Post1: Chunk of CB's text from EW+WW April 1995, page 287.
+---||---+ tsr L | C | o---/\/\---///---+ +---o | Rp | +--/\/\--+Reduces to:-
esr C o--/\/\-------------||--------o
Rp comprises. a) Dielectric loss due to molecular and interfacial polarisation. b) Leakage resistance measured at dc.
tsr is the true series resistance cause by actual metallic resistances in the component makeup.
2 esr = tsr + Rp/[1 + (w.Rp.C) ].esr = tan(delta) x Xc.
esr = cos(phi) x |Z|.
Equivalent series resistance tends to reduce with increasing frequency, but by considerably less than the theoretical halving for each doubling of frequency. Ultimately attaining a minimum value when Xc = Xl.
Also in that article is a listing and graphs of the power loss due to harmonics in a 22nF used to snub a rectangular wave of about 320V peak-peak amplitude. It appeared to suffer about 7.4W losses. CB notes that an X7R ceramic capacitor failed quickly.