7805 regulator, cap on output

[acrophobic]

Hi! First question on the forum..
Is there any advantage of putting a "large" cap on the output of a standard 7805 regulator, apart from the small decoupling cap? Or might it even be a bad idea?

[marshallh]

With 78xx linear regulators, it is much better to put the big cap on the input.
http://www.intersil.com/content/dam/Intersil/documents/an13/an1325.pdf

[acrophobic]

For sure, but already assuming a big enough cap on the input, is there any point in putting one on the output (besides from the small decoupling cap)?
Thanks for the link btw

[JSnyder]

yep, you should follow the datasheet for your particular regulator, one example of a benefit of more output capacitance is better transient response from the regulator with a rapidly switching load (ie. wireless radio, motor, anything that draws current in bursts)

[acrophobic]

Some background: I have two DAC's for converting audio from SPDIF to analog, one FiiO Taishan and one very cheap Deltaco. Both low quality by any audiophile standards, but good enough for me Instead of using the supplied AC adapters, I thought why not build a linear power supply, since I have a suitable transformer lying around. With two 7805's for the DAC's, one for phone charging and one 7806 to supply my digital TV box. That way I get rid of three or four AC adapters, I get a hopefully low-ripple supply and the joy of having built it myself

[VK5RC]

Re the output capacitor,  remember in general,  very large output capacitors also tend to have larger equivalent series resistance and a larger equivalent series inductance which tend to reduce high frequency transient responses,  it is why you occasionally see a set of capacitors of widely different values on a power supply rail for decoupling. 

[jitter]

Re the output capacitor,  remember in general,  very large output capacitors also tend to have larger equivalent series resistance and a larger equivalent series inductance which tend to reduce high frequency transient responses,  it is why you occasionally see a set of capacitors of widely different values on a power supply rail for decoupling.

Should that not be smaller ESR?

[jitter]

For sure, but already assuming a big enough cap on the input, is there any point in putting one on the output (besides from the small decoupling cap)?
Thanks for the link btw

Put enough cap is always good, but superfluous capacitance may get you in stability troubles. Despite rare in 78xx series, the large cap issues show commonly in modern LDOs, so as a good habit, do not put extra cap unless you need it, and always refer to datasheets and application notes.

Yes on the LDOs. They need to "see" a bit of ESR to remain stable.
I have run into a design in which an LM337 oscillated because too low ESR caps were used on its output. The addition of a 0R22 resistor in series with the output pin cured that problem. The 78xx series are not LDOs (Low DropOut regulators), though.

[matseng]

yep, you should follow the datasheet for your particular regulator, one example of a benefit of more output capacitance is better transient response from the regulator with a rapidly switching load (ie. wireless radio, motor, anything that draws current in bursts)

Wouldn't it be better to have the large caps located in close proximity to the device that is causing the load rather that at the vreg?

[TimFox]

Whatever you do at the output of a 7805, be sure to have at least 0.33 uF of good capacitor very close to the input to avoid roughly 100 kHz oscillation.  I normally use a 0.47 uF X7R ceramic at the input.

[Kjelt]

A normal old school 78xx has no problem with lots of capacitance on its output AFAIK. LDO are different!
Indeed make sure to have 100nF or higher caps nearby the ic itself.
The 78xx has a currentlimiter. See for instance the schematic from a Classe CP60 pre amp, and I saw lots of these overdesigned audio circuits. Its purpose probably to lower the noise.
The only thing the 78xx really does not like is if after switching the power off the output voltage remains higher then the input voltage. So in this circuit they balanced the capacitance so this would not happen but you can also put a diode (1n400x) across the 78xx to prevent this from happening.

[acrophobic]

Thanks for all the replies! Although I didn't understand all of it, it was very helpful

[jitter]

Neither LM337s are. Both are normal drop out regulators. LDOs are more susceptible to low ESR caps.

In today's standards they may not seem like particularly low dropout types, but I do count LM337 them among them. I think they should be considered LDOs of the first generation.
Whatever they're called, I consider them LDOs because of their unstable behaviour if the capacitance on their outputs has too low an ESR (a characteristic often overlooked in this ubiquitous voltage regulator, even by experienced designers).

[VK5RC]

Apologies,  jitter is correct re esr and capacitor size,  what I was trying to say (very poorly)  was the 'bigger'  type of capacitors eg routine electrolytic caps have a bigger esr than the 'smaller'  types eg ceramic,  correct me if I am wrong (again!) thanks

[TimFox]

IIRC, LDOs are more sensitive to load capacitance ESR since they have PNP collector outputs (or equivalent) so that their open-loop output impedance is higher than with "normal" regulators that have NPN emitter outputs (or equivalent).  The minimum operating voltage drop for the CE PNP is lower (essentially Vce SAT) than the minimum operating voltage for a simple follower or Darlington follower, hence the lower drop-out voltage.

[jitter]

Apologies,  jitter is correct re esr and capacitor size,  what I was trying to say (very poorly)  was the 'bigger'  type of capacitors eg routine electrolytic caps have a bigger esr than the 'smaller'  types eg ceramic,  correct me if I am wrong (again!) thanks

I see what you mean, but I don't think that is strictly true. Metal film capacitors can be quite big, especially the higher voltage ones, yet have low ESR.