FM Discriminator Coil

The process of reverse engineering the Piper AutoControl II is almost complete.  The oscillator is a Colpitts type with the gain and frequency stabilized with a radio choke.  The various trim, servo, and heading capacitors are in parallel with the heading indicator coil and provide the variable frequency.

The servo motor h-bridge is also straight-forward with a biasing network driving a set of six high capacity transistors.  However there is no protection whatsoever beyond a fuse.  If the discriminator sends a signal to turn on both sides of the bridge the fuse will blow.

The switching transistors provide a high impedance and high gain network for the discriminator.  A very small current into the network is multiplied by two gains (around 2500:1) to provide a way to sink the bias current for the h-bridge and turn either side of it on.

The subtlety is in the discriminator coil.  I have managed to model it and the frequency analysis behaves as described in the various old references.  However, the centre tap is made into a capacitor network, not the secondary itself.  Still, I have modeled it with a more conventional arrangement and the frequency plot seems nearly identical.

The trick is that a very careful selection of coil inductance is required along with the mutual coupling.  The output capacitors are also centre tapped providing for a voltage signal output for either side.  However, if the coupling coefficient is too high then there is a single peak at resonance and both output diodes conduct, turning on both sides of the h-bridge, and melting the transistors.

With a high impedance driver and a low coupling coefficient there is actually a small voltage drop at the centre and the servo motor drivers are safe!

More work is required.