I don't know if this idea was mentioned before, but...

It seems that I have found a way to successfully wire electrolytics back-to-back for AC use without the problems that arise when they are used by themselves or with 'balancing' resistors.

My idea is to parallel with back-to-back wired diodes. The cathodes of the diodes (silver band) should face the positive side of the electrolytics, and the diodes should be able to handle at least as much voltage as the electrolytics.

What will happen is that each diode will short across the capacitor it is wired across when voltage polarity is opposite to what the capacitor is supposed to take. The diode will then directly feed the capacitor that is wired to handle this polarity. Of course there is that .7 volt drop, but in most cases I believe that this shouldn't be a problem, as .7 volts of reverse polarity across most electrolytics shouldn't harm them (perhaps I am wrong).

I have tested this arrangement on my ohmmeter and on the leakage test of my Solar capacitor analyzer (under maximum allowable voltage for the capacitors in question). The meter deflects and then falls to infinity. When the wires are reversed, this repeats. When they are reversed again this repeats again, and so on. Same holds true of the neon light on the leakage tester, which leads me to believe that this arrangement will behave as a regular non-polarized capacitor would behave on AC.

I don't know if this solves the problem of using non-polarized electrolytics as filament dropping resistors, as the overheating experienced here is more likely due to the large amount of current that is rushing into and out of the capacitor. ..It should make for a nice non-polarized electrolytic where applicable, though. I think that if electrolytics are used for filament voltage dropping purposes a fuse of adequately small size should be installed so-as to protect the filament string for the unstable characteristics of electrolytics (possible change in value).

T.

Thomas, you are broaching a subject of interest to me.

Many claim that you just wire two e-caps, back to back, and they will be fine in an a.c. ckt. I had trouble understanding this - since during each half cycle, one of the caps would be applied with a backwards voltage, which I had always understood to be bad.

So, it was explained to me that the cap with the reverse voltage would essentially short circuit (much like the diodes you propose would), leaving the other cap doing its job. OK, fine and dandy, but an e-cap hooked up backwards will overheat - and even for a half-cycle, I'd be a bit worried.

The whole idea of balancing resistors has been debated extensively. Alan Douglas claims they are totally unnecessary, but there are contrarians.

What we need here, in my opinion, is some serious testing using a 'scope, etc. With your proposed diodes and their 0.7-V drop, do the caps conduct any significant current during a backwards half cycle? And with two e-caps back to back, with no balancing resistors or diodes, how do the voltages divide across the two caps, considering cap tolerances?

It's surprising to me that there are so many strong opinions about all this, and seemingly no test data.

Well, so far as I can tell, two 47MFD caps wired in this way work fine with AC, and measure about 23MFD as a whole. I'm testing them at about 40 volts on a 25L6 tube to see if they get warm. So far they don't. A better test would be electrolytics suited for dropping for a 4 or 5 tube radio with the typical 2 25 volt tubes and 2 6 volt tubes. I'll let you know when I get that far.

T.

Here is some interesting information about capacitors used as a dropper for heather voltage. Does explain some about the 90 degree phase shift. About half way down the page.

http://www.vintage-radio.com/repair-restore-information/valve_dropper-calcs.html

Also an Excel work sheet that does all the math for you.

The primary goal of the article which I wrote has little to do with the various methods of reducing line voltage for use in powering heater circuits, but rather regards the ability to safely connect two electrolytics back-to-back to be used as a non-polarized capacitor in AC circuits of any kind requiring a non-polarized capacitor.

As far as I know, the diodes are superfluous; they'll always be biased off. The capacitors will charge to the peak of the applied AC, because of unequal forward and reverse currents in the individual capacitors (in other words, the "backward" capacitor functions somewhat as a diode). Once charged, the capacitors will never see any reverse voltage.

However I've never done the experiment. I'd be leery of using a nonpolar electrolytic as a filament dropper because of the high current involved. If that were possible, someone would be using them as motor-run capacitors to save a nickel, but no one is.

Alan, wait a minute - we're talking about two e-caps, back to back, right? If, during one-half cycle, the "backward" cap acts like a forward-biased diode, then it is essentially shorted, correct? And, I can testify from experience that an e-cap that is hooked up with the wrong polarity will cook.

If, on the other hand, the "backward" cap acts like a reverse-biased diode, it will pass no current, and current through the other cap is totally blocked.

I suspect you may be correct, but help mine unbelief.

:The capacitors will charge to the peak of the applied AC, because of unequal forward and reverse currents in the individual capacitors (in other words, the "backward" capacitor functions somewhat as a diode). Once charged, the capacitors will never see any reverse voltage.:

From what I see at first the capacitors are empty. Current flows one way backward through one, charging the other which is biased correctly for this polarity. Current then reverses, flowing backward into the charged electrolytic, while what was in this electrolytic is now charged into the other electrolytic, which was originally reverse biased and now forward based. This process repeats.

I'm not sure if the capacitor already being charged helps things or not. Current is still flowing backward through the capacitors, but perhaps since current is also flowing out of the capacitor instead of the capacitor simply being charged from an empty state in the wrong direction, damage doesn't occur?

I still want to try my idea, and when I have electrolytics of suitable size for filament dropping in a small mantle radio of mine, I'll let you know how they fair.

Also, I have seen electrolytics used with motors, but I believe that most were just as starting capacitors, with a cut-out switch. I have not studied this in detail, so I do not know.

T.

That's essentially it, it takes a few cycles to charge up the caps, but if the current is higher in one direction than the other, there is a net positive flow.

Nonpolar electrolytics have been used as motor-starting capacitors since at least the 1940s, but they're only in circuit for a few seconds and they have definite ratings for cooling off between restarts.

The problem with nonpolar electrolytics is they can't be reformed. Forming one foil "unforms" the other (they're made with two anode foils). So they have a limited lifetime.

Non-polarized electrolytics are simply two back to back equal value caps in one package. The Cornell Dublier site has lots of information as to how caps are made and their characteristics.

:Thomas, you are broaching a subject of interest to me.
:
:Many claim that you just wire two e-caps, back to back, and they will be fine in an a.c. ckt. I had trouble understanding this - since during each half cycle, one of the caps would be applied with a backwards voltage, which I had always understood to be bad.
:
:So, it was explained to me that the cap with the reverse voltage would essentially short circuit (much like the diodes you propose would), leaving the other cap doing its job. OK, fine and dandy, but an e-cap hooked up backwards will overheat - and even for a half-cycle, I'd be a bit worried.
:
:The whole idea of balancing resistors has been debated extensively. Alan Douglas claims they are totally unnecessary, but there are contrarians.
:
:What we need here, in my opinion, is some serious testing using a 'scope, etc. With your proposed diodes and their 0.7-V drop, do the caps conduct any significant current during a backwards half cycle? And with two e-caps back to back, with no balancing resistors or diodes, how do the voltages divide across the two caps, considering cap tolerances?
:
:It's surprising to me that there are so many strong opinions about all this, and seemingly no test data.
:
:
:

This is very true. However, there has been some concern in the past regarding unequal voltage distribution and possible damage to electrolytics due to reverse polarity on one of the electrolytics at any one time. Particular trouble was noted when non-polarized electrolyitcs were used as heater voltage dropping devices.

The purpose of my idea here is to protect each electrolytic when current is flowing in the direction opposite to which it is designed to conduct safely, and possibly make the non-polarized electrolytic practical for heater current dropping purposes.

I have yet to try it over a long period of time, but mentioned it here for those interested to observe. The heating problem could also simply be due to the electrolytics' inability to handle large inward and outward rushes of current, in which case this idea would not solve the problem of electrolytic heating.

T.

Just a thought about this. Would think the construction of an AC capacitor would be more durable and made to tight specifications. With a regular DC electrolytic the foil and dielectric don’t have to be made so critical in construction. Slight high and low spots though-out the foil and dielectric. When AC is applied to a DC cap this may cause hot spots in areas that have slight irregularities. This uneven charge is where a brake down could start.

The purpose of the diodes would be to assure that only voltage of the correct polarity is applied to each capacitor. Thus AC would not be applied to each capacitor.

T.

Thomas... It seems that the two caps will act as though the diodes were there inherently anyway:

From the Cornell Dublier Application Guide:

"NON-POLAR AND MOTOR START CAPACITORS
If two, same-value, aluminum electrolytic capacitors are connected in series with the positive terminals or the negative terminals connected together, the resulting single capacitor is a non-polar capacitor equal in capacitance to half of the rated capacitance of either of the original pair.

The two capacitors rectify the applied voltage and act as if they had been partially bypassed by diodes. However, at all levels of AC voltage the
apparent capacitance is half of the rated capacitance as you would expect for capacitors connected in series.

The nonlinear performance produces distortion where non-polar aluminum electrolytic capacitors are used in audio AC applications.

In non-polar aluminum electrolytic capacitors and motor-start aluminum electrolytic capacitors a second anode foil substitutes for the cathode foil to achieve a non-polar capacitor in a single case. These are available for momentary-duty AC applications like motor starting and voltage-reversing applications, but the
high DF of aluminum electrolytic capacitors – from 2% to 150% – causes excess heating and short life in most AC applications.

Aluminum electrolytic, motor-start capacitors are non-polar and designed for intermittent operation in starting single-phase induction motors or for other brief AC applications such as motor-run capacitors in electric door openers"

...They do act as a 'partial' diode, which is what is my concern. Essentially the capacitor is momentarily shorting due to reverse polarity. The diodes would do a much better job of shorting for the capacitors, and would save them the stress.

I have yet to try it out due to lacking proper values for use in a small mantle radio of mine. I've only tried this with 47MFD units at hand wired for 23MFD in series with a 25 volt tube at about 40-45 volts. ...Didn't get hot, but not as much stress, either.

I am curious as to what they mean by 'DF' when discussing electrolytics. I know what power factor is, but I don't recall what DF stands for. Dielectric factor? Discharge factor?

...But in the end electrolytics very well might not be suited for long-term high power AC use, but I want to try this idea out. ...Seems I need to place an order with AES or Radio Daze. I need to do that anyway.

T.

T,
Dissipation Factor, so may be related to heat from internal losses. I'm sure there are better explanations though. Google it!!

marv

:...They do act as a 'partial' diode, which is what is my concern. Essentially the capacitor is momentarily shorting due to reverse polarity. The diodes would do a much better job of shorting for the capacitors, and would save them the stress.
:
:I have yet to try it out due to lacking proper values for use in a small mantle radio of mine. I've only tried this with 47MFD units at hand wired for 23MFD in series with a 25 volt tube at about 40-45 volts. ...Didn't get hot, but not as much stress, either.
:
:I am curious as to what they mean by 'DF' when discussing electrolytics. I know what power factor is, but I don't recall what DF stands for. Dielectric factor? Discharge factor?
:
:...But in the end electrolytics very well might not be suited for long-term high power AC use, but I want to try this idea out. ...Seems I need to place an order with AES or Radio Daze. I need to do that anyway.
:
:T.
:

Ya. I tried Googling it, but didn't come up with much. However, I did read about ESR being a limiting factor for electrolytics used in DC applications where ripple is high and current draw is heavy, which would definitely be a problem if used as a dropping device for filament circuits.

Seems I remember someone on here trying to reduce voltage using a diode, but they found that the diode dropped too much, and so I suggested paralleling the filament string with an appropriately sized electrolytic to try to bring the apparent voltage back up. The electrolytic would only see DC, but of a high ripple content, and, as I remember, it was said that the electrolytic got very hot.

Grrr. I'm still going to try this idea, but I see the dead end that it is heading down. Basically you'd have two electrolytics operating on DC with a high ripple content, and they'd probably overheat.

T.

--high ripple content with heavy current draw, so the electrolytics would be charging and discharging quite a bit.

Dissipation factor is the inverse of Q and typically measured at 1 kHz. Power factor is the same thing but measured at 60Hz and expressed as a percentage; not much used nowadays. ESR is measured at 100 kHz or thereabouts, expressed in ohms directly. They're all ways of looking at the same thing.

I have done this for speaker crossover networks (and you see the commercial non-polarized electrolytics used there all the time) but never in a power application. At least for a speaker crossover network this works fine.

:I have done this for speaker crossover networks (and you see the commercial non-polarized electrolytics used there all the time) but never in a power application. At least for a speaker crossover network this works fine.
:

Hi Al;
Although you feel it "works fine" in crossovers..as you can read here however, according to Cornell Dublier ....the distortion created makes it non-desirable in audio application:
"The nonlinear performance produces distortion where non-polar aluminum electrolytic capacitors are used in audio AC applications. "