From: Phil Hobbs
Subject: Re: Ceramic Capacitors
Date: Wed, 25 Sep 2002 10:07:50 -0400
Organization: IBM T. J. Watson Research Center
NNTP-Posting-Date: 25 Sep 2002 14:07:52 GMT
X-Mailer: Mozilla 4.61 [en] (OS/2; U)
Matthew Kendall wrote:
> C0G good, X7R and others bad. In particular, use only C0G in PLL loop
> filters and anywhere else near VCOs unless you want lots of microphony. One
> VHF transceiver I saw was so bad it didn't need a microphone, you could just
> shout at the loop filter.
In order to get a lot of capacitance in a small space, X7R and Z5U caps
use dielectrics containing barium and strontium titanate, which have
huge dielectric constants. Unfortunately, they're also piezoelectric.
Using sintered powder with (we hope) random grain orientation helps a
great deal, but external effects such as mechanical stress due to CTE
mismatch, sustained high-temperature operation under high bias, and that
sort of thing, causes a net piezoelectric response to reappear.
And, of course, since the cancellation is only statistical anyway,
piezoelectricity will always appear at some level.
NPO ceramics are not piezoelectric, and neither are most kinds of
plastic films. Microphonics from actual capacitance variations happen
only with air-spaced capacitors, e.g. panel mount variables or jiggling