Designer’s Guide Community

rajeee1000

Junior Member

Offline

Posts: 15
Bangalore

crystal oscillator simulation in cadence
Jun 06^th, 2007, 3:34am

Dear,

I am designing a 32 KHz crystal oscillator in cadence. I see the problem of the oscillation amplitude depending significantly on the initial condition (inductor current). Is there a way to faithfully estimate the oscillation amplitude or rather to choose the correct initial condition?

Thanks in advance
Rajesh

PS: I don't have access to any harmonic balance simulator

Back to top

IP Logged

ywguo

Community Fellow

Offline

Posts: 943
Shanghai, PRC

Re: crystal oscillator simulation in cadence
Reply #1 - Jun 8^th, 2007, 1:59am

Hi,

I am not aware that the amplitude depends on the initial condition, like the initial current in inductor. However, the amplitude depends on the integration method strongly. Generally, I get smaller amplitude when I choose method = gear than that when I choose method = trap.

However, I have not compared the simulation result with the test results.

Best regards,
Yawei

Back to top

IP Logged

jeffyan

Community Member

Offline

Posts: 47

Re: crystal oscillator simulation in cadence
Reply #2 - Jun 14^th, 2007, 6:06pm

rajeee1000 wrote on Jun 6^th, 2007, 3:34am:

Dear,

I am designing a 32 KHz crystal oscillator in cadence. I see the problem of the oscillation amplitude depending significantly on the initial condition (inductor current). Is there a way to faithfully estimate the oscillation amplitude or rather to choose the correct initial condition?

Thanks in advance
Rajesh

PS: I don't have access to any harmonic balance simulator �

hi
yes, there are some methods to calculate the amplitude theoretically in papers by qiuting huang, and vittoz.

Back to top

IP Logged

vivkr

Community Fellow

Offline

Posts: 780

Re: crystal oscillator simulation in cadence
Reply #3 - Jun 15^th, 2007, 7:18am

Hi Yawei,

That's probably because gear maps some poles on the jw axis to the LHP => unstable systems may simulate
as stable if gear is set. For these cases, trap is better as it provides the best mapping.

However, with gear, you can also achieve the same results with tighter tolerances (atleast that's what the manuals say).

I would say that your measurements should probably match the trap simulation.

Hi Rajesh,

There are several papers by Vittoz and also by RG Meyer (of Gray & Meyer fame), all from the 70s or 80s from IEEE
JSSC. Try searching for crystal oscillators in the search section there

Regards
Vivek

ywguo wrote on Jun 8^th, 2007, 1:59am:

Hi,

I am not aware that the amplitude depends on the initial condition, like the initial current in inductor. However, the amplitude depends on the integration method strongly. Generally, I get smaller amplitude when I choose method = gear than that when I choose method = trap.

However, I have not compared the simulation result with the test results.

Best regards,
Yawei

Back to top

IP Logged