Thermal AM-AM distortion is a memory effect where the gain of a power amplifier at a given instantaneous input amplitude is not fixed, but is modulated by the junction temperature history of the transistor. Unlike memoryless AM-AM distortion, which depends solely on the instantaneous signal envelope, this mechanism arises because self-heating from prior high-power signal peaks alters carrier mobility and threshold voltage, causing the amplifier's gain to sag or expand for subsequent samples. This creates a dynamic, history-dependent deviation from the static amplitude-to-amplitude characteristic.
Glossary
Thermal AM-AM Distortion

What is Thermal AM-AM Distortion?
Thermal AM-AM distortion is a dynamic nonlinear gain compression or expansion in a power amplifier where the instantaneous amplitude-to-amplitude transfer characteristic deviates from its static curve due to the device's temperature history.
This phenomenon is driven by the convolution of the signal's envelope power with the device's thermal impedance, introducing low-frequency lag that standard memory polynomials often fail to capture. The resulting gain modulation manifests as a slow, signal-dependent compression that degrades Error Vector Magnitude (EVM) and creates asymmetric spectral regrowth. Effective compensation requires thermal-aware predistortion techniques, which augment the predistorter with either real-time temperature sensing or an electro-thermal model to de-embed the thermal contribution from the instantaneous nonlinearity.
Key Characteristics
Thermal AM-AM distortion represents a dynamic deviation from a power amplifier's static gain curve, driven by the device's temperature history rather than instantaneous input amplitude alone.
Temperature-Dependent Gain Modulation
Unlike memoryless AM-AM distortion, thermal AM-AM causes the amplifier's gain compression or expansion to shift dynamically as junction temperature fluctuates. Key mechanisms include:
- Threshold voltage drift: As temperature rises, the transistor threshold voltage decreases, altering the conduction angle and small-signal gain
- Carrier mobility degradation: Increased lattice vibrations at higher temperatures reduce electron mobility, lowering transconductance and saturated output power
- Quiescent bias shift: Self-heating changes the DC operating point, moving the amplifier along its load line and modifying the instantaneous gain profile
The result is a history-dependent gain curve where identical instantaneous input amplitudes produce different output amplitudes depending on prior signal envelope history.
Envelope-Dependent Heating Dynamics
Thermal AM-AM distortion is driven by envelope frequency heating, where the low-frequency components of the modulated signal's amplitude envelope fall within the thermal bandwidth of the device:
- The instantaneous power dissipation waveform is proportional to the squared envelope of the RF signal
- When envelope frequency components are slower than the thermal cutoff frequency, the junction temperature tracks these variations
- This creates a thermal convolution effect: the instantaneous junction temperature equals the convolution of dissipated power with the device's thermal impulse response
- Modern wideband signals (e.g., 5G NR with 100 MHz bandwidth) have envelope components spanning DC to hundreds of MHz, ensuring significant spectral content within the thermal response bandwidth
Spectral Asymmetry Signature
A defining characteristic of thermal AM-AM distortion is spectral asymmetry in the output spectrum that cannot be corrected by memoryless linearization:
- The thermal memory effect introduces a frequency-dependent phase shift between the envelope and the resulting gain modulation
- This dispersive behavior causes unequal spectral regrowth on the upper and lower sidebands of the modulated signal
- Unlike electrical memory effects (which operate at megahertz rates), thermal asymmetry appears at kilohertz to low-megahertz offset frequencies from the carrier
- The asymmetry pattern is temperature-history dependent: a long high-power burst creates a different asymmetry profile than a short peak with the same average power
- This signature distinguishes thermal AM-AM from other nonlinear mechanisms and requires thermal-aware predistortion for effective compensation
Interaction with Electrical Memory Effects
Thermal AM-AM does not occur in isolation—it interacts with and exacerbates other nonlinear memory mechanisms:
- Bias network modulation: Temperature-induced changes in quiescent current alter the impedance seen by the baseband bias network, coupling thermal and electrical memory
- GaN trapping synergy: In GaN HEMTs, thermally activated electron trapping in buffer layers and surface states creates a compound memory effect where temperature accelerates charge capture and emission rates
- AM-AM/AM-PM coupling: Thermal gain compression simultaneously modifies the device's nonlinear capacitances, inducing correlated thermal AM-PM distortion that rotates the constellation diagram
- Multi-stage interaction: In multi-stage power amplifiers, driver-stage thermal AM-AM creates a time-varying input signal for the final stage, compounding the overall distortion
This coupling necessitates electro-thermal modeling that jointly captures both thermal and electrical memory for accurate behavioral prediction.
Long-Term Memory Duration
Thermal AM-AM distortion exhibits memory durations orders of magnitude longer than electrical memory effects, governed by the device's thermal time constants:
- Die-level time constants: The semiconductor die itself has thermal time constants in the microsecond to millisecond range due to its small thermal mass
- Package-level time constants: The package substrate, die attach, and heat spreader introduce time constants from milliseconds to seconds
- Heat sink time constants: The external cooling solution can have time constants of seconds to minutes, though these typically affect only the average operating point
- The distributed nature of thermal capacitance creates a multi-time-constant response that cannot be captured by a single exponential decay
- This long memory span requires predistorter models with deep memory taps or recurrent neural network architectures that can track state over extended time horizons
Measurement and Characterization Techniques
Isolating thermal AM-AM from other distortion mechanisms requires specialized pulsed and modulated measurement techniques:
- Pulsed I-V characterization: Applying short-duration pulses with varying quiescent temperatures separates thermal effects from trapping by controlling the thermal state independently
- Two-tone envelope modulation: Using a low-frequency envelope modulation on an RF carrier creates controlled thermal excitation at known frequencies, enabling extraction of the thermal transfer function
- Step-response thermometry: Measuring gain changes following a step in average power reveals the thermal impulse response through the time-dependent gain recovery
- Infrared thermography: Direct spatial temperature measurement across the transistor channel provides validation data for thermal models and confirms the temperature distribution causing AM-AM distortion
- Load-pull with thermal control: Combining active load-pull with precise baseplate temperature control isolates the thermal contribution to dynamic gain compression
Frequently Asked Questions
Clear, technically precise answers to the most common questions about how temperature dynamics create nonlinear amplitude distortion in power amplifiers.
Thermal AM-AM distortion is a nonlinear gain compression or expansion in a power amplifier that is dynamically modulated by the device's temperature history, deviating from the static, instantaneous amplitude-to-amplitude characteristic. Unlike instantaneous AM-AM, which depends solely on the current input envelope magnitude, thermal AM-AM introduces a long-term memory effect where the gain at any given moment is a function of the signal envelope's prior trajectory. This occurs because power dissipation heats the transistor junction, altering carrier mobility and threshold voltage, which in turn shifts the gain profile. The result is a hysteresis-like behavior in the AM-AM transfer curve: the gain for a rising envelope differs from the gain for a falling envelope, creating a distortion that cannot be corrected by memoryless predistorters and requires thermal-aware linearization techniques.
Enabling Efficiency, Speed & Accuracy
Intelligent Analysis, Decision & Execution
We build AI systems for teams that need search across company data, workflow automation across tools, or AI features inside products and internal software.
Talk to Us
Search across company data
Give teams answers from docs, tickets, runbooks, and product data with sources and permissions.
Useful when people spend too long searching or get different answers from different systems.

Automate internal workflows
Use AI to route work, draft outputs, trigger actions, and keep approvals and logs in place.
Useful when repetitive work moves across multiple tools and teams.

Add AI to products and internal tools
Build assistants, guided actions, or decision support into the software your team or customers already use.
Useful when AI needs to be part of the product, not a separate tool.
Related Terms
Understanding thermal AM-AM distortion requires familiarity with the interconnected thermal, electrical, and modeling concepts that govern dynamic amplifier nonlinearity.
Thermal Memory Effect
A distortion mechanism where the device's temperature history, caused by signal envelope variations, alters its instantaneous electrical behavior. This creates a long-term nonlinear memory that cannot be corrected by memoryless linearization. The effect manifests when the signal's envelope bandwidth falls within the thermal bandwidth of the device, typically in the kilohertz to megahertz range.
Self-Heating
The process by which power dissipation within the transistor channel increases its own junction temperature. This dynamic heating leads to shifts in:
- Carrier mobility (degrading drain current)
- Threshold voltage (altering turn-on characteristics)
- Transconductance (modifying gain profile)
The resulting gain compression or expansion is the physical root cause of thermal AM-AM distortion.
Thermal Impedance
A measure of a material's resistance to heat flow, defining the dynamic relationship between power dissipation and temperature rise. Represented as Zth(t), it captures both:
- Static thermal resistance (Rth, steady-state)
- Transient thermal response (dictated by thermal capacitance)
This impedance function is the transfer function that converts the instantaneous power envelope into the junction temperature waveform driving AM-AM distortion.
Thermal Time Constant
The characteristic time required for a device's junction temperature to reach approximately 63.2% of its steady-state value following a step change in power dissipation. Multiple time constants exist in a real device:
- Die-level: nanoseconds to microseconds
- Package-level: milliseconds
- Heat sink: seconds to minutes
The slowest time constants dictate the memory duration that predistorters must compensate for.
Electro-Thermal Modeling
A co-simulation technique that couples semiconductor device physics with dynamic heat generation and dissipation equations. This approach simultaneously solves:
- Electrical equations (carrier transport, current-voltage relationships)
- Thermal equations (heat diffusion, Fourier's law)
The result is a unified prediction of temperature-dependent nonlinearities including thermal AM-AM and AM-PM distortion under realistic modulated signal conditions.
Thermal-Induced Memory Polynomial
A behavioral model structure that augments standard memory polynomials with additional terms specifically designed to capture low-frequency, long-duration thermal lag effects. Key characteristics:
- Standard terms: model short-term electrical memory
- Thermal terms: use heavily decimated samples or integrated power history
- Cross-terms: capture interaction between instantaneous amplitude and thermal state
This structure enables accurate predistortion without requiring explicit temperature sensing.

About the author
Prasad Kumkar
CEO & MD, Inference Systems
Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.
His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.
Partnered with leading AI, data, and software stack.
How We Work
Custom AI workflows for your Business
One-fit-all AI don't work for modern businesses. At Inferensys, we aim to understand your business & custom requirements; which we use to define most efficient agentic workflows, the data, and the tools for your business.
01
Review the use case
We understand the task, the users, and where AI can actually help.
Read more02
Pick the right approach
We define what needs search, automation, or product integration.
Read more03
Build the first useful version
We implement the part that proves the value first.
Read more04
Improve from there
We add the checks and visibility needed to keep it useful.
Read moreThe first call is a practical review of your use case and the right next step.
Talk to Us