Fast start-up crystal oscillator know-how

[hung_wai_ming@hotmail.com]

xtal start-up time gm

Did anyone come across some design techniques for fast start-up XTAL?
I did some previous designs already but always faced a problem for too long
transient simulation for XTAL. What I always did was to simulate AC response
to check the loop gain and got it done, however, it was not comfortable without
a detailed transient performance to check the duty cycle or phase noise.

 

[szekit]

i would be interested to know too....

so in your past experience, you did AC sim and taped out?

 

[hung_wai_ming@hotmail.com]

The first time, I did transient as well, and taped out to be successful.
Therefore, second time, I seldom did transient as it took too long time
and I found AC response is always quite enough to gaurantee good design
if you can give enough margin and simulate ALL corners..

 

[holddreams]

Simulating a crystal oscillator can be a frustrating task.
The higher the Q(originally the quality factor),the longer it takes for the oscillation to build up.
If you want to measure the frequency accurately,you need to do this in very fine steps.

The time need to build up oscillation equals Q*T,where Q is the quality factor=2*pi*f*L1/R1, and T is the period of the XTAL,f is the frequency of the XTAL.
The parameters R1 and L1 are given in crystal model.
So before simulating the XTAL,you can calculte the time for oscillation.

 

[butterfish]

Simulating a crystal oscillator can be a frustrating task.
The higher the Q(originally the quality factor),the longer it takes for the oscillation to build up.
If you want to measure the frequency accurately,you need to do this in very fine steps.

The time need to build up oscillation equals Q*T,where Q is the quality factor=2*pi*f*L1/R1, and T is the period of the XTAL,f is the frequency of the XTAL.
The parameters R1 and L1 are given in crystal model.
So before simulating the XTAL,you can calculte the time for oscillation.

Hi, holddreams,

Do you means,startup time=Q*T=2*pi*(f*T)*L1/R1=2*pi*L1/R1?
that is the startup time only related to crystal?
Does the startup time depend on Gm of amplifier?
It seems startup time strongly depends on amplifier!

 

[holddreams]

refer this

Simulating a crystal oscillator can be a frustrating task.
The higher the Q(originally the quality factor),the longer it takes for the oscillation to build up.
If you want to measure the frequency accurately,you need to do this in very fine steps.

The time need to build up oscillation equals Q*T,where Q is the quality factor=2*pi*f*L1/R1, and T is the period of the XTAL,f is the frequency of the XTAL.
The parameters R1 and L1 are given in crystal model.
So before simulating the XTAL,you can calculte the time for oscillation.

Hi, holddreams,


Do you means,startup time=Q*T=2*pi*(f*T)*L1/R1=2*pi*L1/R1?
that is the startup time only related to crystal?
Does the startup time depend on Gm of amplifier?
It seems startup time strongly depends on amplifier!

 

[butterfish]

Dear hung_wai_ming,
Maybe this paper can give you some comments.

Dear holddreams,
I didnt find startup time=Q*T in your link.could you give more information about your ref?

 

[hung_wai_ming@hotmail.com]

Thanks, I got this already.
Do u mind sharing your design experience with me?

 

[jcpu]

Hello,

Allow me to make sure,
Do ou really want to simulate faster or
do you want a crystal oscillator circuit that start-up faster???

 

[hung_wai_ming@hotmail.com]

[quote="
do you want a crystal oscillator circuit that start-up faster???[/quote]

i am working on this solution. If you have solution, let's share real experience

 

[szekit]

i haven't done any crystal oscillator design. so by using the inverter to generate the 180 degree phase shift, does it mean that the crystal always generates another 180 degree phase shift? will there be 90 degree phase shift from the crystals?

 

[horzonbluz]

Oh, this paper help me.
I haven't designed crystal circuits before.

 

[pseudockb]

Dear All,

I am a new to xtal oscillator design and I hope I can find some useful tips here regarding the start-up transient simulation. I found this website quite useful: **broken link removed**. However, when I tried to use the method in my oscillator circuit by giving a initial condition at the inductor value, I could observe the oscillation with correct frequency initially but the oscillation died down finally. What does that mean? Is my oscillator really working? I am using spectre for the simulation. Thanks!!!

 

[buckaroo]

hi
u can find Q*T in Tom Lee's book, 1st edition, the charpter RLC
good luck

Dear hung_wai_ming,
Maybe this paper can give you some comments.

Dear holddreams,
I didnt find startup time=Q*T in your link.could you give more information about your ref?