@GK!
Greetings from the land of woebegone experimenters.
Finding white papers on the following is difficult and understanding them even more so.
Look into 'para-metrically pumped toroidal capacitors with sequential poloidal pump windings'. It is the only way, AFAIK, to produce a DC output from a varying control signal. Basically, it is a mix of magnetic amplifier theory with dielectric charge pumping vs. iron core saturation control.
This is possible when applying a clean varying signal, sequentially and in quadrature fashion to segmented poloidal windings wrapped around a toroidally wound capacitor.
I made my first by slicing a 40,000uF 400VDC Aluminum electrolytic capacitor into doughnut shaped segments.
There is a great deal of difficulty trying to maintain electrical isolation between the layers where the slicing is done. I think I went through a couple of dozen caps before I found that the sharpest possible blade works when slicing the chunks off on a slow lathe. (I have an almost endless supply of large caps being replaced during the PM processes where I work.)
Simple light handling will cause the layers to bend and short out but I did succeed after many tries.
The next step was to protect the edges of those layers from physical damage by coating them with Robroy Class I Div I ceramic cement. Later, I found out that super fine grout compound also worked.
After making four connections (one each on each end of the two metal film layers - In & Out) was even more difficult.
Charging the dielectric is most easily done by sitting the resulting toroid shaped device centered around the output point of an old fashioned lab room air ionizer.
The ionizer is intended to provide charge for the capacitor dielectric vs. applying voltage to the former capacitor's layers.
The quadrature, poloidially wound, windings were fed from a simple MOSFet H-Bridge circuit driven by a signal generator.
So, the simple hypothesis to be proven was that the small toroids were actually dielctric core with toroidially wound capacitor plates and DC was obtained from the former capacitor plates by pumping (in quadrature fashion) that charge around thus creating an IN and OUT for a completed load circuit.
Unfortunately, the best I obtained was ~1VDC with the pump frequency superimposed upon it. It had a very low impedance ( I could dimly light a 20W 6V incandescent lamp) but when all was added up it was only about 10% efficient.
I've seen studies from CERN using similar methods for multiple reasons but I'm not looking to create a black hole in my man-cave.
Not much time on the bench anymore but there were some interesting results
EDIT>>
BTW, I could imagine using two such devices over a two port lab air ionizer with opposite quadrature rotation to quadruple the combined output voltage but never went that far.