Problem about crystal oscillator (32.768k)

[slchen]

Dear all:

I am designing a 32k crystal oscillator.
And I have done the AC simulation to ensure the loop phase and loop gain!
The followed figures are the schmatic and the simulation results.
Please see the picture of the simulation results.
We can see there are two operation points A and B when phase=0.
At point A, the loop gain is larger than 1.
At point B, the loop gain is smaller than 1.
Are this results correct?
If yes, how do I ensure that the operation point of the crystal osc. is at A (not at B) in real chip.

Sincerely,
slchen

 

[davidwong]

I think operation point B will not happen as the gain is not larger than 1.
it can not oscillate at point B.

 

[butterfish]

B is the series resonance point of LC network,
It can oscillate only if the phase shift of amplifier is 360degree(actually it is very hard to control).
Now your phase shift of amp is about 180degree, it only can work on parallel resonance point A.

ragards
fish

 

[vrkrao]

If you find out the input impedance of the oscillator, the resistance at the freq. of oscillation should be much larger than the external crystal's resistance to ensure oscillation. The resistance value also influences the oscillator settling time.

 

[sk1105]

Dear all:

I am designing a 32k crystal oscillator.
And I have done the AC simulation to ensure the loop phase and loop gain!
The followed figures are the schmatic and the simulation results.
Please see the picture of the simulation results.
We can see there are two operation points A and B when phase=0.
At point A, the loop gain is larger than 1.
At point B, the loop gain is smaller than 1.
Are this results correct?
If yes, how do I ensure that the operation point of the crystal osc. is at A (not at B) in real chip.

Sincerely,
slchen

what's the R11 used for in your schem? for sim or one part of your osc circuit?
your loop gain is vdb(v2,v0)?
i tried almost same schem, but failed. it looks flat near resonate freq.
anyone can give some sugestion. Thanks.

 

[Hero]

Hi,

I think that Dalas 32768Hz oscilator with temperature compensations is better solution. It is very difficult to design temperature compensated oscilator with long term stability.