Plz Help: The Design of Crystal Osc (32.768KHz)

walkingsun said:

I have the oscillation circuit:
inverter + shunt res(1MOhm) +2 load cap(30pF, at the input of crystal)

The crystal is 32.768KHz. (R1=35K, C1=1.9fF, L1=12255H, and C0=800fF)

But the phase shift is not enough at the freq.

What can i do?

thank you!

Click to expand...

xxargs said:

walkingsun said:

I have the oscillation circuit:
inverter + shunt res(1MOhm) +2 load cap(30pF, at the input of crystal)

The crystal is 32.768KHz. (R1=35K, C1=1.9fF, L1=12255H, and C0=800fF)

But the phase shift is not enough at the freq.

What can i do?

thank you!

Click to expand...

Incrase shunt resistor to couple of MOhm - to low value can act as RC-oscillator with load-cap.

Use serial resistor on inverters output in value between 220 KOhm to 1 MOhm or you vill overload 32768 kHz crystal and can starting and working on spurr and overtone-frequency. Serie resistanse on output and load-cap working as low-passfilter outside qrystal frequency.

(spurr can never simulated unless you have spurr modelled in your crystal model)

crystaloscillator with high Q-value take loooooooooong time to starting if you simulate... we talking hundreds of thousen periods in high resolusion simulation and struggling with diffrence frequency and amplitude variation long time before oscillating inside crystal going to big enough and make dominating current and voltage in circurits. - check C1 in beginning alway rise in amplitude over long term time under simulating, and can be very slow start in beginning and amplitude ringing and ripple.

And simulator not have termic noise, have limited numeric resolution and limited timestep resolution can find state (machine) to never starting oscillating on right frequency in infinite time simulate - never happing in real world circurits..

If C1 not make couple of thousen Volts, your crystal have not starting work yet in you simulation.

... and one metode shorter simulate time is charge C1 to around 100000 Volt before simulate and if voltage over C1 still incrase or lowering in simulation - adjust voltage again to find steady state working point little faster.

In steady state working - measure current out from crystal and calculate power with crystals resistanse (or measure power over crystals resistance) and check this value are lower than allowed from crystals datasheet - to high, you must incrase serieresistance from inverter output.

Click to expand...