Strange capacitor specification (to me)

Thread Starter


Joined Aug 21, 2008
I copied this from a film capacitor datasheet. No, it is not polarized, its metalized polyester.


My question: Why is it not ok to pass AC current through this part?

Edit: Come to think of it the capacitor sees the current as AC and it would see the voltage as drawn above. I might have to contact their "technical engineer" (as opposed to the artistic engineer?) to get an answer.
Last edited:


Joined Oct 7, 2019
"or DC-blocking circuits" What? Like the input and output of a amplifier? In a DC blocking circuit, some circuits, the voltage could cross zero.
Last edited:


Joined Mar 2, 2015
I copied this from a film capacitor datasheet. No, it is not polarized, its metalized polyester.

My question: Why is it not ok to pass AC current through this part?
Beats me. I've never encountered any such restriction on film capacitors, and reading the data sheets on the ones I have in the lab I don't find anything like what you posted.

'Tis a mystery...


Joined Sep 17, 2013
'Tis a mystery indeed. If there is no polarity marking it can clearly accept DC of either polarity. It can also accept high ripple, judging by the graph. That's getting to look a bit like AC.
I wonder if the manufacturing process leaves the dielectric with a residual polarised property (but then you'd expect a specified orientation in use)?


Joined Oct 2, 2009
In the original Apple Macintosh computer there is a 4.7μF non-polar capacitor in the Hor Sync power circuit.
This was notorious for going bad. Perhaps this is a similar example where even though the capacitor is designed to be non-polar it doesn't endure having repeated reverse polarity.


Joined Aug 27, 2009
Just a guess, it might be a dielectric property frequency limitation.
The frequency at which a dielectric is used has an important effect on the polarization mechanisms, notably the relaxation time displayed by the material when following field reversals in an alternating circuit.

  • Case 1: If the relaxation time for polarization is much longer and slower than the field reversals, the ions cannot follow the field at all and losses are small.
  • Case 2: If the relaxation time is much faster than the field reversals, the polarizing processes can easily follow the field frequency and losses are small.
  • Case 3: If the relaxation time and field frequency are the same, the icons can follow the field but are limited by their lag, thus generating the highest loss with frequency.
Therefore, dielectric losses are highest at the frequency where the applied field has the same period of the relaxation process. Ceramic dielectric formulations always show a range of relaxation times over the frequency spectrum, since these materials consist of polycrystalline matter. In high frequency applications, this parameter is often known as the Q factor, which is the reciprocal of the loss tangent: Q = 1 / (tan δ)
The dielectric component tan δD is a measure of the losses associated with the dielectric (i.e. the energy wasted to polarize and repolarize the dielectric in two opposite directions for successive half-cycles of the AC voltage). It determines self-heating at low frequencies: In polypropylene capacitors, tan δD remains approximately constant with frequency and will typically result in a value of approx. 10-4. In polyester capacitors, tan δD is considerably greater and increases with frequency. So these capacitors display noticeably higher, overall dissipation factor at lower frequencies (cf. figure 14).


Joined Jan 23, 2018
So who is going to ask their technical engineer?
The reason that some capacitors are not able to work with ripple currents is that the effective current flowing through the internal (equivalent) series resistance will produce enough heat to cause probolems, such as failure of the capacitor. It has to do with the polarization changing back and forth and that atomic level motion causes heating in some materials. Also called "dielectric absorbtion", but I think that is actually something different.

Analog Ground

Joined Apr 24, 2019
My guess would be the self healing characteristics are different between DC and AC applications. I recall the result of a self healing event for AC was not as reliable as for DC. Or, in other words, the characteristics of the healed location are different. Self healing is the ability of the capacitor to "clear" or burn away any tiny short through the film. This is really a fine point and would take some serious research to confirm.
EDIT: This article eludes to a difference between self healing characteristics for DC and AC but is not very clear (no pun intended).
Last edited:


Joined Oct 29, 2017
I'll go with the translation poblem explanation. Plus, the tech author who drew up the datasheet wasn't in his element when it comes to electronic engineering - the formula for Irms doesn't work for a rectified sine wave. Irms will be zero if there's no load. And to pick another nit, either the tech author wasn't great at drawing sine waves by hand, or the harmonic distortion of the input would invalidate the Vrms formula.

To be slightly more serious, I wonder if the self-healing properties wouldn't be up to spec for AC input. In other words, don't use it as a transient filter on the raw mains input.