Plausibility of BFO discrimination scheme

[Teleno]

I'm submitting this idea for comments. Perhaps the most knowledgeable would care to point to any misconceptions.

System structure:

Free running LC oscillator -> Peak detector -> Amplifier -> Output A (amplitude signal)
. . . . . . . . . . . . . .-> Frequency to Voltage conversion (PLL) -> Output F (frequency signal)

Even without a target, a change in the frequency of the LC tank also translates into a change in the oscillator's amplitude. This is due to the oscillator's characteristics.
Therefore the amplitude signal is a combination of two factors:

1. The resistive component of the taget (R)
2. The frequency deviation caused by the target.

The system can be calibrated to eliminate the 2nd factor by using a purely magnetic target (a ferrite), this way the relationship frequency/amplitude can be established (brown line in the graph).
Similarly, a diamagnetic good conductor (pure Ag) is used to calibrate the other extreme of the frequency deviations (green line).

We now got two linearly independent vectors, though not orthogonal, and any target falls in between.

In order to separate the two components of the amplitude we need to convert the R and F signals to an orthogonal coordinate system where the x-axis is only affected by frequency and the y-axis is only affected by resistivity. This can readily be done using Algebra, in particular a change of basis, once the two vectors above are known. In the new basis the X and Y components would be equivalent to those produced by an IB system.

Any weak points in my reasoning?

[Davor]

Yes. For weak targets the frequency change is small.
There is also another frequency measuring approach worth mentioning: pulse counting. It had its peak popularity in the tubes era, but compared with PLL it offers larger dynamic range, and better S/N. Pulse counting device may be built stupidly simple.

[Teleno]

Yes. For weak targets the frequency change is small.
There is also another frequency measuring approach worth mentioning: pulse counting. It had its peak popularity in the tubes era, but compared with PLL it offers larger dynamic range, and better S/N. Pulse counting device may be built stupidly simple.

Are you referring to a charge pump like the LM2907?




I believe that PLL is better at detecting small frequency changes because the VCO gain in terms of V/Hz can be set very high, for example the CD4049B by playing with the ratio of R1 and R2.

[Davor]

It can be realised with LM2907, but also with other simple circuitry. Try google with "pulse counting FM demodulator" and keep open mind because the majority of circuits out there are from tube era. There were also some patents that improved on the basic idea by introducing multiphase detection etc. but all quite old.