hey guys, quick question which just occurred to me after forgetting it for ages. On voltage regulators (or other electronic equipment) where noise removal is necessary, I often see 2 capacitors instead of just one. We recently constructed a frequency meter as a project and the voltage regulator we used (i forget the model number) had two capacitors on the output of the regulator. One quite large, around 4.7 uF and a smaller one.

Why are both capacitors necessary? I know that capacitive reactance is calculated with the formula 1/(2πfc) and parallel capacitances add.

My lecturer told me this was to filter different signals and if i recall correctly, the capacitors were of different types.

Could someone perhaps just nail this down a little more for me please?

 

A big capacitor has inductance so it is poor at high frequencies. But it is good at low frequencies.
A little capacitor has low inductance so it is good at high frequencies. But it is poor at low frequencies.

 

Hello,

Do you have a schematic?

Most times the capacitors are there to prevent the regeulators from oscillating.
Oscillations can make the regulator go hot and drive them in thermal shutdown.

Greetings,
Bertus

 

thanks for that, makes perfect sense now.
One thing, would you mind elaborating on the small capacitor. You said that it was poor at low frequencies. Why is that?

 

Because it is a small capacitance. At low frequencies it has a very large reactance, and might as well not be there.

 

This looked interesting in that it shows that paralleling caps might have unexpected results ...

http://books.google.com/books?id=nZ...3oCrDw&sa=X&oi=book_result&ct=result&resnum=1

 

it is primarily for ESR reduction. large (eletrolytic) capacitors tend to have high ESR, which makes them unable to deliver lots of current and it may cause oscillation through the parasitics in the circuit.

smaller capacitors, especially tantalum or film capacitors have very low ESR and by paralleling a large cap with a small (fast) cap, you can reduce the effective ESR of the combined "capacitor".

whether you need the smaller one depends on your design. I typically use as small as a capacitor for decoupling (4.7u - 47u typically) as I can to stay with just one capacitor.

Because it is a small capacitance. At low frequencies it has a very large reactance, and might as well not be there.

wrong answer.

 

One additional gotcha that does not apply in this case, but I have seen it in other complete schematics, is the practice of showing multiple, parallel decoupling capacitors (such as 0.1 uF ceramics) on the output of the voltage regulator. In that case, one must read the description or just know that they belong close to the IC's they are intended to decouple.

John

 

Thanks guys, I am feeling enlightened!

 

One additional gotcha that does not apply in this case, but I have seen it in other complete schematics, is the practice of showing multiple, parallel decoupling capacitors (such as 0.1 uF ceramics) on the output of the voltage regulator. In that case, one must read the description or just know that they belong close to the IC's they are intended to decouple.

John

this is particularly true for motherboard manufacturers. you will see capacitors and sometimes regulators strategically located on the board to help provide dynamic current locally.

 

well, i use 3 on output of my regulators. a larger electrolytic (around 220uF), a xr7 1k-10kpF ceramic, and a 10k-100kpF poly film. more so for noise suppression over transient 'well' supply. i also try to use a ceramic and poly next to each IC's supply pin.

 

well, i use 3 on output of my regulators. a larger electrolytic (around 220uF),

(linear) regulator datasheets typically give out the value for the output capacitor.

in general, you should stick with that far more strictly than you would with input capacitors. most linear regulators are emitter followers. and the output capacitor, together with the load, is the load for the emitter follower.

so putting tons of output capacitance isn't necessarily a good idea.

 

(linear) regulator datasheets typically give out the value for the output capacitor.

in general, you should stick with that far more strictly than you would with input capacitors. most linear regulators are emitter followers. and the output capacitor, together with the load, is the load for the emitter follower.

so putting tons of output capacitance isn't necessarily a good idea.

well, my Fairchild 78XX doesnt require any caps to run stable at maximum ratings. they do suggest a input cap to kill input ripple (as large as required), and they do suggest a output cap for stability and transient reponse (some datasheets have noted 0.33uF min for generic 78XX). and yes, in some cases having too much capacitance on the output can be a problem if the input voltage has a transient drop and the output cap doesnt discharge fast enough (hence Vo is suddenly greater than Vin), but this can be handled with a bypass diode across Vo and Vin.

National makes a note for output cap "Although no output capacitor is needed for stability, it does help transient
response. (If needed, use 0.1 μF, ceramic disc)."

good info, thanks for spotting this caveat.

 

the concern here is more on loop stability because a (linear) voltage regulator is simply an opamp with an emitter follower that function as the regulator - we are not talking about shunt regulators here.

adding capacitance to the output (the emitter follower's emitter) will always introduce phase shift and weakens stability. there is also an issue of large current going through the b-e junction of the regulator if the ESR of the output capacitors is too low.

for that reason, you may see people using shunt regulators in high-end audio but that topology has its own issues as well.

end of the day, too much of a good thing is bad.