RF Simulation Time

I have been simulating the IP3 response of the buffer circuit I am designing lately. I noticed that it is very long to simulate the IP3 using the transient and DFT method. Here I list the time to simulate:



Using the QPSS method is far faster than using transient and DFT method. I can save about 13 minutes. Utilizing the QPSS means I am using an RF license though.

RF for the First Time

I met my soon to be boss two days ago, that was March 9, 2011. He is a Japanese engineer working in our company for almost 16 years already and he is now the department's manager. Early next month, April, he will join our division as our General manager. This young brilliant man is only 38 years old and have been to many countries like China, Singapore, USA, and Philippines to develop cutting edge technologies and innovative products. He is responsible in creating RF designs used in RF applications such as audio portable devices.

Having been in circuit designing for 8 years, developing multi-purpose Integrated Circuits like linear regulator, motor driver, operational amplifier, dc-dc converter, and LED driver Ics, it would be my first time to be involved in RF design.

Since last week, I was asked to develop a super buffer. So I shall be getting good IP3 and 1dB compression from my design. I have two-tone signal in my input, f1=98MHz and f2=100MHz. From the equation (f1*2)-f2 and (f2*2)-f1, I shall have 96MHz and 102MHz as my 3rd order frequencies.

Simulating the circuit, I can get the 3rd order frequencies, which is shown in figure below:


To get the IP3 of the circuit, line is extended to the 3rd order graph through the 1st order line. Their intersection is the IP3, which is shown in the figure below:



To verify the data, I also get the transient response at the emitter and collector of the transistor. The transient response of the collector and emitter of the transistor is not distorted until input signal is 111dBuV.

Quasi-Periodic Steady State Analysis

I have learned that there are many ways to get the IP3 (input referred 3rd order intercept point) such as using the transient and DFT method, and the QPSS or quasi-periodic steady state analysis. It's my first time to use; I found it easy and interesting though.

How to use the QPSS? Using Virtuoso Analog Design Environment, here are the things to be set up:

1. First, make a test circuit creating 2 voltage source with frequency 1 and frequency 2 for the positive and negative input of the amplifier. For my analysis, I've used f1=98MHz and f2=100MHz. Of course, I should have the same amplitude for my 2 signals.

2. Engine setting is shooting.

3. Conservative is used for accuracy.

4. Pre-determined Sweep Range for the input signal. For example from 60dBuV to 120dBuV.

5. For the sweep type, I usually use the linear type with step size of 2.

Saving the netlist and running the program results to the following graph:

image credit by silvaco

Periodic AC Analysis

I am simulating the same circuit and getting its IP3 characteristic. This time Periodic AC analysis (PAC) and Periodic Steady State analysis (PSS) have been utilized.

Unlike in QPSS (quasi-periodic steady state) analysis, one tone is used. It means that one frequency will be applied in the input signal, e.g. 98MHz. Thus the positive (+) and the negative (-) inputs will have the same frequency and magnitude but the other input will have 180 degrees phase shift.

Source type is sine wave (voltage source) and the small signal parameter is employed to input the value for the PAC magnitude and PAC phase as well.

PSS analysis set-up:
1. The name of the frequency F1 with a value of let's say 98MHz is a large signal.
2. Ticking the auto calculate will give you a beat frequency of 98MHz.
3. For the output harmonics, you can select from range, let's say from 1MHz to 1GHz; and make 10 orders perhaps.
4. Select the desired signal frequency (98MHz).
5. For accuracy, tick conservative.
6. The input signal should be variable so you can sweep a range, maybe from 60dBuV to 120dBuV.
7. For the sweep type, I usually use the linear type with step size of 5.
8. Furthermore, the integration method parameter should be gear2only. This is optimized in the PSS Options section.

PAC analysis set-up:
1. Since 98MHz is used in the PSS analysis, then the Beat frequency is 98MHz.
2. Sweep is Absolute.
3. Single Point frequency is 100MHz.
4. For the sidebands, you can select from range, from 1MHz to 1GHz perhaps.
5. Select index frequency and some order frequencies like 96MHz and 100MHz.