Ferromagnetic materials have very high magnetic permeability, but saturate at some point; in some cases, especially for extremely ferromagnetic materials like the electrical steel used in transformers, the transition is quite abrupt.
A major problem in the historical development of radio was the "detector": some way of converting the high-frequency AC signal of the detected radio wave into a DC or low-frequency signal that could be used, for example, to activate a solenoid. This was eventually solved by the development of the vacuum-tube diode and later the semiconductor junction diode, but before that, there were a number of Rube Goldberg contraptions, some of which remained in use for a long time in special circumstances: the "coherer", which sintered metal particles together with the RF energy and then measured the DC resistance of the result; the "cat's-whisker detector", a delicate Schottky diode made with a point contact between a finely-pointed wire and a crystal of a semiconductor such as galena, iron pyrite, carborundum, or even the iron oxide on a razor blade of a "foxhole radio"; Marconi's "magnetic detector", which used the nonlinear hysteresis behavior of moving iron wire to convert an RF magnetic field into a tiny DC voltage; and Fessenden's "electrolytic detector", which used the electrolytic formation of a layer of bubbles on a fine platinum wire electrode to preferentially impede current in one direction. Somewhat related is the "mercury-vapor rectifier", which uses the enormous difference in work function between mercury and graphite to conduct in only one direction.
It occurs to me that the saturation transition in low-hysteresis electrical steel could be used to form a detector for frequencies up to some limit, as follows. You bias the primary winding of an iron-core transformer almost to saturation with DC, then superimpose the AC signal on it. The part of the AC signal opposing the DC bias will dip into the high-permeability region and will therefore see strong inductive effects --- a high impedance, either inductive (in the case of an open-circuit secondary winding) or resistive (in the case of a dummy load connected across the secondary). The part of the AC signal in concert with the DC bias will experience much smaller inductive effects, perhaps two orders of magnitude less.
This should give you an entirely-solid-state "detector" that works without any semiconductors or vacuums.
I think this device is limited in frequency only by hysteresis and eddy-current losses, which increase linearly with frequency. In this application, though, much larger losses are acceptable than in the usual applications of transformers: a 1%-efficient detector is still usable. Wikipedia tells me that laminated-steel transformers with especially thin laminations are still sometimes used at 10kHz, so I am guessing that this detector should work usably with a steel core up to some 100kHz, and with ferrite or powdered iron, up to 1GHz.
You should be able to substitute a permanent magnet for the DC bias, eliminating the need for a power supply.