Micro strip capacitor issues and questions

Most of the capacitance is in the FR4, not air. FR4 has a dielectric constant around 4.6

Why not use a proper surface-mount ceramic cap?

The dielectric constant of FR4 is not well controlled, but such a capacitor will work fine. Figure about 15 pF per square inch for

The temperature coefficient of an FR4 capacitor is bad, ballpark 900 PPM/K. I can't recall the sign, probably negative. Dielectric absorption is rotten, too.

Tempco might be considerably better with the interdigitated capacitor. The TCE of FR-4 is very bad in the Z direction; not so bad X-Y, because of the way the glass fibers run.

Best regards, Spehro Pefhany

Besides John and Spehro's answers, there may also be incentive to keep everything on the top layer (or the signal layer, anyway) due to an internal metal core, or contiguous ground plane requirement. Typical of RF stuff.

The downside, of course, being: you get so much more capacitance to the ground plane than the coupled conductor, i.e., the local transmission line impedance is so much lower, that such a construction probably isn't useful.

Tim

-- Seven Transistor Labs Electrical Engineering Consultation Website:

Are you sure it is a capacitor ?

If it is an interdigital filter, those "fingers" are about 1/4 wavelength long (consider velocity factor) thus each "finger" acts as an independent 1/4 wave resonator. The spacing between fingers (of opposite sides) just determine the coupling coefficient between various resonators.

If it is a capacitor and not an interdigitated filter, then the capacitance is the "fringe" electric field, and the majority of the field will be going through the FR4 laterally.

Google "micro strip filter" and "transmission line filter" -- what you find may tell you more about what you have.

Even surface mount capacitors have self inductance, so at some frequency it is going to have a series resonance, but above that it will have inductive reactance, not nice feature e.g. as a bypass capacitor.

For this reason various micro strip line tricks are often used.

I don't mind if my bypass caps are net inductive above some frequency. All I want is low impedance. The bigger the capacitance, the better, even if it means a lower SRF notch.

Not for bypass caps! The best high frequency bypass is a copper pour near a ground plane, sprinkled with real caps.

Cheap 0603 ceramics work fine at 5 or 10 GHz.

On a sunny day (Mon, 06 Apr 2015 14:32:36 -0700) it happened John Larkin wrote in :

An at the end shorted half wavelength line will be low impedance, and so can be used to short RF at that frequency, etc.

For wideband, mHz to light for example, you need the caps.

I just grabbed a random piece of FR4 copperclad.

It measures 48 mils thick overall. Capacitance is 23.5 pF per square inch with a positive TC.