About 37 years ago, I built, as a test, a low power AM generator. Its output signal was:
 double sideband suppressed carrier (DSB-SC) and,
 the frequency of its ‘suppressed carrier’ varies, at a rate of 6 Hz (sinusoidal), between 405 and 505 KHz.
I chose this range of spectrum to emulate the output of an AM IF stage (at 455 KHz), though having a much wider bandwidth than of standard ones which is about 10 KHz.
In this test, I recovered the audio modulating signal by a linear demodulator that uses:
 One standard PLL IC, CD4046.
 Digital flip-flop and gates.
 Linear comparators.
 Non-precise resistors and capacitors.
 No inductances (LC tanks) or selective active filters
I did this test because, at that time when I was at the American University of Beirut, it was known that a simple low cost linear demodulator for a DSB-SC signal cannot be made. In other words, it was also universally accepted that recovering the modulating signal of a single sideband suppressed carrier (SSB-SC) one is much easier than of a DSB-SC one.
So I liked to prove, by my test, that the previous statement is false and its opposite is true even with a ‘varying frequency’ of the suppressed carrier. And this is natural, because it would be illogical that receiving two sidebands is worse than receiving just one of them ;)
By the way, the demodulator of this test works for any AM index; from m=0 (no modulation) to infinity (DSB-SC). And it is suitable for hi-fi audio signals as well (as long the bandwidth is wide enough). Also its circuit, as mentioned above, could be integrated in one low cost IC.
So, after about 3 decades, I expect that what I did as a test became known in a few faculties of electronics in the least. And, therefore, some members here who are interested in communications also know its topology. Otherwise, it will die with me, sooner or later, since I am already 67 and I wasn't born in the West ;)