The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> RF Design >> 3rd overtone xtal oscillator
https://designers-guide.org/forum/YaBB.pl?num=1330113243

Message started by ali kotb on Feb 24^th, 2012, 11:54am

Title: 3rd overtone xtal oscillator
Post by ali kotb on Feb 24^th, 2012, 11:54am

iam new to this site , hope i get the best of it
iam trying to make a crsytal oscillator as reference for PLL
the crystal has fundamental 40Mhz which is not suitable to work at for certain specs in PLL, it should work at 3rd overtone (120MHz)
i need help in modeling the xtal plz
knowing that co (shunt cap)=5pF
cload=15PF
i got the values at 120MHz : Lm=3.52m , cm=0.2f
actually it is not oscillating because at high freq impedence of sunted cap c0 is too small (shorting the series Lm , cm , Rm)
the solution is to use another Lo to be shunted to co to remove that effect but the value of Lo would be 0.35u which is not integrated on chip , what can i do ? help me plz
can any one post me a practical on chip schematic for such oscillator, thnx

Title: Re: 3rd overtone xtal oscillator
Post by loose-electron on Feb 24^th, 2012, 1:58pm

ali kotb wrote on Feb 24^th, 2012, 11:54am:

With a little research you can find the equivalent LC model for a crystal.

Essentially, it is nothing more than a tuned tank, but with a very high Q
(roughly 10E5 if I recall.)

Now comes the interesting parts:

In a simulation environment, the Q is so
high that it will not simulate easily and quickly.
So you may want to adjust the resistance to represent a
value of Q = 100 (give or take)

That will make the initial simulation work easier to run.

Now, one other thing with the device is starting the oscillation
process for test and simulations.

Use an initial condition of a DC current in the inductor
that is the electrical equivalent of the crystal.

The start of oscillation process in a crystal is very slow and
takes a long time to simulate (you need to introduce noise
into the model in order to get the oscillator to start.)

The IC of current in the inductor easily gets you around this.

Try it, you will like it.

Title: Re: 3rd overtone xtal oscillator
Post by ali kotb on Feb 24^th, 2012, 3:12pm

sir , i already did plenty of simulation on spectre using fundamental xtals , and it succeeded , but when it comes to overtone , no o/p is produced
1) may be due to parallel co (having low impedance at high freq.- thus short the series L, C, R)
2) i think we need a filter to suppress the fundamental and allow the overtone (l,c) filter but the value of L will be large to be integrated , i have no solution to this
ur efforts are deeply appreciated , thank u

Title: Re: 3rd overtone xtal oscillator
Post by loose-electron on Feb 25^th, 2012, 12:31pm

ali kotb wrote on Feb 24^th, 2012, 3:12pm:

It sounds to me like you have included the electronic equivalent
of the resonance point associated with the fundamental and only the fundamental.

That probably looks like some kind of tank circuit
allowing signals to only pass at that resonance points.

You are creating the electronic equivalent of a mechanical resonance.

Your model require the inclusion of a tank circuit that models the resonance at the
harmonics as well.

so you need multiple tank circuits at the overtone frequencies.

what you add as external circuitry selects which of those tank circuits
has not been filtered out.

Title: Re: 3rd overtone xtal oscillator
Post by ali kotb on Feb 25^th, 2012, 3:02pm

u mean that i need a xtal model of multiple L, C, R as motional arms ,
to represent the harmonics. u are right sir, do u have any document to share with me for this model (i need to see their calculations).??

i use spectre cadence tool , actually when u apply pss analysis using only a single L, C,R calculated at fundamental --> i can see harmonics too

iam very confused , should i use multiple xtal modeling arms , or single one for overtones

can u help me in details and pass by an 3rd overtone xtal model
thank u ,,

Title: Re: 3rd overtone xtal oscillator
Post by loose-electron on Feb 27^th, 2012, 2:14pm

Yes, exactly, you need a structure that Looks like a bandpass filter
But - Has the capability to pass the 1F, 3F, 5F frequencies.

If you are trying to get the 3F to work, you need the 1F and 5F in the model.
To demonstrate what you attach the crystal to does not respond.

Suggest contacting the crystal vendor for a Spice model,
or an AC Bode plot of the crystal response.

If they do not have the model, the Bode plot will allow you to create your own.

You may also see some even harmoinics in what the crystal vendor has.
If so, these resonance points will need to be modelled too.