Important capacitor specs ?

[solo2racr]

When building a linear PSU with LM317/LM337 regulators from a schematic, and the only specs shown are for capacitance and voltage, what other properties are important, or particularly matter? Things like temp, ESR, ripple, etc. I would assume higher temp, lower ESR and lower ripple would be a better choice than the opposites of that. And which specs are more important, if any, than another?

 

[Gryd3]

When building a linear PSU with LM317/LM337 regulators from a schematic, and the only specs shown are for capacitance and voltage, what other properties are important, or particularly matter? Things like temp, ESR, ripple, etc. I would assume higher temp, lower ESR and lower ripple would be a better choice than the opposites of that. And which specs are more important, if any, than another?

What do you plan to use it for? Any spec can be important if your circuit has specific requirements. Otherwise if your just lighting LEDs you can get away with almost anything.

 

[(*steve*)]

Ripple current defines how fast you can charge and discharge the capacitor without damaging it. A higher ripple current is better (all other things being equal).

The ripple current experienced by filter capacitors can be surprisingly high (the peak current can easily be 4 or 5 times your load current).

 

[KrisBlueNZ]

Generally if nothing else is specified, you can use pretty much anything that meets the specifications that are given.

For decoupling, for values of 1 µF and less, ceramic capacitors are traditionally used, because they have low ESR and are cheap and compact. But avoid Zxx dielectrics (e.g. Z5U) because it has a limited temperature range and the capacitance isn't very stable. I normally use X7R.

For capacitors of 10 µF and higher, aluminium electrolytics are normally used, and if no ESR or ripple current specifications are given, the requirements are not strict. ESR and ripple current are relevant for output capacitors in switching supplies, but not normally for anything else. Ripple current specifications apply at 100 kHz, not at 100 Hz or 120 Hz.

Edit: that last sentence is wrong. Thanks Steve for the correction (see next post).

PNP-based LDO regulators have minimum ESR requirements for the output capacitor - i.e. your output capacitor can't be "too good" - for stability, but the LM317 doesn't care.

 

[(*steve*)]

My main point (not clearly enunciated) was that a higher ripple current rating is better (not worse) when all other things are equal.

Before you read any further, the rest of this post (almost 100% guaranteed) does NOT apply in your situation. Follow Kris's advice.

Ripple current applies at all frequencies. See here where it s discussed for tantalum capacitors at frequencies from 100Hz to 100kHz.

The important thing to consider is that the ripple current rating is dependant on frequency (and not in a simple way either). It is an I²R (i.e. heating) issue, and the frequency dependence comes from the fact that the R is the ESR of the capacitor and that is frequency dependant (in a likewise non-simple way).

Ad Kris says, it's generally not an issue at low frequencies, but this is because the ESR is much lower (leading to a higher permissible ripple current), not because ripple current per se is not an issue.

This means you have to read the specs very carefully *if* it is an issue.

(I think I need to do a resource on reading specs. There's so much that is well hidden in even the most clearly stated specifications if you don't know *exactly* what they mean or under what conditions they apply.)

Here is another article showing how different assumptions can give very different ripple current ratings for the same capacitor.

Even determining the ripple current can be tricky. Here and here are examples of manufacturers trying to make the selection of the correct capacitor an easy process D). Note that ripple current is just one aspect.

You might notice that I find this stuff fascinating.

 

[solo2racr]

Thanks guys. The use would be to power audio circuits. So, as you can imagine, the cleaner the better. Of course, there is a point of diminished returns. So, given the option, HIGHER ripple, lower ESR and higher temp would be the preferred choices.

 

[KrisBlueNZ]

Thanks guys. The use would be to power audio circuits. So, as you can imagine, the cleaner the better. Of course, there is a point of diminished returns. So, given the option, HIGHER ripple, lower ESR and higher temp would be the preferred choices.

Yes, but the ripple current specification is irrelevant for caps on the output of a linear regulator.

Thanks for the info Steve. I stand corrected!

 

[solo2racr]

Will keep that in mind, Kris. I'm working on a microphone preamp based around a THAT1510. It will also be my first time trying to etch a PCB as well. When I was looking at cap dimensions and pin spacing, it occurred to me that I didn't really know what difference the other specs meant for a given application. Which is what prompted this question.

 

[Harald Kapp]

While at first glance it seems reasonable to use capacitors with as low an ESR as possible, This is not true for many linear regulators. Many linear regulators have been designed with known limitations of the available and affordable capacitors in mind. This leads to the seemingly paradox effect that the output capacitors for many linear regulators are expected to have and accordingly should have some minimum ESR to achieve stability of the regulation loop.

As Steve wrote, nothing beats reading the datasheet (and knowing how to interpret the data you find there).

More can be read in this application note.

 

[KrisBlueNZ]

While at first glance it seems reasonable to use capacitors with as low an ESR as possible, This is not true for many linear regulators. Many linear regulators have been designed with known limitations of the available and affordable capacitors in mind. This leads to the seemingly paradox effect that the output capacitors for many linear regulators are expected to have and accordingly should have some minimum ESR to achieve stability of the regulation loop.

PNP-based LDO regulators have minimum ESR requirements for the output capacitor - i.e. your output capacitor can't be "too good" - for stability, but the LM317 doesn't care.

 

[solo2racr]

As Steve wrote, nothing beats reading the datasheet (and knowing how to interpret the data you find there).

As well as knowing how to apply that data to the circuit you're working on. That's why I asked.

 

[(*steve*)]

As well as knowing how to apply that data to the circuit you're working on. That's why I asked.

Yeah, that's a great point.

We tell people to look at datasheets, but it's important that people know what to look for and when to apply it, and (even more importantly) *how* to apply it.

If you take a look at my developing heatsink resource you'll find that some of the required parameters can be called different things, can be derived several ways, and sometimes are not even mentioned on the datasheet. It can be a case of "Oh, you looked at *that* datasheet -- no, you want the other datasheet for that component" -- AAAGH!)

Many of us have been reading datasheets for so long that we almost forget that it may as well be written in foreign language for many people (or an even more foreign language for those who don't have English as a first language).

 

[solo2racr]

Yea. I'll look at a datasheet and sometimes wonder what does that unit of measure mean.

 

[solo2racr]

For decoupling, for values of 1 µF and less, ceramic capacitors are traditionally used, because they have low ESR and are cheap and compact. But avoid Zxx dielectrics (e.g. Z5U) because it has a limited temperature range and the capacitance isn't very stable. I normally use X7R

Kris...... Instead of ceramic, how would film caps work? On the particular PSU I'm working on, The first cap after the bridge is a 470uF polarized cap followed by a 100nF non-polar cap. For some reason, I have a fondness for Wima film caps. I know...weird. LOL.

 

[KrisBlueNZ]

A 100 nF film capacitor will be fine. Film capacitors are not very space-efficient but they perform well.

 

[solo2racr]

Thanks again, Kris. I am using a mixture of film and X7R ceramics to keep the size and specs in check, depending on voltage requirements.