Practical RF Circuit Designing

Anurag Bhargava Application Consultant Agilent EEsof EDA Agilent Technologies

Email: [email protected] You Tube: www.youtube.com/user/BhargavaAnurag Blog: http://abhargava.wordpress.com

Agenda What is RF? Approaches to RF circuit designs Practical considerations Successful methodology for RF circuit designs

Agenda What is RF? Approaches to RF circuit designs Practical considerations Successful methodology for RF circuit designs

What is RF?
Radio frequency (RF) is a rate of oscillation in the range of about 3 kHz to 300 GHz, which corresponds to the frequency of radio waves, and the alternating currents which carry radio signals. Electric currents that oscillate at radio frequencies have special properties not shared by direct current or alternating current of lower frequencies. The energy in an RF current can radiate off a conductor into space as electromagnetic waves (radio waves), this is the basis of radio technology. RF current does not penetrate deeply into electrical conductors but flows along their surfaces; this is known as the skin effect.

Scope of current discussion

RF Circuit design is very wide topic and we shall limit ourselves to discuss about Lumped/Discrete and Distributed circuit design aspects in this presentation.

Techniques which are discussed in this Webinar can be applied to any type of RF/uWave circuits i.e Active or Passive.

All the simulations have been carried out using Agilent ADS (Advanced Design System) software.

Agilent EEsof EDA Solutions

ADS
MMIC, RF Board, SiP, Printed Antenna, Arrays

ADS
Signal Integrity

Genesys
RF Board

SystemVue
ESL

Golden Gate
RFIC

EMPro
3D EM (FEM & FDTD)

ICIC-CAP
Device Modeling

Agilent ADS for MMIC, SiP and RF Board Co-Design

ADS Main Features
1. 2. 3. 4. 5. 6. Complete Front Front-to to-Back RF Board Designs Complete Front Front-to to-Back MMIC Designs Best Best-in in-class simulators for Active and Passive Circuit/System Designs Superior capabilities for High Speed Signal Integrity Analysis Links to 3rd party tools like Cadence, Mentor Graphics, Zuken etc Direct Links to Agilent EMPro for full 3D EM analysis and Agilent SystemVue for RFRF-Digital system design & Verification Widest support for Vendor Component Models Biggest market share in RF/ RF/uWave uWave Design segment

7. 8.

Day in Life of a RF Designer..few examples!!

Example 1: Low Pass Filter Design

Designed Measured

Day in Life of a RF Designer..few examples!!

Example 2: MMIC Amplifier Design

Day in Life of a RF Designer..few examples!!

Example 3: MIC LNA Design

NF Spec: 0.7 dB
FET

Measured:

8 dB

Day in Life of a RF Designer..

Some familiar statements (Do you hear them often?).

RF/uWave is black magic.

i.e. you dont know why things work and why they dont

RF/uWave is too tough to handle

i.e. I am better off in some other field

No matter how much work you do in software, you still need to take out
screw driver (just an analogy..!!!) and fix it on the measurement table to make RF circuits work.
i.e. Dont waste your time using RF design software

Design software is really bad, it showed me good response but

measurement results are really bad, not sure what to do.
i.e. blame it on the software

Day in Life of a RF Designer.. Lets think

Did you account for all things which shall affect circuit performance during the design phase? i.e. Did you. Account for parasitic behavior of Inductors, Capacitors etc at RF? Account for component tolerances and its effect on your circuit performance during your design phase? (Our focus today!!) Account for radiation, parasitic coupling effects of transmission lines at higher frequencies? Account for Etching tolerances in fabricating your PCBs/MICs? Account for Parasitic effects of transmission lines at higher frequencies?

Simulated your virtual design as close as to the real life assembly, accounting for most of the things which might make difference in the performance?

Most often you will find that we missed something during the design phase, lets try to understand these things better so that we have better predictability in our RF circuit performances after fabrication

Agenda What is RF? Approaches to RF circuit designs Practical considerations Successful methodology for RF circuit designs

Approaches to RF Circuit Design

RF Circuit Implementation Techniques There are various implementation techniques for RF circuits e.g: a. b. c. Lumped / Discrete components based circuits Distributed Lines (e.g. Microstrip, Stripline etc) Waveguide based structures

Approaches to RF Circuit Design

How to select implementation technique for RF Circuits?
Implementation technique for a RF circuit can be selected by considering various factors:

a. Frequency of Operation:
a. b. c. Circuits under 500MHz are usually designed using Lumped / Discrete components Circuits between 1 GHz 30 GHz can be designed using distributed lines such as Microstrip etc Circuits over 30 GHz are usually designed using waveguide based components

b. Power Handing Capacity:

Another important aspect of RF circuit design is power handing or operating power of the circuit which may call for hybrid implementation technology.

c. Q-factor requirements (usually referred for filters)

Sometimes high-Q structures (e.g. Sharp rejection) can be implemented using waveguide based techniques even at lower frequencies e.g. Cavity Filters

Agenda What is RF? Approaches to RF circuit designs Practical considerations Successful methodology for RF circuit designs

Practical Considerations for Discrete Elements

What could make difference in real world? Component Q: Also known as Unloaded Q which introduces loss in circuit HF Parasitics: Each component has some HF equivalent circuit which could affect circuit performance (see next slide) Layout parasitics: These parasitics could play a role in changing circuit performance and they originate due to dispersive nature of transmission lines which are used as a connecting medium for layout Component Tolerance: All the discrete components will have some inherent tolerance associated and it is not practically possible to produce a discrete component without any tolerance and actual component value can vary within their specified tolerance range.

HF equivalent circuit of Inductor and Capacitor

Cmain -> Main Capacitance Lseries -> Series Inductance caused due to bonding at both side of Capacitor internal assembly, causes Series resonance (SRF) Rseries -> Equivalent Series Resistance Cpar -> Capacitance caused due to parasitics between capacitor plates and assembly metals in SMT comps., causes parallel resonance (PRF) Rpar -> Parasitic parallel resistance

Lmain -> Main Inductance Rseries -> Equivalent Series Resistance (ESR) Cpar -> Capacitance caused due to parasitics between Inductive coils and assembly metals in SMT comps. Rpar -> Parasitic parallel resistance, typically varies with frequency of operation

Guidelines for selection of Inductor and Capacitor

Before selecting any component from Inductor and Library it is always preferable to consult manufacturer manual to note their Q and SRF specifications. Ideally Frequency of operation should be <= 20% - 30% of SRF of each component, which poses the upper limit on the Discrete component values which can be used for circuit design. Meeting these SRF guidelines ensures some kind of flexibility in design process so that the parasitic effect dont dominate the natural response of the components, resulting in more tolerant design to account for component value variation due to manufacturing limitations. If it is not possible to restrict component values as per SRF guidelines then it is preferable to choose discrete components with tighter tolerance levels as far as possible to ensure yield variations are kept to minimum and components value do not vary much due to various reasons and this will help in producing high yield manufacturing circuits.

Discrete Components Tolerance Reference Table Tolerance Level B C D F G J K M Value +/- 0.1 +/- 0.2 +/- 0.5 +/- 1 % +/- 2 % +/- 5 % +/- 10 % +/- 20 % Comments Absolute Value Absolute Value Absolute Value -------------------------------

Distributed Circuits
At sufficiently high frequencies it is not possible to design circuits using discrete components as their physical size become appreciable fraction of wavelength All the components used for circuit design needs to be realized using printed techniques, still the basis of all circuits would be L, C, R but realized using printed techniques Designers have option to select printed transmission line technology of his choice, few of them are highlighted on the next slide. Inductor can be realized with High Impedance (Narrow width) line, Capacitor can be realized with Low impedance (Wider width) line, resonators can be realized using coupled transmission lines.

RF/uWave Printed Transmission Lines

Microstrip Lines: Easy to fabricate, Easy tuning after fabrication, Easy mechanical assembly, Relatively Low Q, Quasi TEM mode

Striplines: Complex assembly, No Tuning, Higher Q than microstrip lines, TEM mode

Suspended Stripline: Complex assembly, Mechanical Intensive, High Q , No Tuning

Picture source: www.microwaves101.com

Microstrip Lines
Microstrip lines have been extensively used for RF/Microwave circuit design over the years and they are still preferred in most of the cases because of their flexibility Substrate selection is an important criteria in designing circuits based on Microstrip technology Electric field needs to be confined in the substrate for proper circuit operation else it will face radiation problem (Antenna is reverse where we need only radiation) Lot of references available on theory of Microstrip lines, one of the most popular being the book: Foundations for Microstrip Circuit Design T.C. Edwards

Selecting Substrate for Microstrip Designs

There are few key parameters which helps in deciding right substrate for the RF/Microwave designs: a. Substrate Height: This needs to be selected as per the frequency of operation, it can cause Surface wave radiation. Higher the frequency of operation lower should be the height of substrate b. Relative Permittivity / Dielectric Constant: Usually referred as dielectric constant (Er) is the measure of opposition placed by Dielectric substrate towards any electric field. c. Loss Tangent: It is the quality measure of dielectric purity (depicts lossy nature of dielectric). Lowest possible value is always desirable to ensure that dielectric is lossless. Represents the Q factor of the dielectric. Lesser the better d. Conductor Material: Conductivity of the conductor used in Microstrip design contribute to the transmission losses. (e.g. Gold: 4.1E7, Copper: 5.8E7, Silver: 6.1E7, all units in Siemens/meter)

** Substrate performance can be controlled/optimized with proper selection of Height and Er

Practical Considerations for Distributed Circuit Designs What could make difference in real world?
Boundary conditions: All the printed line components are modeled using Analytic equations in schematic/ circuit design environment which may not always represent true behavior of transmission lines and they can be truly characterized in EM (Electromagnetic) domain using EM solvers. Parasitic Cross Couplings: All the transmission line structure can observe cross coupling between adjacent & non-adjacent sections and it is almost impossible to take care of these parasitic coupling by any of the circuit simulators as all the lines/sections drawn in schematic are independent of each other. Radiation Problems: Circuit simulator will have limitations on accurate prediction of the radiation problems which might occur due to the circuit layout size (Space wave radiations) or because of Substrate properties (Surface wave radiations).

Does it mean circuit simulators are useless for Distributed Circuit Designs?

Electromagnetic vs. Circuit Simulations

EM simulators
Analyzes physical structure based on EM theory (Maxwells Equations) Considers all coupling Passive components only Generally slower simulations Optimization is not easy

Circuit Simulators
Analyzes circuit schematic using built-in models Considers explicit coupling Passive & Active components Generally faster simulations Optimization is very easy Designs can be tuned in real time
L= 10 mil W= 10 mil L= 15 mil W= 10 mil L= 15 mil W= 10 mil 3 L= 120 mil W= 10 mil S= 15 mil L= 15 mil W= 10 mil L= 10 mil W= 10 mil L= 10 mil W= 10 mil 4

L= 10 mil W= 10 mil 1

It is recommended to start with Circuit Simulators and later turn to EM for more accurate analysis. ADS offers hybrid simulation to combine Circuit and EM analysis.

Agenda What is RF? Approaches to RF circuit designs Practical considerations Successful methodology for RF circuit designs

Recommended Lumped / Discrete Circuit Design Flow

Step 1:

Start with circuit design obtained by manual calculations /ADS Designguide / Genesys Synthesis Replace ideal components with actual vendor Models or use Qfactor or Spice Models or S2P file Statistical (Yield) Analysis Prepare Layout

Run Optimization Run Optimization

Yield Optimization

Step 2:

Step 3:

Step 4:

EM/Ckt cosimulation

Step 5:

Generate Gerber / DXF etc

Recommended Distributed Circuit Design Flow

Step 1:

Start with circuit design obtained by manual calculations /ADS Designguide / Genesys Synthesis Prepare Layout and run EM simulations (Mandatory for accurate prediction at high freq) Statistical (Yield) Analysis Prepare Final Layout Generate Gerber / DXF etc

Run Circuit Optimization

Run EM Optimization (if possible)

Step 2:

Step 3: Step 4: Step 5:

Yield Optimization

Note: Type of EM simulation technology can be selected as per the circuit under design. Designers usually have choice of using Method of Moments (MoM), Finite Element Method (FEM) or Finite Difference Time Domain (FDTD), each technology has their own advantages and disadvantages.

Step 1: Initial Circuit Design

We start our circuit design by calculating element values using equations. Following circuit shows the circuit component values obtained after optimization to meet the desired S11 and S21 specifications

Step 2: Circuit Design with Q-factor included

As a next step we change the ideal L & C components to the models with Q-factor so that we can take lossy nature of L & C into account We can see that in this case there is no appreciable change in circuit performance with inclusion of Q-factor so we are good to proceed to next step

Step 3: Yield Analysis

Yield Analysis is a 3 step process: Set the Goals / Performance Criteria to be met by the circuit Set the tolerance in the components Set Yield Controller for desired number of iterations

Inductors are having +/- 10% tolerance and Capacitors are having +/- 5% tolerance

Step 3: Yield Analysis

Yield analysis shows that this circuit has a poor yield of 45% that means only 45% of the circuits fabricated will meet the desired specifications and rest 55% will fail. From the simulation results we can observe that S21 rejection criteria (>40dB) and S11 (<-15dB) is getting worse due to component tolerances hence causing lower Yield.

Whats wrong with the circuit..??

Questions: a. Why circuit yield is so bad? b. Which component or components in my design are creating problems? c. How to correct the problem and improve the yield?

Sensitivity Histograms in Agilent ADS S/W

Sensitivity Histograms in ADS are very useful to find out how each component is making a effect on a specific Yield Goal or specification. We can place Sensitivity Histogram measurement functions/equations from Yield library palette in ADS as shown below:

Equation shown here is for checking C1s tolerance effect on S21s rejection goal from 250MHz 500 MHz Similarly, we can place these measurement functions for all different type of specifications we have in our design e.g. S11, S21 etc Remember these equations apply equally for active circuits performance criteria to check Output Power, Noise Figure, Power Added Efficiency etc

Sensitivity Histogram Result for S21 response

How to read histogram results: Y-axis is Yield % and X-axis is the component value with its +/- tolerance range If we closely look at these plots we can notice that which component is affecting performance metrics for S21 From the graphs we can conclude that component values will need to re-adjusted to improve the Yield performance..

Initial Value C1-> 43pF C2-> 43pF L1-> 75nH L2-> 139nH L3-> 70nH

Modified Value 48pF 48pF 83nH 174nH 88nH

S21 Performance Correction

As suggested by Histograms, components values were readjusted and after Yield Analysis we find that circuit performance has improved and we are able to achieve compliance for S21 (rejection criteria) of >40dB in 250-500 MHz band so remaining problem can be thought of coming from S11 failures.

Still S11 results seems to be giving problem

S11 Sensitivity Histograms

By including Measurement functions for S11 histograms, we can notice that L1 and L2 inductors are still causing S11 performance to degrade and we will need to reduce the tolerance of these 2 components. Change the tolerance of L1 and L2 inductors to 5%

Yield Optimization in ADS

Now as we fixed one specification and reduce tolerance of few components to take care of the other specification, we can use Yield Optimization to centre our components values in one direct way to achieve design centering for various performance criteria together in a more reliable manner . Yield Optimization is different than normal circuit optimization as it centers the components values using Statistical Method and we should have optimized circuit before we run this kind of optimization. Also, remember that Yield Optimization can take longer time because it performs Yield Analysis at each iteration to check the circuit Yield and centers the design performance based on the tolerance of the components

Steps for Yield Optimization

1. Define the component values to be Optimizable: So that ADS has the right to change the values. We can always set the min and max of the values under which ADS changes the component values Set the Yield Optimization Controller and number of iterations

2.

Circuit Yield after Yield Optimization

We can see that RF circuits Yield improves dramatically after our component value modifications and Yield Optimization. We can now proceed for Layout generation and perform EM-Circuit cosimulation for checking if Layout parasitics will affect our circuit performance

Step 4: EM-Circuit Cosimulation (Virtual PCB Assembly!!)

EM Circuit cosimulation is like performing virtual PCB assembly whereas ADS will simulate PCB layout using EM simulator (Momentum i.e. Method of Moments or Finite Element Method (FEM) based on the selection and then hybrid simulation with Circuit simulator. It allows designers to take care of the parasitic effect which may be caused by interconnecting PCB lines etc on top of usual electrical component designs. We can include transistors etc as well.

EM-Circuit cosimulation results

EM-Circuit cosimulation result here show very close matching with our final design hence our circuit is ready for fabrication.

Example: Yield Analysis on GaN Power Amp

As we stated before, technique which we learn can be applied on all type of RF circuits i.e. Active or Passive e.g. Power Amplifier, Mixer, Oscillators etc.. Here is one example of GaN Power Amplifier where widths of the transmission lines have etching tolerance of +/-35um based on the printing technology used and we can observe the variation in PAs Output Power and Power Added Efficiency (PAE) as shown below.

What we learnt today..?

Successful RF circuit design is the result of careful design process Nothing can be taken for granted..simulate when in doubt !! Agilent ADS software provides all the key capabilities to make you successful RF circuit designer Simulation tools are usually as good as designers (barring their own inherent limitations...) Software tools are different in their capabilities, accuracy and algorithm implementations so designers need to make the choice according to what they need. Agilent ADS is a mature RF design software with many years of unmatched innovations in technology, features etc and it provides best in class technology for RF designers to achieve 1st pass design success

Looking Ahead..!!
Kindly send your requests on what we can cover in our future webinars to: Mukul Pareek Marketing Engineer, Agilent Technologies Email: [email protected]

Resources to help you...

Subscribe to the Blog: http://abhargava.wordpress.com to receive latest news, tips & tricks on RF design and industry news.. Subscribe to the You Tube Channel: www.youtube.com/user/BhargavaAnurag to see various ADS tutorial videos and learn the ADS concepts better Agilent EEsof Home Page: www.eesof.com Agilent ADS software evaluation license or for any other queries:

E-mail: [email protected] Toll Free Number: 1800-11-2929