# RE: [SI-LIST] : Capacitance in a Ceramic changes ?i usedf

[email protected]
Tue, 25 May 1999 11:08:56 -0400

> the iron was 750F or so / we used MuRata and Kemet as
>vendors... what's stranger yet today we measured the caps and
>they, on the average, are about 0.055uf nom. when 2 days ago
>the same caps were reading > 0.06uf

Keith,
According to the ceramic capacitor section of the 1982 IBM Purchased Components
Manual Passives:
"Aging Characteristics

A characteristic of ceramic capacitors is their loss of capacitance with time.
The loss is associated with the ferroelectric state and geometry of the
crystalline structure of the lattice. Figure 4-21 presents capacitance loss per
decade of hours for class I and class II dielectrics. The aging rate can be
increased by a factor of 10 by applying a dc dias. Although the aging can be
increased with a dc bias the slope of the stressed and unstressed curves will be
the same. ==> The aging effect can be reversed by application of thermal energy
(~150C), but begins again when the energy is removed. <==" (my emphasis)

Basically, for every decade increase in a ceramic capacitor's life, its
capacitance will drop by a certain amount. For example, right after soldering a
ceramic capacitor to a card (effectively annealing the barium titanate
dielectric) it might measure 0.60uF. After one hour it might measure 0.59uF,
after 10 hours 0.58uF, after 100 hours 0.57uF, after 1000 hours 0.56uF, etc. I
remember a professor back at the University of Michigan who told about feeding a
"radio pill" to some large animal to measure the temperature inside its stomach.
When they started the experiment they came up with some incredible temperature
based on their calibration curve for the radio pill! Then they realized that
they had resoldered a ceramic capacitor in the temperature-measurement circuit
between the time that they had calibrated the radio pill and when they fed it to
the critter-- resetting it on its aging curve. They took another capacitor
from the same batch, soldered it on a card the same way, and tracked its change
in capacitance for a while. From this they were able to calculate a
time-offset to their original calibration curve, and finally make valid
measurements for the experiment.