What is ripple current?

[DMcMahon]

What exactly is the ripple current number (such as 300mA for example) shown on some electrolytic can capacitors, is it the AC current component (due to the mains ripple) of the DC HT voltage say from the reservoir capacitor (which to me seems high) or something else ?

David

[GrimJosef]

It is, as you say, the AC current flowing in the capacitor. In a power supply with a full-wave rectifier there will be a 100Hz component (120Hz in the USA) and the waveform won't be sinusoidal (short charging spikes heat the capacitor up more than a sinusoid would). Depending on the application there may also be higher-frequency components if the load has a time-varying demand.

Cheers,

GJ

[duncanlowe]

I agree it's the current flowing into and out of the capacitor, due to the rectified AC supply putting current in during peaks, and the load taking current out during troughs.

But why is it a number on a capacitor?

Depending on the current taken out by the load, to be replaced by the AC, the amount of current flowing in and out, the ripple current will vary.

Why does that matter? Because capacitors aren't perfect and the current flowing backwards and forwards, in and out, will cause some heating. So the maximum ripple current is specified so that you don't overheat the capacitor.

Or at least that's what I was taught, albeit a very long time ago.

[Radio Wrangler]

Ripple current markings are maximum ratings. They are a value you shouldn't exceed. You don't have to run them at this current, but if you do go over, self-heating due to the equivalent series resistance of the capacitor will accelerate the rate of loss of moisture from the electrolyte, shortening the life of the capacitor. Sealing is never perfect and water vapour can make its way out.

For good life, you may want to run your capacitors to about half the marked value. This will usually much more than double their life.

David

[ms660]

As an approximation:

For half wave Ir = Vr*2*pi*f*C

For full wave Ir = Vr*2*2*pi*f*C

Ir = Ripple current

Vr = Ripple voltage (RMS)

pi = 3.141

f = AC supply frequency

C = Capacitance in Farads

Lawrence.

[kalee20]

Originally Posted by DMcMahon

What exactly is the ripple current number (such as 300mA for example) ...

is it the AC current component (due to the mains ripple) ...

(which to me seems high)

Yes you're right, as others have put.

Does it seem high? Well, just do some basic calculations.

Suppose you stick your 'scope across a 50μF reservoir capacitor, and see 300V but with 40V peak-peak of ripple at 100Hz, which are all quite plausible figures.

Then, very crudely, 40V p-p is 20V peak, or 14V rms.

Now at 100Hz, reactance of 50μF is 32Ω. So if you have 14V of AC across this capacitor, you'll have 438mA of AC going through it (partially discharging and recharging as the voltage falls and rises). This is a very crude figure, but it gives you an idea. The 300V DC pedestal has absolutely no effect on this.

Why does it matter, a capacitor is supposed to be lossless, innit?

Well, internally, the capacitor is made of wires, foils etc and we all know that if you push enough current through a wire it'll get warm. So, to avoid excessive temperature rise, manufacturers put a limit on how much current their capacitors should pass. Less current is always good, and if you want your capacitors to last a long time you'll not put them next to a rectifier or output valve, but somewhere cool. The manufacturers also tend to supply derating curves too, if your capacitor is kept at a low temperature it's happier to accept more ripple current - after all, it is the maximum temperature (ambient temperature plus internally generated heat) which matters.

[Craig Sawyers]

There is a classic paper that covers all this, Otto Schade "Analysis of Rectifier Operation" July 1943. You can find this is downloadable if you google carefully.

There are other good references that cover that, including databooks, but looking through them, they all refer back to the design graphs of Schade.

A really useful piece of simple power supply design software is https://www.duncanamps.com/psud2/download.html

But, by and large the RMS capacitor ripple current is between 1.7 and 2 times the load current. So if your supply delivers a steady current of 1A, reckon on the ripple current rating of your smoothing capacitor to be at least 1.7A and preferably significantly higher.

If it is supplying AC current to a power amplifier, class B or AB, the calculation is very different. See for example http://www.signaltransfer.freeuk.com/powerout.htm

Good luck

Craig

[DMcMahon]

Originally Posted by Radio Wrangler

Ripple current markings are maximum ratings. They are a value you shouldn't exceed. You don't have to run them at this current, but if you do go over, self-heating due to the equivalent series resistance of the capacitor will accelerate the rate of loss of moisture from the electrolyte, shortening the life of the capacitor. Sealing is never perfect and water vapour can make its way out.

For good life, you may want to run your capacitors to about half the marked value. This will usually much more than double their life.

David

Thank you David.

When replacing say an old reservoir capacitor and say it is marked as 300mA ripple, ideally the new replacement should have similar ripple current spec.

Some modern (and some old) can capacitors have no ripple current value identified on the can. In that situation would you recommend to steer clear of them (as replacements) assuming that the supplier cannot provide a manufacturers spec.

David

[Craig Sawyers]

Originally Posted by Radio Wrangler

For good life, you may want to run your capacitors to about half the marked value. This will usually much more than double their life.

David

I suspect that capacitor life obeys the Arrhenius equation. So the hotter it runs, the lower the lifetime.

Aha - thought so http://www.rubycon.co.jp/en/products/alumi/pdf/life.pdf

Craig

ideally the new replacement should have similar ripple current spec.

At least as or greater unless you analyse the circuit and make an "executive" decision. Mind you capacitors (as are all components) are getting better all the time, it may be hard to find one with a ripple rating anywhere as low as the original.

[Ed_Dinning]

Hi Folks, modern (especially SMPS) caps also have a temp rating, say 85C or 105C. This temp is the max that the cap should be operated at to achieve its design life.
A reduction of 10C on cap temp will double the design life at the higher temp, so 105C caps are often used well below max temp when extended lifetimes are needed.

Actual cap temperature is directly related to ripple current, which, if being measured must be with a true RMS meter

Ed

[Radio Wrangler]

Some capacitors may even give you two different ripple ratings, one at 100/120Hz and the other at tens of kHz (typically) for SMPS frequency current ripples.

David

[trobbins]

Care is needed when looking for ripple current rating in some datasheets - especially if there is a 'ripple current multiplier' given for frequency. Some e-caps can have about half the ripple current rating at 100Hz as they have above 10kHz.

The larger the value of e-cap used for common applications the less likely that ripple current is going to be a concern, but do be careful if using low value ecaps in the first-filter position in power supply circuitry, especially if aiming to use the same low cap value like 4 or 8uF for vintage equipment.

The other aspect of concern is to distance an e-cap from heat sources, like a 5W resistor for audio output stage cathode bias, or on top of the chassis near rectifier and output stage valves (especially when the equipment has a cover and gets little air flow).

[Radio Wrangler]

In valved equipment, where a valve rectifier is used, many rectifier valves have a maximum capacitance value they are rated to drive. This is a maximum capacitance for the total oof capacitors connected to its cathode. Further capacitors isolated by resistors or smoothing chokes don't count.

You'd have thought the more capacitance the merrier? More certainly means lower ripple voltage on the HT supply, and less hum. But a bigger capacitor discharges less in voltage between the peaks of the mains cycle, and this means the rectifier has to rplace the amount of charge in a smaller time at the peak of the next cycle. So as the smoothing capacitor value is increased, the charging currents it demands from the rectifier get more and more like narrow pulses, with the current having to increase to make up for the shortened width. High pulse currents shorten the life of rectifier valves, so each rectifier valve type is specified to drive a maximum size of reservoir capacitor. Exceed this, and the coating on the cathode will fail.

This serves to establish a minimum amount of ripple voltage on the reservoir capacitor. If this is too much for the set to not hum a lot, then a second stage of HT smoothing is needed, with more capacitance, but fed from via a resistor or choke from the reservoir.

So, when replacing a smoothing capacitor, if you have to use a larger reservoir capacitor on availability grounds, check that you aren't exceeding the specs of a valve rectifier.

If the capacitor you're replacing is a can with both the reservoir and smoothing capacitor in it, the reservoir capacitor ( the first one fed from the rectifier) gets the harder life with more ripple voltage and more ripple current than the smoothing capacitor (the one fed via the resistor or choke) Consequently, dual capacitor cans are designed with the reservoir capacitor rolled-up around the smoothing capacitor, so that it isn nearer the surface of the can and benefits from better cooling. Getting them the right way round helps with the life expectancy of the capacitors. You'll see that one section of the dual capacitor has a better ripple current rating than the other, or is marked as the 'Outer'.
The markings may be written on the can, or you may have to look them up in the data sheet from the manufacturer.

So, not only do capacitors have ripple current specs, but rectifier valves do too. Helpfully, the valve manufacturers have done the calculations for you and state it as a maximum reservoir capacitor value. But you have to know there is a limitation, and to look it up.

David

[Craig Sawyers]

Semiconductor diodes also have a maximum repetitive current rating, IFRM. Depending on the diode that can be between 4 and 8 times the average current rating.

So, in the same way that you can get a smoothing capacitor wrong with a valved rectifier, you can do exactly the same thing with a semiconductor rectifier.

Even Tektronix got this wrong from time to time. The bridge rectifier in the 475 was know to fail for that reason (the 475 has a conventional non-switched mode supply). They replaced it with a higher rating and that failed too. It took them four attempts to get it right.

Craig

[DMcMahon]

Thank you to everybody for the very interesting and informative inputs.

David

[Ed_Dinning]

Note also the the same caps from the same manufacturer but different shapes can have quite different ripple current ratings. It is all down to the surface area.
Most are spec'd for still air but some do specify airflows

Ed

[turretslug]

I recall a Philips data sheet for their larger, screw-terminal PSU-orientated electrolytics stating that the ripple rating could be increased by 20% if the blue plastic sleeve was removed to leave bare aluminium. I generally trust Philips as knowing what they're on about, but it struck me that if you end up resorting to this sort of measure, then surely things are a bit marginal anyway!

[cmjones01]

Ripple current anecdote: we know how important it is for capacitors to be adequately rated for ripple current, and the data sheets for modern components give useful information to help the designer choose the right part. However, sometimes a "best guess" is necessary and things go wrong. Properly wrong.

A couple of years ago I designed a 3-phase motor controller for an electric bicycle (the application is relevant here, trust me). The main bulk capacitor in such a controller has a hard life, because the motor windings are inductive and so current is constantly flowing in to and out of them via the poor capacitor. I didn't have an easy way to work out how large the resulting ripple current was going to be, so just used a "best guess" component in the prototypes in order to make some measurements and optimise it later.

It turned out that this capacitor was working harder than I thought. A prototype was fitted to a bike in a convenient location more-or-less between the rider's legs. During testing on the road, a bug in the motor control software came to light, which scrambled the switching of the power to the motor windings so the ripple current became even higher than usual. At this point the bulk capacitor, which had already been running very hot, gave up the unequal struggle and exploded with a loud bang and cloud of white smoke! I wish I'd been there to see the test rider's reaction. He was apparently a bit shocked!

There followed a period of bench testing and measurement of the ripple current, and some changes to the way the bulk capacitance was provided...

Chris

[Diabolical Artificer]

Is ripple current relative to capacitance? EG a higher capacitance cap will have lower ripple but at the same time a higher ripple current rating? I must admit to winging a lot of times when buying caps,the datasheets often don't have a current ripple rating at 100hz only at higher frequencies.

Andy