I just saw the first circuit in a drawing by Bill Marsden and wonder...why have I been using diodes all these years? My mistake or his?

 

Howdy!
The two diodes will certainly equalize the voltage across the caps, but they can create havoc where linearity is important...such as audio coupling circuits! It depends on the application.

 

It seems to me that the voltage will be divided equally between the 2 capacitors unless the dielectric fails. Have I got this belief wrong?

 

It seems to me that the voltage will be divided equally between the 2 capacitors unless the dielectric fails. Have I got this belief wrong?

That is true if the values of the two caps are identical. However, as you know the tolerances of most electrolytics are notoriously bad! So commercially made NP capacitors are SUPPOSED to be better matched than "do it yourself ones." though I'm not sure how true that is.

 

I'm not so much concerned with the equality of the forces as much as the idea that one capacitor would seem to be stressed in a direction opposite to its labeled polarity (during each half cycle) without the diodes. Do aluminum electrolytics just give up and short out the instant the charge gets backwards? That would prevent a backwards charge accumulating on either capacitor.

 

Something I did not know. From Wikipedia:
Two identical polarized electrolytic capacitors are connected back to back to form a bipolar capacitor with half the nominal capacitance of either.[16] However, the anode film can only withstand a small reverse voltage.[17] This arrangement can lead to premature failure as the anode film is broken down during the reverse-conduction phase and partially rebuilt during the forward phase.[18] A factory-made non-polarized electrolytic capacitor has both plates anodized so that it can withstand rated voltage in both directions; such capacitors also have about half the capacitance per unit volume of polarized capacitors.

 

There's a recent thread on the subject:

http://forum.allaboutcircuits.com/threads/polarized-to-non-polarized-capacitor.91406/

 

Look at these NP electrolytic capacitor in two realms, DC and AC.
In the DC world you are dealing with the leakage current of each capacitor. It is the leakage current that keeps the DC voltage equal, if both capacitor sections have the same leakage current that is. If one section has been damaged, the leakage current is the first thing to be affected, so the DC bias will shift so the healthy capacitor will see most of the voltage. Putting diodes in parallel with the caps does not keep the voltage equal, it just keeps one capacitor from seeing over .7V of reverse voltage.
In the AC world the balance of the capacitance determines the balance of the voltage between each capacitor. A capacitor with high leakage current will not change capacitance much.

........ Do aluminum electrolytics just give up and short out the instant the charge gets backwards? ......

When an electrolytic capacitor has reverse voltage applied to it the leakage current begins to increase. The higher the reverse voltage the quicker the leakage current increases. If not too much of this occurs then the capacitor generally will recover. But if it continues, heat will physically damage the electrolyte thus destroying the capacitor.

 

@ The Electrician:

That was a really good thread, and a big Thanks to RB for doing the work to find out what happens at different voltages!
Of course, I have no idea how I missed it...it was only a year ago. Probably one of my absences that last for a week or so when my real life demands my attention.

All in all, I think the diode method is better for the longevity of the capacitors and will continue to use that.

 

#12, consider the following:
Two electrolytic capacitors go into a bar...... no. no just kidding.

Two electrolytic capacitors are connected plus to plus as NP pair. One capacitor is healthy and the leakage current is modeled at 90 meg ohms, the other cap has had a rough life and leakage is modeled at 10 meg ohms. 10 volts DC is applied to a caps. This puts the voltage at 9 volts across the healthy cap and 1 volt across the lesser cap. At this point, do the diodes do anything to help this condition?......No, diodes are reversed biased and have no effect. Now we apply a 4 volt peak to peak sine wave across the caps. Because their capacitance is the same, the voltage is the same across each cap. The lesser cap sees 2 volts, which puts it 1 volt into the reversed voltage region except that the diode conducts at .7 volts. The result of using diodes is that one capacitor still sees a reversed voltage and the signal is distorted.

Now lets do it right:
Exact same circuit except that the diodes are replaced by 1 meg resistors. Because the resistors pass 10 time more leakage current than the worst capacitor, they hold the voltage equal between the capacitors. Now the 2 volt peak to peak that the leaky cap sees is no problem. No reverse voltage, no distortion. Yes, you have more leakage, but if leakage is important, you should be using a ceramic capacitor or something else.

 

Or I can just buy the right N.P. cap for the finished product.

 

That's what I do.