The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> How can I accelerate the PSS simulation of VCO?
https://designers-guide.org/forum/YaBB.pl?num=1309103994

Message started by Patrick520 on Jun 26^th, 2011, 8:59am

Title: How can I accelerate the PSS simulation of VCO?
Post by Patrick520 on Jun 26^th, 2011, 8:59am

Hi
I am design a vco for zigbee receiver, whose oscillating frequency is 2.4GHz. And the torpology is LC oscillator.
In Spectre ADE, I choose the PSS (shooting) simulation.
But I find that simulation speed is too slow.

The simulating netlist presents as:
pss: time=13.19us (3.23 m%), step=1.839ps ( 200 p%)
pss: time=13.26us (3.25 m%), step=2.38 ps (400 p%)
.......

My question is how can I accelerate the speed of PSS simulation?
The step is too slow like only 1.839ps.

How can I set the PSS parameter?

Title: Re: How can I accelerate the PSS simulation of VCO?
Post by Ken Kundert on Jun 26^th, 2011, 1:47pm

You did not give much information, so it is hard to give any specific suggestions. Given what you have said, I recommend:
1. First set all tolerances to their default values. Many people get themselves into trouble tightening tolerances and accuracy settings before they know they need them. In the best case it results in unnecessarily slow simulations, and it the worst case it prevents the simulation from completing.
2. Try setting the integration method (transient option 'method') to gear2only. Occasionally trapezoidal rule can generate excessive ringing that causes the simulation to run slowly.
3. Try to determine what is going on at 13.19us to 13.26us that is causing the simulator to run slowly.

-Ken

Title: Re: How can I accelerate the PSS simulation of VCO?
Post by rfidea on Jun 27^th, 2011, 1:17pm

Is 13.19 and 13.26us part of the tstab or fundamental oscillation search. It looks like the oscillation search from the log. It looks like you have used a very long tstab since your oscillation frequency ic 2.4GHz, which is a period time of 0.417ns. Have you checked that your oscillator is starting in either the tstab or in a pure transient simulation? It does not need to seattle but it must start to oscillate.

Title: Re: How can I accelerate the PSS simulation of VCO?
Post by neoflash on Jun 30^th, 2011, 11:36am

I would recommend you first run a transient simulation and make sure every node looks normal.

And also clear out all warnings in your transient log file. Sometimes, the floating nodes give huge undebuggable trouble.

Title: Re: How can I accelerate the PSS simulation of VCO?
Post by sheldon on Oct 9^th, 2011, 1:33am

Patrick,

Is there any reason you need to use Shooting Newton PSS analysis?
The HB engine has also matured and is another option that may
provide better performance. This will speed up the PSS portion of the
analysis. Since it seems like you have an LC oscillator without a
divider, you might also want to try the oscinit=linear option. When
using the oscinit=linear, transient is not used for the either the initial
conditions or for the PSS analysis. Don't forget to set tstab=0.

Best Regards,

Sheldon

Title: Re: How can I accelerate the PSS simulation of VCO?
Post by loose-electron on Oct 29^th, 2011, 5:59pm

http://www.amazon.com/Designers-Guide-SPICE-Spectre%C2%AE-Book/dp/0792395719/ref=sr_1_1?ie=UTF8&qid=1319936275&sr=8-1

and, you need the above book on your shelf!

:D