Designer’s Guide Community

‹ Previous topic | No topic ›

Pages: 1

IIP3 Simulations using Spectre RF and Transient Me (Read 5766 times)

Frank Hayden

Guest

IIP3 Simulations using Spectre RF and Transient Me
Apr 30^th, 2003, 2:12pm

Hello There;

I am wondering why I get a 10 dB difference when I simulate IIP3 using the method that was published on this site ( Accurate and Rapid Measurement of IP2 and IP3, version 1b, 22 May 2002) and the method of two tone IIP3 simulation using transient analysis. I trust the transient analysis approach more than the PAC/PSS method, as this is the way I would measure it.

Has anyone had any experience with either of these methods, and can you make any rcommendations?

Thank you in advance.

Frank Hayden

IP Logged

Thomas_Hansson

New Member

Offline

Posts: 4

Re: IIP3 Simulations using Spectre RF and Transien
Reply #1 - May 2^nd, 2003, 12:22am

Hello
My experience is that these two methods should give about the same results. Just som comments:

* I have senn sometimes the built in functions in fuctiousn using direct plot, having wrong names on the axes. Could somtimes cause me to use input as ouput or vice versa.
* PSS and PAC is a small signal analysis only valid upto about 10dB less than 1dB-compression point. This means you have to check this first. Still useful for checking mixers. For better possibilities I would suggest you to use QPSS (Pdisto). This is much better and the best way I know to simulate IP3 for amplifiers. IT's however very important to have the recommended accuracy settings and other settings. QPSS can handle "moderate signals" meaning more nonlinear.

* It's important to sweep from a low value (where it's linear). Check different starting point to see that this is not affecting the result.

Thomas

IP Logged

Frank Hayden Guest	Re: IIP3 Simulations using Spectre RF and Transien Reply #2 - May 2^nd, 2003, 11:05am Thomas; Thanks for your reply. Since making this posting I have found that I must set the errpreset option in the PSS analysis form to "moderate". Failing to do so causes an error in the analysis. I am running a test case on both methods at the moment. Again, thanks for your reply. Frank Hayden
Back to top	IP Logged

Pages: 1

‹ Previous topic | No topic ›