Difference between Bipolar and Non-polar caps

The first are polarized, the second are not.

It has everything to do with getting sufficient capacitance into a small enough package. For larger values, electrolytics are really the only means of doing that. But electrolytics by the construction are polarized.

For most places electrolytics are used (because of a need for higher capacitance), polarized capacitors are not a problem. Either they are filtering power supply lines where they are seeing pretty much a single polarity signal, or they are carrying AC where one terminal is clearly more positive than the other.

Rare is the case where high values of capacitance are needed and they must be non-polarized. But for those few occasions they can manufacture electrolytics so they are not polarized, and there you go. In effect non-polarized electrolytics are the equivalent of two electrolytics in series inside the package:

---||-||--- + - + which of course can be emulated by putting two discrete electrolytics in series with the proper polarity (and remembering that it will be half the capacitance).

Note that most capacitors are "non-polarized". But since they don't come in polarized versions, since the construction doesn't make them polarized, they don't need to be referred to as "non-polarized". We are talking about ceramic, paper, polystyrene and any other type of capacitor except for electrolytics and tantalums. YOu will also find that it's rare to find these types of capacitors in anything much larger than 1uF or so.

Michael

Well, there are times one will use large capacitance that is non-polarised. Note it's not because it's non-polarised, but because of undesirable effects of electrolytics (and tantalums).

Large capacitance at a high voltage is the realm of electrolytics, of course, and anything over about 100uF is too, but expect the high values of ceramics to increase.

The issues with electrolytics (and to a certain extent tantalums, which have other undesirable traits) is the high ESR (Effective series resistance), their poor stability vs time and their horrendous stability vs. temperature.

If you are filtering a large chip (let's say a power hungry processor), it's distinctly possible there's up to 100A going on, but at 0.9 -

1.2V. As ripple performance is dominated by ESR, you need ultra low ESR as well as some bulk. There are 100uF ceramics available at 6.3V that are perfect for this sort of thing (and as the bulk output of power supplies, replacing 1000s of uF of electrolytics) and they are cheaper too. Of course, you could use bulk electrolytics with lots of small ceramics, but you still don't get all the benefits of that ultra low ESR as soon as you put an electrolytic in (it will still source/sink the bulk currents).

As noted, switch mode power supplies are another example of where I need low ESR (although it significantly complicates the loop stability, but that's another issue) to maintain the output stable in the presence of impulses, be they on input or output.

A bonus is the size - 100uF ceramics come in a 2220 case (0.22 x 0.2 inch) and very low height, and the fact they are MUCH cheaper than electrolytics.

(One of the issues with tatalums is the chemistry involved - if they overheat, they give off toxic fumes. One large company we OEMd equipment for insisted we could never use electrolytics in the equipment destined for them).

The following manufacturers make large ceramic values in MLCC (MultiLayer Ceramic Capacitors) and there may be others:

Panasonic TDK Kemet

Cheers

PeteS