15
The Impact of Electro- Thermal Coupling on RF Power Amplifiers Matthew Ozalas Keysight Technologies SANTA ROSA, CA

The Impact of Electro- Thermal Coupling on RF Power Amplifiers

Embed Size (px)

Citation preview

Page 1: The Impact of Electro- Thermal Coupling on RF Power Amplifiers

The Impact of Electro-Thermal Coupling on RF Power Amplifiers

Matthew Ozalas Keysight Technologies SANTA ROSA, CA

Page 2: The Impact of Electro- Thermal Coupling on RF Power Amplifiers

Self Heating Mutual Heating Transient Heating

Review of Heating in Active Devices:

(+ different approaches to modeling these)

Page 3: The Impact of Electro- Thermal Coupling on RF Power Amplifiers

Device Heating Basics:

PdissRTT thambientdevice *+=

⎟⎟⎠

⎞⎜⎜⎝

⎛++

−=

)2()2(ln

)(21

sub

sub

tLWtWL

WLRth

κ

⎥⎥⎥

⎢⎢⎢

⎡=

333231

232221

131211

RRRRRRRRR

Rij

Self Heating Terms

j

iij P

TRΔΔ=

1 2 3

Self Heating

Mutual Heating

Substrate Spreading Term (for rectangular geometries L>W)

Thermal Conductivity

Page 4: The Impact of Electro- Thermal Coupling on RF Power Amplifiers

Transient Thermal Response Keysight GaAs HBT Test Device (Probe) Transient pulse applied to Vcollector

Resulting Thermal Response in Ic

(Log Scale for time)

Fall Time ~ 0.1 us

(Current drop shown here is due to dev. cooling)

Page 5: The Impact of Electro- Thermal Coupling on RF Power Amplifiers

Lumped Modeling Approach

ththth CR=τ Intrinsic Device temp tracks RF power burst well

Thermal Network is modeled as a RC Network(s) for fast pulses

Intrinsic Device ~ns to us

Die or Wafer

Package

~ms

~s

Die + Device temp delayed in responding

Page 6: The Impact of Electro- Thermal Coupling on RF Power Amplifiers

Iteration loop is done automatically

until powers and temperatures are

self-consistent

Circuit Simulator Read temperatures

Solve electrical equations Write power dissipation

Thermal Simulator Read power dissipation Solve thermal equation

Write temperatures PDISS

TDEVICES

Thermal technology

files

Electro-thermal Modeling Approach

Page 7: The Impact of Electro- Thermal Coupling on RF Power Amplifiers

Lumped Thermal RC vs. Full 3D Electro-thermal Single HBT Device, Beta Drop due to heating

vs

AHBT_Model HBTM1

Xth2=0.0 Cth2=0.0 Rth2=0.0 Xth1=0.0 Cth1=5.0e-10 Rth1=1000.0 Tnom=25.0

Page 8: The Impact of Electro- Thermal Coupling on RF Power Amplifiers

Self Heating Mutual Heating Transient Heating

–  Long Term Failure

–  Thermal Stability / Balance

–  Device and Circuit Performance

–  Memory Effects

Review of Heating in Active Devices:

Areas where heating is impactful to PA Design:

Page 9: The Impact of Electro- Thermal Coupling on RF Power Amplifiers

Context: Commercial WLAN PA (HBT)

PA

Sw

itch Ctrl

GaAs HBT PA Layout

PA Schematic

Product Diagram

Page 10: The Impact of Electro- Thermal Coupling on RF Power Amplifiers

PA Bias Topology

rfRbmir VbeVVbe +=

Vbe vs. Temperature

Well known characteristic of PN junction diodes: Diode voltage is CTAT

Hold RF Device Temp Constant, vary Tmirror:

MIR

OUT

MIR

MIR

MIR

OUT

dVbedI

dTdVbe

dTdI *=

Iref

Imir

Vbemir

Qmir Qrf

Iout

Vberf

Rb

)()(

)(

rfrfT

mirrRb

VVbe

rfTMIR

OUT eVIs

dVbedI π

=

CmVdTdVbe

MIR

MIR !/1.1−≈

If TMIR ↑, VbeMIR ↓ If VbeMIR ↓, IOUT ↓

If Qmir is cooler than Qrf, the bias current Iout will increase, which can increase the RF gain

Page 11: The Impact of Electro- Thermal Coupling on RF Power Amplifiers

S2

S3

S1 S2

S3

S1

Thermally Coupled Bias Networks Thermally Decoupled Bias Networks

S1 bias S1 PA

S2 PA S2 bias

S3 PA

S3 bias

S1 bias

S2 bias S3 bias S1 PA

S2 PA

S3 PA

•  Here we have two schematically identical WLAN PA layouts to demonstrate the differences between modeling approaches. The three conditions are:

-  Isolated Self Heating Simulation (Static Thermal Network, Internal to AHBT Model) -  Full Electro-thermal Simulation with moderately thermal-coupled bias networks -  Full Electro-thermal Simulation with a poorly thermal-coupled bias network

Test Case: Two Different IC Layouts, Identical Schematic

Page 12: The Impact of Electro- Thermal Coupling on RF Power Amplifiers

Bias Configuration

Tmir < Trf, Ibias ↑, Gain ↑

Mirror RF

Pdmir << Pdrf

Temperature Coupling and Memory Effects …Apply a low frequency RF power pulse

T (°C)

TfTr G

ain

Pin

Gain @ ΔTr

Gain @ ΔTf

Resulting Gain Hysteresis

Trf

TmirrorΔTfΔTr

Time

Gai

n

Pin

Gain @ TriseGain @ TfallTmirror

Ther

mal

ly

Isol

ated

Pin

ΔTr~0

ΔTf~0Tf

Tr

RF Input Pwr

Trf

Pin

Time

Thermal Response to RF Burst

Ther

mal

ly

Cou

pled

RF Input Pwr T (°C)

Page 13: The Impact of Electro- Thermal Coupling on RF Power Amplifiers

Apply Time Dependent RF Power Burst, Plot gain curves at rising and falling edges

Input RF Pulse

RF Array Temp

Bias Temp thermally coupled case

thermally decoupled case

Due to thermal time constant, device temperature lags the RF Power Pulse

0.45 dB

0.2 dB

0 dB Self heating model

Pout (dBm)

Hysteresis in Gain Response (upramp/downramp)

ETH, Bias T-Decoupled

ETH, Bias T-coupled

Gai

n (d

B)

Placement of Bias Devices relative to RF Devices can enhance or minimize thermal memory effects!

Dev

ice

Tem

p (°

C)

Time (ms)

Transient Simulation with mutual heating between RF and bias device for different layout configurations

These plots compare:

Isolated Self Heating Model Mutual Heating Model, thermally decoupled Mutual Heating Model, thermally coupled

Simulation Results

Page 14: The Impact of Electro- Thermal Coupling on RF Power Amplifiers

– Modeling distributed heating is critical in RF Power Amplifier Design, especially for low frequency phenomena such as memory effects or device-to-device interaction

– Memory effects at low frequencies might be due to:

• Cross-circuit, layout-dependent mutual heating

• Distributed heating on longer time scales

–  This is a case where full Electro-thermal analysis is useful in the design process because it predicts problems that the intrinsic thermal RC network inside the device model does not

Closing

Page 15: The Impact of Electro- Thermal Coupling on RF Power Amplifiers

Feel free to contact me with any additional questions:

[email protected]

Thanks!