Non polarised electrolytics

[ppppenguin]

I need to replace a number of these in my Thorn 2000. The usual approach is to put a pair ordinary electrolytics in inverse series. I've done this many time before and had no problem. But I've a nagging doubt.

Is the capacitance halved as in the normal series conenction of caps? Or not, as only one of them is really working at any one time? In previous applications I've assumed halving and to be honest it didn't matter if that wasn't the case. Here they are in convergence circuits where it might matter to some extent.

For voltage rating I'd always be conservative and rate each cap at the full voltage. After all, the cap that's reverse biased isn't really helping.

[Crackle]

Jeffrey
What sort of values you are after? Have you considered film capacitors, they can be quite small now.

Mike

[ppppenguin]

50uF would be rather too much for a film cap. Even if available it would be too expensive and too big.

100uF and 150uF even worse.

[peter scott]

I suspect that their value is not halved but you could always try measuring them. How about a diode in parallel with each half?

Peter

[ppppenguin]

Diodes like that would probably give some hideous non-linearity.

I suspect that in an DC application (where you wouldn't need a non-polarised part) one of the caps would tend towards a short circuit. So the value wouldn't halve. In an AC application each cap would stay formed despite spending half its time reverse biased so the value would be halved.

Or is that a load of rot?


From Wikipedia, sounds plausible:

Special bipolar aluminum electrolytic capacitors designed for bipolar operation are available, and usually referred to as "non-polarized" or "bipolar" types. In these, the capacitors have two anode foils with full-thickness oxide layers connected in reverse polarity. On the alternate halves of the AC cycles, one of the oxides on the foil acts as a blocking dielectric, preventing reverse current from damaging the electrolyte of the other one

[pwdrive]

Reactance measurement?

Lawrence.

[Kalee20]

Halved or not? Good question!

If you apply 1mV AC to your series pair, of say 100uF each, it is inconceivable that the oxide film be wrecked by such a small inverse voltage. So, you'll see the pair behave as a single 50uF component.

If you raise the AC to 20V peak, then by equal voltage division, you'll get 10V across each capacitor. One of these will be the wrong way round, and 10V reverse is enough to upset things. So this capacitor would easily look leaky, and start clamping its voltage, leaving the majority of the volts across the other. So, the pair will behave like something between 50uF and 100uF.

What were your concerns about non-linearities, Jeffrey? :-)

True non-polarised electrolytics are deliberately formed with two equal foils (I don't know how!) so they never start passing any significant reverse current. Whereas, normal electrolytics have just one foil formed and the other has just a very thin oxide film on it (ideally there'd be nothing, but aluminium likes to cover itself with something), so has very little reverse withstand capability. Two of these in series therefore don't give much chance of behaving like a nice linear capacitor! But that's just my thoughts.

[ppppenguin]

If the electrolytic is working as a coupling capacitor then it shoudl have a minimal AC voltage across it though it may or may not have significant DC. However these are in the convergence circuits as shown in this clip from the Thorn 2000 manual.

I'd be hard pressed to guess whether they are meant purely for coupling decoupling or are meant to be doing some kind of filtering or waveform shaping. On reflection, seeing as that whole lump of circuitry is sitting effectively in series with the bottom end of the frame scan coils then there is some wave shaping going on.

[pwdrive]

It's been a long time for me but I would say parabolic shaping.

Lawrence.

[Kalee20]

I'd almost taken it for granted that there was some filtering or waveform shaping going on! Because if the AC voltage across the capacitor was negligible, then the value wouldn't have been critical to a factor of 2 or so, and so the original question wouldn't have needed to be asked!