[ARC5] Diode effect in Caps.

[J. Forster]

The effect George describes is called 'dielectric relaxation' and is a result of
something akin to a piezoelectric effect.

If you consider a capacitor as a pair of plates on each side of a dielectric
slab, there is a force between the plates when a voltage is applied that
squishes the plates together. Just like when you squish a piece of cheese, when
the force is relaxed the dielectric slab does not fully return to its original
dimensions right away, but will recover over some period of time, perhaps hours,
perhaps days. This relaxation causes a slight redistribution of charge in the
dielectric block, yielding a terminal voltage. High voltage capacitors, such as
those from 'Plastic Capacitor Corp.' can develop quite a charge due to
dielectric relaxation effects. Note that this is a linear effect...  reverse the
polarity of the HV and the 'relaxed' voltage will reverse also.

A similar effect can occur in ceramic capacitors. Their incremental capacitance
can increase markedly when a DC bias is applied. Some years ago, I built a
triangle wave form generator which produced a slightly curved wave form. It
turned out to be caused by the voltage coefficient of capacitance. Moral: Use
high quality caps when making integrators. Again, this is linear...  that is the
capacitance v. applied voltage curve is symmetrical about zero applied bias.


The effect Keith originally described is non-linear..  the leakage is in only
high for one applied bias polarity.

-John