Another Q&A on this site ("Ceramic caps vs electrolytic. What are the tangible differences in use?") contains good information on capacitors in general, including parameters I wasn't aware of, such as the sometimes-considerable change in capacitance of a multilayer ceramic cap as DC voltage is applied.
My search into this realm was prompted when considering use of a 10uF ceramic cap to replace a couple of 22uF series-connected electrolytics in a radio receiver AGC 'servo' application. An op-amp comparator has full open-loop gain for DC errors between its inputs, but a 10uF integrating cap between the output and the inverting input drops the gain to unity for anything much above a few Hz.
The circuit is such that the polarity across a single cap would often reverse in normal operation. I've been using a couple of electrolytics in series, plus-to-plus, and included a diode to their junction from the positive supply rail to force proper polarity of the caps. The circuit of course glitchs as the DC level at the amp's output changes abruptly, but once each of the caps charges to its new DC value the midpoint holds still enough that the diode is essentially out of the circuit. Nevertheless, a good deal of current flows in and out of those caps when the comparator corrects for a large incremental DC change, and I'm wondering what sort of trouble I'd be in if I simply removed that diode. From what I understand, electrolytics have the equivalent of a built-in diode pointing to their + terminals and should remain happy.
So my question has two parts:
- Does this use of a 'nonpolar' electrolytic violate any conditions of its use when DC is involved? and
- Does this seem a proper spot for a 10uF multilayer ceramic?
