Infinite impedance detectors got their name from their extremely high input impedance. Of course, the real input impedance is anything but infinite. But it is relatively large. Vaccum tube based infinite impedance detectors have been around for a long time . They promise high-fidelity and low harmonic distortion compared to simple diode detectors when used for AM-demodulation .
The following circuit was used for testing purposes:
The working principle is rather simple: The active device, in this case a BF256B JFET, is self-biased by R103 in a way, that the gate to source voltage is essentially equal to the pinch-off voltage of the JFET. Therefore, the negative half-wave of the incoming AM-signal will have little to no effect on the voltage seen over R103. The positive half-wave, however, allows the JFET to conduct more. Thus the source voltage will follow the positive half-wave of the AM-signal applied to the gate of the JFET. The impedance “seen” by the AM-source is solely the (typically very high) gate impedance of the JFET in parallel with R102 (100 K). The previous stage is, therefore, not loaded all too much by the infinite impedance detector.
C104 plays a critical role in the demodulation process: Its job is to low-pass filter in order to remove the rf-component (AM carrier) contained in the positive half-wave. If the value of C104 is chosen too high, the higher audio components will also be surpressed. If, on the other hand, the value of C104 is too low, the AM carrier may not be surpressed satisfactorily. 10 nF works fine for AM-voice transmissions with a cut-off of about 3 kHz. For broadcast AM-signals, the value should be lowered significantly. Literature on vaccum tube implementations of this circuit suggest that the time-constant of R103 and C104 should be chosen to be several times the period of the lowest carrier frequency.
R101, C102 and C103 have been omitted in my test-circuit. R101 should be around 100 Ohms. C103 should be chosen so that the lowest frequency components of the audio signal will not be impeded. C102 is not too critical, 100 nF will be fine for an intermediate frequency (IF) of 455 kHz or higher.
For the first test I used a common IF-frequency of 455 kHz. The 455 kHz was amplitude modulated with a single 1.5 kHz tone at a 80 % modulation depth. The signal was then fed directly from the SDG 1032X signal generator into the gate of the previously shown circuit.
As shown on oscilloscope, the 1.5 kHz tone was demodulated perfectly. The amplitude of the input signal could be varied between about 500 mVpp and 4 Vpp without any visible distortion. Out of curiosity I checked if this circuit can also handle a IF-frequency of 10.7 MHz.
As shown above, an AM-modulated signal at an IF of 10.7 MHz is also not a problem fo this circuit. As a matter of fact, the circuit actually worked well up to 30 MHz. However, a drop-off of the output amplitude was observed starting at around 20 MHz.
Due to popular demand, I published a YouTube-Video about the JFET-based infinite impedance detector on my channel:
Links and Sources:
 Chaffe, E. L. Cobine, J. D., Mimno, H. R., Cooke, S. P., Morris, L. W., Cornett, R. O., Stockman, H., Githens, S., Tatum, G. R., LeCorbeiller, P. E., Wing, A. H., (1947): Electronic Circuits and tubes: By the Electronics Training Staff of the Cruft Laboratory, Harvard Univ., New-York, U.S.A.: McGraw-Hill Book Company, inc.
 Scerri, F. (2006). High Fidelity AM Reception. Elliott Sound Products. Retrieved July 29, 2022, from: https://sound-au.com/articles/am-radio.htm