Jim Wilkins wrote:

"Michael A. Terrell" ? wrote in message
news ?
? He, and several
? others couldn't be convinced of SRF in resistors and capacitors
? threw a
? hissy fit when I designed a broadband DC block to replace the pile
? of
? unlabeled junk that was in use. It consisted of a stack of surface
? mount caps. I use .33 uF. .033 uF .0033 uF 330 pF and 33 pF.

Like most I learned capacitors as ideal components.


Unfortunately, capacitors are the weakest link in most electronics.


It was an
eye-opener to sweep them with a network analyzer and see how much they
deviated, especially the high-Q MLC ones at high frequencies. The
design rule at Mitre (where DC was defined as below 1 GHz) was to do
as you did even with bypass caps, like 0.1uF in parallel with 1000pF,
sometimes stacked on the same pads.


Some engineers swear by that idea, and others ridicule them. I
worked with very low noise RF gear, and you had to keep all the crap you
could out of the power rails. Any noise, in any rail including ground
would cause excessive phase noise in the wide band synthesizer. It was
band switched with forward biased diodes to reduce the length of the
inductor in the VCO. I've had idiots try to tell me that that method
doesn't work, or that a fraction of a mA is enough to turn it on. You
need enough current to turn it on hard, without damaging the silicon to
keep the RF from modulating the current through the diode. I just
missed buying one of the modules on Ebay by a few bucks, when my browser
crashed 30 seconds before the auction ended.


As an experiment I attached a .1uF MLC to a capacitance meter and
heated it with a hot air gun. The capacitance dropped to ~ 1/10th of
its room-temp value. A silver-mica changed almost too little to
detect.


You should have seen the huge, uncased silver mica caps used in RCA
UHF TV transmitters. They were over $600 each, in the '70s. They were
in the interstage RF coupling, at 1 KW.

Can you do that heat test with any NPO/COG ceramic caps?