Can a non polarised capacitor replace an electrolytic?

[DMcMahon]

I think I know the answer but interested to hear other peoples thoughts.

Can a non polarised capacitor be used to replace an electrolytic when they have the same capacitance value and voltage rating, for example a 1uF 350V film capacitor to replace a 1uF 350V electrolytic when used as a de-coupling capacitor.

David

[lesmw0sec]

Yes.

[paulsherwin]

Yes, but they usually cost more for no benefit.

[GrimJosef]

Film capacitors can be longer-lived than electrolytics. The little ones (electrolytics) in particular can have a tendency to dry out.

Cheers,

GJ

If there is space and funding go for a film type, these don't wear or dry out. In certain designs the ESR (series resistance) of an electrolytic is used in the design but only in new highly cheapend jobs.

[paulsherwin]

A plastic film cap will normally be much larger than a modern polarised electrolytic of the same value.

[DMcMahon]

Thank you for all the inputs.

David

[Radio Wrangler]

Sorry for raining on the parade, but there is one little whoopsie to beware of. The electrolyte in an electrolytic capacitor is somewhat resistive and this gives the part a significant ESR component. In lots of power supply decoupling circuits, the inductance of wires/PCB tracks can resonate with the capacitance of capacitors. Having a few electrolytics around the place can damp the inevitable resonances. Replacing the electrolytics with film capacitors or modern high value ceramics can bring out all the resonances previously damped.

So there isn't a generic yes or no answer which just works. You have to have your wits about you. ESR is seen as an unwanted stray effect, but an awful lot of circuitry relies upon it in this case. Try things by all means, but be suspicious enough to check things are really OK.

David (A somewhat bitten old engineer)

[G.Castle]

I suppose series resistance could be added in that case if the job was being done because a non electrolytic was specifically required for some reason?

[kalee20]

Yes, it could - Merlinmaxwell alluded to this in post #5.

It's only in a few circuits that actually HAVING some ESR is desirable. As a component characteristic, it is rather poorly defined (manufacturers don't put a tolerance on it, for instance) and moreover is highly temperature dependent.

[G.Castle]

Thanks for that. I will have to read the prior posts in the threads more carefully

[kalee20]

Originally Posted by G.Castle

I will have to read the prior posts in the threads more carefully

I wasn't meaning that! I just like to acknowledge other posts who have already alluded to the same point as I'm going on about.

In fact David Radio Wrangler also posted about ESR, which although being an imperfection, is sometimes doing an important job. It's all circuit-dependent.

Me, I'd generally go for non-electrolytic where possible, but sometimes space does not allow. Life of modern plastic or ceramic capacitors is virtually indefinite, and I like my creations to be built to last. That said, in old equipment, it's quite usual for paper capacitors to need replacement while the can-type electrolytics are still doing their job...

[Radio Wrangler]

At various times in the past, different capacitor technologies have been seen as the only good choice.

AR88s were built without a single electrolytic capacitor... because they were considered to have limited lifetimes. So the sets were filled with paper capacitors which we are now beavering away changing.

Then people from Ericsson (the telecomms firm) were round touting their self-healing paper capacitors for mains filtering use... under their brand name 'Rifa' - enough said!

We've got the advantage of 20:20 hindsight, and we now know how a lot of these things turned out, simply because enough time has gone past. Some makes of film capacitors turned out to have internal corrosion issues, some were just fine.

Unfortunately, now we know which ones were really really good, they aren't making them any more. As knowledge arrived, availability went.

If you're putting a new type of capacitor in, it's helpful if you know the foibles of both the old type and the new type.

Relying on unspecified parameters like ESR, which are really imperfections, looks like crappy design, but it's widespread. Some people did it knowing what they were doing, others used electrolytics in certain positions because they were cheaper and they were unaware of the damping effect, others used electrolytics simply because everyone else did and were also unaware.

The way things were and are decoupled, with decent high-frequency capable ceramc capacitors for each IC, leaves you with a mesh circuit at RF, having multiple resonance frequencies and multiple modes. Undamped, the power supply quality seen by each IC is not necessarily good. It isn't decoupled, it's resonated.... Ugh!

So over the past decade, there has been a trend to avoid electrolytics and replace them with ultra high 'k' ceramic parts. Their ESRs are remarkably lower than the electrolytics they replace, and so the damping of all the resonances we're not aware of is less. So you can find ringing on IC power supply pins that wasn't there before.

One nasty critical application that only the bitten are fully aware of, is the decoupling of linear 3-terminal regulators. Cheap as muck semiconductors and equally as common. They are feedback devices. You can analyse them as amplifiers. Driving a capacitive load with a feedback amplifier is a known problem, it can add an extra pole within the feedback loop, using the effective source resistance of the output stage. Oops. Stability is impaired.

So, many of the well known 3-terminal regs are designed for electrolytic capacitors, with their ESR to go before and after them. The chip was designed around the ESR. Ooops change them at your peril. You may indeed want to fit a spoiling resistor to keep them tamed. Some advantages exist for using two capacitors, one with a spoiler, one without. This needs a bit of fancy design because you are using it to tune the damping to the part of the spectrum where the instability is.

I designed a modulator for an AM transmitter. The PA stage ran into low effective load impedance, so it was a high current, low voltage thing. THe PA really wanted good low-Z decoupling i.e. lots of C. But the modulator was effectively a 1-quadrant audio output stage with compound (multiloop) feedback. Stability was an issue, and the impedance presented to it by an upstream switcher was critical. So the circuit in a couple of places had a decoupling capacitor bridged by a second identical one with a series resistor. The boss kept questioning this, but he didn't understand the issues which were being balanced to get a compromise calculated to handle a range of part tolerances.

Often, decoupling gets done with no thought... done the way everyone seems to do it. Most often, it works. But few people are looking out for the pitfalls. The trend to high value ceramics has created a range of rather commodious pitfalls. They have rather large tolerances, but that's no problem, so did the electrolytics. They have lower ESR/higher Q as above. Their capacitance is rather voltage dependent! sufficiently so to need considering at your working voltage, and they can cause distortion. To top it all off, many of the little devils are strongly microphonic... the ceramic materials are piezo-electric.

Used with open eyes, they are very useful parts and cal allow us to dodge tantalum based stink bombs.

And everyone always thought capacitors were easier than inductors....

David