IIP3 and NF measurements in cadence spectre

[jimito13]

Hello all,

I want to measure IIP3 and Noise Figure with cadence spectre.My application is a fully differential Low-Pass filter.Could somebody guide me with details how to setup the simulations?If possible i would appreciate a screenshot of a similar testbench for filter application.

At the input of my testbench i use a port (port component from analogLib) with a balun (to convert the signal to fully differential) with two tones (f1 & f2).I select f1 to be at the center of my passband as a colleague suggested me,is it correct or i should choose f1 as the corner freq. of my filter?

At the output of my filter i have connected as load a fully differential amplifier (that is the stage that connects next to me in our project).For the measurements of the 2 quantities i want i suppose that load should be present,right?Do i need to connect a port at the output (that is the output of the amplifier that is my load) as well for the simulation purposes?

The impedance of the input port is set to 500Ω (the impedance of the previous stage) and the power is given to me to be sweeped from -40dBm to -10dBm with 2dBm step.If output port has to be connected at the output of my load what impedance should that have?

I am looking forward to a helpful and detailed help to my issue.

Thanks in advance,

Jimito13

 

[Milad-D]

**broken link removed**

 

[jimito13]

I already have this document,thanks although.

Could somebody answer my questions on the first post separately?Even with showing a pesonal working testbench with a snapshot (schematic with the ports locations for a filter and the simulation settings) so that i will be 100% sure that my results and measurements are correct.

I tried a qpss analysis with 2 tones sweeped prf from -40dBm to -10dBm and the 2 curves do not cross.....What this result means?

 

[Milad-D]

First, the efficient way for iip3 simulation is pss+pac. About your circuit, are you measuring out of band iip3 or in band? In other words, the f1 and f2 are in band or out of band?

It is good to have a look on the ports and qpss properties window of your test bench plus the result graph.

They do not cross. you are talking about the curves or the extrapolated lines? if the iip3 of your circuit is high, then the extrapolated lines will not cross till -10 dbm.

As you know, when plotting these two tones, by the method described in document, the simulator will give you the value of iip3. What value, do you get?

 

[jimito13]

Milan-D,

I am interested in measuring in-band IIP3.As i already said in my first post i select f1 to be at the center of my passband as a colleague suggested me,is it correct or i should choose f1 as the corner freq. of my filter?So f1 and f2 are in the band of my application and not out of it.

My testbench schematic has the following form as chain :

INPUT PORT --> IDEAL BALUN --> FULLY DIFFERENTIAL LOW PASS FILTER --> FULLY DIFFERENTIAL AMPLIFIER (that is the load of my filter) --> OUTPUTS

I should note that at the OUTPUTS i do not use a port...should i insert such an instance over there as well?

At the port properties i set :

Port Resistance : 500Ohm (that is the output impedance of the previous stage of the filter that is being designed by someone else)

Frequency Name 1 : frf
Frequency 1 : frf
Amplitude 1 : prf

Frequency Name 2 : frf2
Frequency 2 : frf+1M
Amplitude 2 : prf

I have no more settings under the port instanse.

For the QPSS setup i will show you a snapshot with the settings and the result i came up with.

I am looking forward to your help,it is very precious to me because those measurements are critical for my chip and i want to be absolutely sure that the result of the simulations is the correct one.If you or somebody else can give me help for NF measurement setup it would be great.

Thanks in advance.