14

I work with low-power DC voltage regulators. I am already aware of the formula to calculate the size of smoothing capacitor(s). This can be an iterative process of testing one size with a scope and then using a larger size or adding more until the scope shows acceptable (very low) levels of ripple and noise.

Besides the cost of the capacitors, is there any tradeoff to rounding up (a lot) and just using a very large capacitor(s) rather than trying to calibrate the sizing to "just enough" but not more than that?

Downgoat
  • 105
  • 6
PhilM
  • 291
  • 1
  • 2
  • 7
  • Are you talking about capacitors on the output of voltage regulators? – Andy aka Mar 24 '16 at 10:49
  • Yes, the output side of a voltage regulator. Thanks for all the good answers. My takeaway is that there can be some oversize margin without consequence but not too much for all the reasons stated. – PhilM Mar 24 '16 at 17:36
  • 2
    The capacitor on the output side of a voltage regulator is not a smoothing capacitor. It is a decoupling capacitor for transient behaviour and circuit stability. Te smoothing capacitor is on the input side, and that is where all the answers about transformer conduction angle apply. – user207421 Mar 24 '16 at 21:53
  • I didn't realize that. Thanks for explaining the distinction. – PhilM Mar 24 '16 at 22:44
  • This reason may be buried in one of the many good answers Some (only) regulators can be damaged by the output capacitor discharging back through the regulator if Vin is lowered to below Vcap. In particular, if power is turned off and other loads rapidly reduce Vin to near zero, or if Vin is set rapidly to zero by a fault or crowbar circuit, then most of the capacior energy may be dissipated in the regulator. Some regulators are protected against this. Older ones are less liable to be (eh LM7805/LM340). A 'fix' in most cases is a reverse biased diode from Vin to Vout that conducts when needed. – Russell McMahon Mar 30 '16 at 05:32

7 Answers7

13

As far as caps go, there are two competing requirements:long-term (ripple) and instantaneous (spike). A big electrolytic can give you the former but not the latter. Generally you parallel your large electrolytic with a smaller 0.1uF capable of supplying that instantaneous spike whilst the electrolytic lumbers into action. Or the 0.1uF may be for local decoupling to stabilise that regulator. If the specified capacitor is actually 0.1uF or smaller, then the intention of the capacitor is to supply small amounts of charge very fast. Do not replace this with a bigger electrolytic - that's definitely a case where larger is worse not better.

Going past that, you'll have to tell us what kind of regulators you're dealing with. If it's just a basic linear regulator then it doesn't really matter. If you have a switching regulator though, the capacitor will affect the resonant frequency of the switcher, so be very careful there.

Graham
  • 6,020
  • 13
  • 20
12

A larger than minimum smoothing capacitor on the output of a transformer and rectifier will give you lower ripple, which is a plus. It's a small plus however, as even doubling the size of the capacitor will only (roughly) halve the ripple. Anything downstream of a large capacitor will need to have significant Power Supply Rejection Ratio (PSRR) to cope with the ripple. There are cheaper ways of improving this by a factor of two than doubling the size of the Big Filtering Capacitor (BFC).

The downside to a larger BFC is that it will draw larger, shorter current pulses from the input transformer and rectifier.

This can cause a number of problems, though most are small, or can be mitigated.

a) Higher electromagnetic interference generation, due to larger current pulses, and higher currents being switched off in the diodes.

b) Slightly hotter diodes and transformer, due to larger RMS current.

c) Poorer input power factor.

A sniff of inductance somewhere in the supply (AC input, transformer leakage inductance, post transformer or post diode) will reduce the magnitude and extend the length of the rectifier pulses, improving all of the above.

Neil_UK
  • 158,152
  • 3
  • 173
  • 387
  • 1
    In some (e.g. battery powered) applications the leakage current of large electrolytic capacitors might become problematic as well. If something goes wrong with the circuit, a larger capacitance can provide more energy for more destruction. – jms Mar 24 '16 at 08:42
  • 2
    "The downside to a larger BFC is that it will draw larger, shorter current pulses from the input transformer and rectifier." this only really applies to capacitors connected directly to a rectifier, not really to capacitors on the output of a voltage regulator. – Peter Green Mar 24 '16 at 12:24
  • 2
    Some DC-DC converters have maximum capacitance ratings. If you are regulating the output of a DC-DC with an LDO then a large capacitor at either the input or the output (if the LDO current limit is too high) may cause problems. – Spehro Pefhany Mar 24 '16 at 17:58
7

Note: my interpretation of the OPs post is we are talking about capacitors on the output of voltage regulators, some other posts seem to assume the asker is talking about capacitors on rectifiers.

The main downside of a bigger capacitor is that the switch on rise time and switch off fall time will be greater. That means more stress on the regulator during startup and in extreme cases may even cause an overcurrent shutdown of the regulator. It can also cause problems for loads which don't handle undervoltage very well.

Having said that I don't think there is any point trying to micromanage the size of such capacitors. In most cases allowing a generous margin (a factor of 2 or more) over what you think you need is unlikely to be a problem.

Peter Green
  • 21,158
  • 1
  • 38
  • 76
  • 1
    I agree with your interpretation of the question. You might also consider how adding too large a capacitance may cause output voltage instabilities. – Andy aka Mar 24 '16 at 13:10
7

From Andy akas comment:

If the supply you are using has specific output capacitor requirements, then make sure you follow them. For all these types of regulator linked (LDO), there is usually a minimum capacitance only. (search the datasheet for ESR).

If you are using a switch-mode regulator, then the output capacitor (in current mode controllers) determines the output pole and zero. In voltage mode converters, it forms a resonant circuit with the output inductor. In both cases, we must provide loop compensation and that is partly determined by the value of the output capacitor(s).

(Note: I am aware that using ceramics on the output of a current mode device requires other techniques to provide an output zero as a ceramic capacitor zero is too high in frequency to be useful).

These capacitor(s) must be carefully chosen; changing these values requires re-assessing the loop compensation components, or it is quite possible loop instability can result.

This re-assessment may also reduce the loop bandwidth of the supply, reducing transient performance.

Peter Smith
  • 21,923
  • 1
  • 29
  • 64
6

Here's another point: many modern converters are protected against shorts or overloads in the output circuit. Such protection is a must for lab PSUs and a nice feature for all PSUs with connectors, since the ability to connect different loads increases the risk of shorts and overloads.

Having a big cap on the output reduces the effectiveness of such protection, since more energy is available to do the damage before the protection cuts the power off.

Dmitry Grigoryev
  • 25,576
  • 5
  • 45
  • 106
5

On the face of bigger is better for reasons that are well documented elsewhere.If the cap gets really big there will be problems with inrush current .On a small power supply the transformer should keep this down to a reasonable value .When rectifying mains into a cap filter the peak currents in the diodes can be several times the average DC output current .This is well documented elsewhere.This peakiness of diode current causes poor power factor and bad line current THD .If your source impedence is low the bigger cap will make this worse .Generally you can use the bigger cap on a small transformer based system without having to add any other parts.Larger systems can be made to work well by employing a line reactor on the AC or a small choke on the DC .If you are putting a very large smoothing cap on the output of a buck convertor there is a risk of instability which may need a small inductor to mitigate by divorcing the big cap.

Autistic
  • 14,235
  • 2
  • 27
  • 65
1

Larger capacitors also have more parasitics (eg equiv. series resistance and inductance.) This is what "slows them down" so to speak.

stanlackey
  • 11
  • 2