Polarized capacitors for coupling

[NIC1138]

I often see electrolytic capacitors being used for coupling in situations were the sinal can reverse the polarity on it, and even a reverse-polarity DC level might be present. People say it doesn't matter, it won't blow the thing, because there are no big currents going on, or whatever excuses they have...

What is your experience? Has anybody ever had trouble for using electrolytics for coupling in the input of the circuit? Am I the only person afraid of doing it?...

[sherelec]

NIC1138,

Your question is a very good one. In my experiences in power supply design I have seen configurations that foldover electrolytic capacitors during operation. Usually this happens under startup or fault conditions. The capacitors seem none the worse for wear.

In particular a capacitor input voltage doubler will experience a negative bias startup cycle 50% of the time. This same circuit will experience almost 100% foldover every cycle under short circuit conditions. My experience is that no premature failures are encountered in these circuits during these operating conditions.

Note, these conditions subject the cap to reverse bias for short time periods or in rapidly alternating periods of positive and negative bias. I believe this is key. In general an electrolytic seems to behave well in reverse bias if it has just experienced a period of positive bias. Thus, I believe, allowing it to have a properly formed insulating dielectric. The short period of reverse bias does not deplete the "gas layer dielectric" and thus the cap appears to work for both positive and negative bias conditions.

In some design work, circa 1995, I encountered an app note from a major capacitor manufacture on using electrolytics under intermittent reverse bias conditions (unfortunately I did not copy and save it). The manufacture in question would have been Mallory, CDE, Nichicon or Panasonic. In general electrolytic manufacturer spec sheets say do not reverse bias their caps.

Finally, in one of Douglas Self's books he has a preamp where most of the coupling caps are connected up with a small negative bias most likely across them. Later on in the same book he has a DC blocking cap on a diff input with a couple of diodes protecting it from some undefined failure condition-go figure!

Thank you for starting this thread,
VSR

[gootee]

It seems to also be fairly common practice to place two polarized electrolytics in series, oriented "back to back", to attempt to create a non-polarized electrolytic, which seems to work. I have read that in that case it might be advisable to place a large resistance in parallel with each cap. If anyone has more knowledge of, or experience with, this type of configuration, please comment.

- Tom Gootee

http://www.fullnet.com/~tomg/index.html

-

[AndrewT]

Hi,
a few electrolytic manufacturers state that their caps behave as bipolar upto 1500mV or a little less.
This would allow 1Vac across the cap, and that just happens to be similar to the input of most pre-amps and amplifiers.

If the cap has a DC bias across it then orienting it appropriately extends the usuable voltage range even further.

If Sherelec's experience were supported by the manufacturers for short term AC conditions then there appears to be no problem.

Many amplifier designers/manufacturers use polarised this way and they offer guarantees with their products.

The main argument against electrolytic coupling caps is poor quality sound.

[Christer]

Originally posted by AndrewT
Hi,
a few electrolytic manufacturers state that their caps behave as bipolar upto 1500mV or a little less.
This would allow 1Vac across the cap, and that just happens to be similar to the input of most pre-amps and amplifiers.

I think that applies to all normal electrolytics. The voltage figure differ slightly between manufacturers, but my guess is that that has more to do with how cautios they are than in real differences, since they all state a general figure supposed to hold for all their products.

For normal operation we are probably not likely to see as much as 1 V over a coupling cap. However, it should also be able to cope with fault behaviours. Suppose the preceeding piece of equipment breaks and its output sticks to one of the rails, then we want our coupling cap to survive this and protect what comes after it. Also in the extreme case of fairly high voltage, very low frequency components in the signal, we might get higher reverse voltages over the cap. Maybe that should be considered a fault condition too, but then, the cap should cope with it.

[Keruskerfuerst]

This can be used, when a sufficient DC offset (=larger than the peak of the input signal) is added to the "input" of that capacitor.

Usually a bipolar capacitor can be used.

[NIC1138]

Originally posted by Keruskerfuerst
Usually a bipolar capacitor can be used.

By "usually" you mean how much input resistance? More than 500k, I suppose??...

I bought a few 470nF polimer caps here... These are the largest bipolar capacitors with small mechanical size I could find. I was thinking about making them the "standard" input cap in my projects... This gives about 16Hz of lower cutoff frequency for a bad input resistance of 20k... OK, perhaps, for a "standard" input?

What's the deal against electrolytic caps? Noise? Low parallell resistance? High series resistance? What are the actual figures?... What to do if I want, say, 200uF of decoupling?

[gootee]

Originally posted by NIC1138


By "usually" you mean how much input resistance? More than 500k, I suppose??...

I bought a few 470nF polimer caps here... These are the largest bipolar capacitors with small mechanical size I could find. I was thinking about making them the "standard" input cap in my projects... This gives about 16Hz of lower cutoff frequency for a bad input resistance of 20k... OK, perhaps, for a "standard" input?

What's the deal against electrolytic caps? Noise? Low parallell resistance? High series resistance? What are the actual figures?... What to do if I want, say, 200uF of decoupling?

How small are those 470nF caps?

The AVX 63V polyester "box" caps are fairly small. The 2.2uF ones have a footprint of about 0.237" x 0.295". Their 470nF one is about 0.2" x 0.295".

If you must have polypropylene (and sometimes you "must"), the WIMA (63V I think) polypropylene "box" caps are the smallest I have found. The 0.22uF one has a footprint that's about 0.275" x 0.275".

You can find the datasheets for those through http://www.mouser.com , for example.

Regarding electrolytic caps for signal coupling... Some people don't seem to like the idea. But I have used Nichicon UHE-series electrolytics (220 uF 10V for example) for DC-blocking caps for high-quality sine signals in instrumentation and haven't seen any bad effects, having measured THD (below .005%, agreeing with simulation results). Yet, many people claim that they are bad for the sound. And I've never tried listening to them. So maybe you'd just have to test some.

- Tom Gootee

http://www.fullnet.com/~tomg/index.html

-