Help me fix my crystal oscillator

[newbie_1]

help! crystal oscillator

i have setup a crystal oscillator using inverter
and the netlist is shown as below:

m0 xo xi vdd vdd pch_33 l=0.5u w=2u m=1
m1 xo xi 0 0 nch_33 l=0.5u w=1u m=1
c1 xi 0 16p
c2 xo 0 16p
rf xo xi 1meg
v0 vdd 0 pwl 0 0 100u 3
xi0 xi xi xtal

.subckt xtal n1 n2
rs n1 n5 55k
ls n5 n7 11.233e3 ic=1e-7
cs n2 n7 2.1f
c2 n2 n1 1.1p
.ends xtal



and i have simulated the open loop gain and it's about 16dB, and the gm at steady state is about 500uS(<gmmax and >gmcrit), however, after powerup, the oscillator will enter the steady state(xo and xi remains at about 1.45) and never oscillate. （ and i have used option accurate and method=gear, and transient step is 1ns)
So anyone help me!!Thanks a lot

 

[Teddy]

Re: help! crystal oscillator

Homestly for the osc startup you should use the conservative method (taks forever I know) .
Also depends on the model of you xtal. Most of the xtals are arround 40-60 ohm resistance in the model. But on the internet you find models with R=10 and so on which is too low.

Also remember that to build up simulation can take hundreds of us so try to zoom in a lot.

 

[newbie_1]

help! crystal oscillator

thanks teddy,
i have set conervative in spectre, but failed,
and i tried hspice,
still fail,
and mine xtal is 32k and the the resistance is 55k, a very large value.

 

[holddreams]

Re: help! crystal oscillator

Give VDD a pwl stimulus,and run a .tran analysis,
It will take a long time to see the final result,maybe several days.
You can print the osco plot,and check whether its magnitude is decreasing .If decreasing(maybe very slow),then it will oscillate till the end.

 

[newbie_1]

help! crystal oscillator

thanks holddreams,
i have done that, but failed,
and the osco should be increasing when oscillating, right?

 

[holddreams]

Re: help! crystal oscillator

Give L an initial current ,and check whether it can oscillate.This is not a good way.
Just try .

 

[newbie_1]

Re: help! crystal oscillator

i have tried to give an initial current to L,
and the amplitude of oscillation will attenuate, so still failed
any other suggestions?

 

[John Xu]

Re: help! crystal oscillator

Hi,
Startup a 32K xtal oscillator will largely depend on the capability of simulator, xtal model. Try the below options to kick off your 32K osc:

chgtol=1e-18
reltol=1e-6
kclref=min
method = trapezoidal
maxord=2

sometimes, method=bdf will take effect.

This is the options I used before for my xtal 32K osc simulation. The simulator I used is the internal ones for our company, which have special algorithm to deal with it. So I can not guarrantee the options can be applied to your simulator.

Can you send me the model you used, then I can try it in our side.

 

[Teddy]

Re: help! crystal oscillator

Well as much as I agree that based on calculations the xtal should oscillate at about 32.8kHz I can't imagine that you can have xtal with Ls=11.233kH
It is a decent coil there. But I am not saying I know everything.

The issue there is that in general you have Pierce oscillator "gm stage"
and you need gain. In this case there is nearly no gain at all and it is not able to drive enough current to such a huge L.

I am quite sure that in case you increase the inverter size to hundreds (Wp=400u Wn=200u) it will start. Also add on one terminal little current pulse as a "noise" source. I use cca 2mA for 100ps and then 0

Have fun

 

[arsenal]

help! crystal oscillator

i think the cry will never oscillate if width is increased too much,
and since the gain and phase have been verified, then maybe sth has to be checked in the setup of the simulation environment

 

[newbie_1]

help! crystal oscillator

thank you all, anywy, it still cannot start up

 

[alok_ky]

help! crystal oscillator

what the negative resistance fo your amplifier ?

Try reducing the Gm of the amp.

Added after 51 seconds:

Not only reducing but try to find out the optimum vakue of Gm to have sufficient negative resistance.

 

[Eric Best]

Re: help! crystal oscillator

...
xi0 xi xi xtal
...

I haven't tried the simulation but in my opinion the problem is in the above line of your netlist - your crystal is shorted!
Both terminals are connected to one node xi: xi0 xi xi xtal

Best Regards
Eric

 

[holddreams]

Re: help! crystal oscillator

i have setup a crystal oscillator using inverter
and the netlist is shown as below:

m0 xo xi vdd vdd pch_33 l=0.5u w=2u m=1
m1 xo xi 0 0 nch_33 l=0.5u w=1u m=1
c1 xi 0 16p
c2 xo 0 16p
rf xo xi 1meg
v0 vdd 0 pwl 0 0 100u 3
xi0 xi xi xtal

.subckt xtal n1 n2
rs n1 n5 55k
ls n5 n7 11.233e3 ic=1e-7
cs n2 n7 2.1f
c2 n2 n1 1.1p
.ends xtal



and i have simulated the open loop gain and it's about 16dB, and the gm at steady state is about 500uS(<gmmax and >gmcrit), however, after powerup, the oscillator will enter the steady state(xo and xi remains at about 1.45) and never oscillate. （ and i have used option accurate and method=gear, and transient step is 1ns)
So anyone help me!!Thanks a lot

Where did you get this crystal model equivalent parameters?
55k seems to be too large.

 

[Eric Best]

Re: help! crystal oscillator

CAN'T YOU READ???

The circuit, as it is, WILL NEVER OSCILLATE!

The crystal IS SHORTED!!!

The line in the netlist, as I mentioned above, MUST BE:

xi0 xi xo xtal

Note: this crystal has Q ≈ 42050 @ (series) resonant frequency ≈ 32.768kHz

Eric