After some fumbling, a Colpitts oscillator sorta-kinda works at 32 kHz:
Even though the Darlington provides enough DC gain for megohm-scale bias resistors, I think they still loaded the
crystal tuning fork a bit too much:
It didn’t start reliably and a JFET would probably produce better results. As it turns out, my transistor assortment currently lacks JFETs, a situation shortly to be resolved by an ePacket from halfway around the planet.
The proto board has two oscillator circuits behind the crystal test fixture, with the 32 kHz resonator hanging from the header pins on the lashup toward the rear:
My copy of Crystal Oscillator Circuits (Mattys, 1983) suggests useful component values:
The value of of the recommended shunt resistor Rhi shows why I figured the Darlington and its bias resistors wouldn’t work:
The other Colpitts circuit worked just fine for a 3.58 MHz crystal and reasonably well for a 60 kHz tuning fork, so some fine tuning should improve the results.