Wasted about an hour using Baby GOOgle and got nowhere. Experimentally tried the CMOS Pierce circuit with C1 = C2 = 470pF and
5K from output to xtal. VERY sensitive and crystal dependent as to what supply voltage to use. Started at zero and slowly increased it until i got a stable frequency and power off / on restart. Every crystal, with out regard to frequency (23Khz or 100Khz) had different criteria. They all stopped oscillating above a given voltage.Yet, stealing the PCB from a mouse, removing its 32Khz crystel and putting a 2-pin socket allowed me to try every tuning fork crystal (including the original one) at the 5V that the mouse likes. Not "instant" on; took 0.5-2 seconds for stable start-up. There is a "funny" IC there that does it all. A 10Meg scope probe to either xtal pin will kill the oscillation.
Anybody have access to a (discrete part) working circuit?