The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> xtal oscillator
https://designers-guide.org/forum/YaBB.pl?num=1121183264

Message started by qadeer on Jul 12^th, 2005, 8:47am

Title: xtal oscillator
Post by qadeer on Jul 12^th, 2005, 8:47am

Hi All,

I've just joined this forum and found quite interesting and useful.
I'm currently working on xtal oscillator and facing a startup current problem. Can anyone tell me how do we chose the initial current of inductor which is given in the inductor propery when it's picked from samGeneric library. My AC response shows 5dB gain (and 0 phase shift) at resonace frequency but when I run transient simulation, the circuit fails to oscillate at lower initial current of the inductor. The circuits start oscillating at higher current (1mA) so I'm not sure how much startup current the inductor will get on actual silicon.
Looking forward towards the reponse.

Thanks and Regards,
Qadeer

Title: Re: xtal oscillator
Post by Ken Kundert on Jul 12^th, 2005, 10:46am

Did you do anything to excite the oscillation in transient? Apply initial conditions? Hit it with a short pulse?

-Ken

Title: Re: xtal oscillator
Post by qadeer on Jul 12^th, 2005, 8:46pm

Hi Ken,

Thanks for the response.
I'm already applying the initial condition but it still does not oscilate if I don't provide the ic in inductor. I also tried to simulate the circuit with vdd ramp-up ( which is likely to be the case on actual silicon) and found that there is a correlation between supply ramp rate and inductor initial current (which is expected). At slower ramp rate I need to provide higher initial current in the inductor to make it oscillating. My only conern is that how do we decide that how much initial current would the xtal get on actual silicon. Other oscillator which are RC or GmC based starts oscillating with intial condition only I mean if we give a 0V or vdd at any node then it start oscillating while in case of xtal, the same is not true.

Regards,
Qadeer

Title: Re: xtal oscillator
Post by Ken Kundert on Jul 13^th, 2005, 10:31am

People often do not properly start oscillators when simulating them. They either forget to start the oscillator, or they use a poor method to start it. In this case, the fact that your oscillator does not start or starts slowly is not an indication that your oscillator is flawed.

I give instructions on how to start an oscillator in my "Spice and Spectre" book. I suggest you read them. It is important to choose a method and place for applying the initial stimulus so that it efficiently couples into the oscillatory mode of the oscillator. Generally, ramping the power supply is a poor approach.

-Ken

Title: Re: xtal oscillator
Post by ywguo on Jul 13^th, 2005, 10:25pm

Hi, Qadeer,

Quote:

My only conern is that how do we decide that how much initial current would the xtal get on actual silicon.

Don't worry about the initial condition of the actual silicon. If the AC analysis proves that your xtal osc has enough gain at the 360^o phase shift, the actual silicon must oscillate because there is noise in the circuit. The noise makes the xtal osc deviate from the balance point and initiates the oscillation, which is different from the SPICE transient simulation.

There isn't any noise in transient simulation. Thus sometimes you have to initiate the oscillation using an initial current for inductor, or kick the oscillator with a small pulse. In one word, the xtal oscillator needs some energy to start. But it is depends on your circuit that the xtal osc needs the initial current or so. Sometimes the xtal osc starts oscillating without any initial condition because the SPICE transient simulation has small intrinsic noise due to the iteration and integration algorithm. For example, it is easier to start for method=trapzoid than method=gear.

Do you have any other concerns?

Best regards,
Yawei

Title: Re: xtal oscillator
Post by qadeer on Jul 14^th, 2005, 8:26am

Hi Yawei,

Thanks for the answer.

Best Regards,
Qadeer