Polarized vs. Non-polarized caps

[swb1]

Hi,

When do I know when to use a polarized capacitor vs. a non-polarized capacitor?? Is there a rule of thumb??

Thanks,

swb1

 

[MikeHalloran]

Use non-polarized for AC or for fast-moving signals.

Explanation:
1. Reverse polarity usually blows up a polarized capacitor.
2. Polarized capacitors are usually electrolytic, i.e., part of the capacitance has to do with chemistry, which is slow, so they can introduce phase errors.



Mike Halloran
Pembroke Pines, FL, USA

 

[swb1]

Thanks for your reply Mike! So are you saying that the value has nothing to do with it?? Are you saying that if I have a slow moving signal (less than 30 MHz) and there is a 10 uF cap on the signal, I should always use a non-polarized cap?? At what speed is a signal considered "fast"??

Thanks!,
swb1

 

[Warpspeed]

Polarised capacitors use a very thin oxide layer as the insulating medium between the plates, so that the capacitance per unit volume can be made very high. They usually require a dc polarising voltage to stabilize the oxide layer, and being a chemical process, either very high or very low temperatures can be a problem. Another difficulty is that the actual value of capacitance may be neither very accurate or very stable.

The bottom line is, use polarised capacitors where you need a lot of capacitance in a small volume, but are not too fussy about the exact value. The main application is for coupling and decoupling, and bulk energy storage.

Non polarised capacitors are superior in every respect, except they start to become really huge and expensive if a really high value of capacitance is required.

 

[itsmoked]

Ceramic, mica, polypropylene capacitors are also considered non-polarized. It's just a class of capacitors. It's an important classification for people who lay out circuit boards. This is because of the need to put down silk screen foot prints that correctly depict the polarity for subsequent assembly.

 

[MikeHalloran]

To some people, 3kHz is a fast signal. I have a friend who's into music in a big way. He has replaced all the electrolytics in the signal path of his audio equipment with nonpolarized caps. I think I can hear the difference.

High capacitance per unit volume is really electrolytics' only virtue.



Mike Halloran
Pembroke Pines, FL, USA

 

[swb1]

Thanks for all the replies guys! My big newbie question is why is a capacitor needed anyway?? Sorry, for my inexperience :-(

swb1

 

[MikeHalloran]

A capacitor is used to:

- store charge

OR

- pass an AC signal while blocking a DC signal



Mike Halloran
Pembroke Pines, FL, USA

 

[itsmoked]

And while storing charge delaying signals. Or shorting high frequencies to kill them.

 

[Warpspeed]

And making the magic of tuned resonance possible.

 

[swb1]

"Storing Charge"?? For example, one side of a 4.7uF cap on a +5V line and the other side to ground. Why have the cap there and not let the +5V go straight into the IC??

Thanks,

swb1

 

[itsmoked]

A 4.7uF hooked up as you describe is for three reasons.

1) For logic hold up during shutdown (unlikely in your case)

2) To provide a low resistance,(impedance), supply for brief but large,(relative), current draws demanded by the logic that cannot be discovered and met by regulator quickly enough. (very likely in your case)

3) To provide the required capacitance needed to prevent the regulator from oscillating fatally for the circuit. (very likely in your case)

 

[swb1]

Okay..so how does one come up with the value 4.7 uF?? Will any value work??

swb1

 

[MikeHalloran]

Put another way, the +5V 'line' has finite impedance, as does the ground return to the power supply. The 4.7uF (other values would do) serves as a local supply.

A local supply is needed because logic chips draw large currents during transitions between states, and are sensitive to the supply voltage. Without these 'bypass caps' on each logic chip, the transient current spikes from one chip would pull down the line voltage (and/or pull up the ground voltage) and interfere with the operation of other chips.

If you build a circuit board without the bypass caps, it will probably mostly work, but it will exhibit strange and logically inexplicable behaviors that you will need a very fast oscilloscope to even capture.

If you look at precision analog circuits or op-amp circuits, you may find 4.7uF caps at the downstream end of low value resistors in the supply, effecting RC low-pass filters. Same idea, better filtering.

The 4.7uF is not entirely arbitrary. It's a large value that can be had in a tiny package. You can compute its cutoff frequency by estimating the resistance of the supply path and counting that as a resistor, again in an RC circuit.

For sensitive circuits where high frequencies are present, you may find several bypass caps of widely different values, parallelled right across the chip, to deal with transients of different speed. At GHz speeds, 4.7uF caps don't really behave like caps, so you need a few pF to bypass the really fast transients around them.

To not quite answer your last question, "what will work" depends on what you are trying to do, and in what environment you are trying to do it. But in general, if a circuit doesn't seem to be working like you think it should, and you didn't bother to bypass it, you should start debugging by doing it.



Mike Halloran
Pembroke Pines, FL, USA

 

[itsmoked]

Lets also keep clear that huge values like 4.7uF are generally for "energy storage" whereas bypass caps need to be smaller on the order of LESS THAN 0.47uF

The standard is pretty much 0.1uF

Larger caps have too much inductance to do effective bypassing.

 

[MikeHalloran]

We have sort of run off the end of what I've learned from a couple semesters of circuits, blowing up capacitors, and hanging around sparkys.

swb1, listen to itsmoked.



Mike Halloran
Pembroke Pines, FL, USA

 

[swb1]

Thanks for your time and patience guys!

swb1

 

[itsmoked]

MikeHalloran do okay man!

 

[Higgler]

Tantalums work if you install them backwards too. At least for a while. Long enough to get thru a circuit card tester (cuz they use low voltage to test at room temp). Then comes temperature testing where they fail. That was at the systems test level, lots of unwiring, unsoldering, and resoldering and rewiring and retesting later, it worked. Ahh, I remember all the failures so vividly.

Two tantalums in parallel is supposed to be a no-no (unless you put a tiny resistance between them) according to Mil Std's. I had one blow up in a production run, that's when we found out that tidbit of knowledge. Never did change that, never heard of anymore blowing up in the field.

kch

 

[felixc]

Adding a series resistor to Tantalum caps, what a strange idea. Aren't tantalums used for their low-esr characteristics? (also for their long-life, but...) I use many tantalums in parallel on my PC boards. Never saw any problem like that. I wonder what could have been the reason behind this Mil-std.