[Prev][Next][Index][Thread]
Re: Non-linear capacitance
Gary and all,
Shazaamm!! This just gets curiouser and curiouser! Dug a little bit
deeper and found another interesting tidbit about ferroelectric (Class
II) capacitor dielectrics such as barium titanate. It turns out that
their behavior is quite complex, and changes versus just about every
parater including Frequency, temperature, voltage, and possibly the
phase of the moon. While the capacitance declines with an applied DC
voltage, the capacitance actually INCREASES with applied AC voltage
stress... and the magnitude of the positive C versus V change is
actually greater than the effect for DC voltages - the curves I looked
at this morning indicate a 15% or more change!
Perhaps this is because of hysteresis effects (the same effects that
make a dielectric amplifier possible as in a recent post by Bert Pool).
The fact that you see identical behavior with the unit immersed in
vaseline implies that this effect appears to be coming from the HV caps
alone. Very interesting! Apparently, having non-linear elements in a
voltage divider chain would make accurate measurements quite a
challenge! :^)
BTW, one of the best sources of information for both resistors and
capacitors is a book called "Capacitive and Resistive Components" by D.
S. Campbell and J. A. Hayes, Gordon and Breach, 1994, 381pp, ISBN
288124999X... and the GOOD news is that it's still in print and at about
$28 (paperback) although not cheap, is affordable! Unfortunately, the
hardback version is $79. :^(
Safe cappin' to you!
-- Bert --
Tesla List wrote:
>
> Original Poster: "Lau, Gary" <Gary.Lau-at-compaq-dot-com>
>
> Hi Bert:
>
> You point out that in the non-linear, higher k ferroelectric (barium
> titanate) dielectrics, the capacitance decreases with increasing voltage.
> What I observed was, a 120V RMS signal read just that, but a 15KV RMS NST
> gave me a reading of 33KV peak, or 23.3KV RMS. I know that's way higher
> than it actually is. For the NST voltage to read so high, the capacitance
> of my string would have had to increase with the increasing voltage. And
> the capacitive load imposed by the probes across the NST secondary is only
> 7.5pF, so I don't think that near-mains-resonance is at play to ring up the
> voltage.
>
> I built two such capacitive dividers so that I can do a true differential
> measurement with the scope on A-B. Today I encapsulated one of the
> capacitor strings in melted Vaseline jelly to see if corona was responsible
> for the apparent decrease in impedance with increasing voltage. No
> difference, both encapsulated and open strings performed the same.
>
> Guess I better concentrate on getting those fiber optic probes built!
>
> Regards, Gary Lau
> Waltham, MA USA
>
> >Original Poster: Bert Hickman <bert.hickman-at-aquila-dot-com>
> >
> >Great post, Gary! Last year, I was thinking about doing the same kind of
> >thing, since HV ceramics are pretty inexpensive and easy to find at
> >Hamfests. However, after reading up a bit on ceramic capacitors, I
> >decided to look at other alternative.
> >
> >There are two different classes of ceramic dielectrics. The first,
> >called Class I, is a non-ferroelectric, lower-k dielectric. Capacitors
> >fabricated from this formulation do not show capacitance change versus
> >voltage, and have predictable capacitance versus temperature
> >characteristics. If you made the high-voltage section from this type of
> >capacitor, you shouldn't see the capacitance change as a funtion of the
> >applied voltage stress, and if you used NPO type (minimal change versus
> >temperature), they would work fairly well.
> >
> >The second type (Class II) are made from a higher k ferroelectric
> >dielectric (barium titanate). This formulation shows marked capacitance
> >changes with temperature and applied voltage, and (analogous to
> >ferromagnetic materials) they also show hysteris effects(!). The
> >capacitance of a Class II cap decreases significantly with increasing
> >applied voltage. For example, as the DC voltage stress is increased from
> >0 to 20 volts/mil, the capacitance may decrease by as much as -10% or
> >more!
>
> <snip>