How Does DPD Optimize SSPAs for Better Linearity?

Getting maximum power from your SSPA without creating signal distortion is a constant challenge. This trade-off between power and linearity often forces unwanted compromises in system design and efficiency.

The best way to optimize an SSPA is by using Digital Pre-Distortion (DPD).¹ DPD digitally models the amplifier's non-linear behavior, including memory effects², and creates an inverse distortion signal. This pre-corrected signal, when amplified, results in a clean, linear output even near saturation.

I've spent years in the lab pushing amplifiers to their limits. The goal is always the same: get as much clean power as possible. You need high linearity for today’s complex modulation schemes, but you also need high efficiency to manage power and heat. It’s a difficult balancing act. This is where understanding different linearization techniques becomes critical. Let's look at why DPD has become the go-to solution for high-performance systems and how it compares to older methods.

Why Does SSPA Linearity Matter So Much?

Your amplifier is pushing out a signal, but is it the right signal? Non-linearity distorts your waveform, causing it to spill into adjacent channels, which can ruin system performance.

SSPA linearity is critical because it directly impacts signal quality and spectral efficiency. A linear amplifier faithfully reproduces the input signal, preventing data corruption and interference with nearby channels. This is essential for modern communication systems that use spectrally dense modulation schemes.

Deeper Dive: The Real-World Impact of Non-Linearity

When an SSPA operates in its non-linear region, it generates unwanted distortion products. This isn't just a theoretical problem; it has serious practical consequences that I've seen derail projects. The two main types of distortion, AM-AM and AM-PM, directly degrade signal quality.

Key Performance Metrics Affected

The most common metric we use to measure this is the Adjacent Channel Power Ratio (ACPR)³. It tells us exactly how much of our signal power is leaking into the bands where it shouldn't be. Another is the Error Vector Magnitude (EVM)⁴, which measures how far the actual received symbols deviate from their ideal locations in the constellation diagram. Poor linearity leads to high spectral regrowth (bad ACPR) and a scattered constellation (bad EVM). For systems like 5G, satellite communications, and military radar, this is unacceptable. It can mean dropped calls, slow data, or even a complete loss of signal lock. The table below shows how linearity impacts different applications.

Ultimately, maintaining linearity is about ensuring the integrity and efficiency of the entire RF system.

What Are the Limits of Analog Pre-Distortion?

Engineers have tried to fix non-linearity for decades. An early solution was analog pre-distortion (APD). But it often creates more problems than it solves, especially with modern wideband signals.

Analog pre-distortion is limited because it uses simple, memoryless circuits. These circuits cannot model or correct for an SSPA's dynamic memory effects, which are a major source of distortion. As a result, APD only provides a small improvement of a few dB in ACPR.

Deeper Dive: Why APD Falls Short

I remember a project where we were working on a high-power amplifier for a client. We needed to meet a tight spectral mask, and the first attempt involved an APD circuit. The results were disappointing. We got a small improvement, but not nearly enough to pass the requirement. The core issue lies in how APD works compared to the problem it's trying to solve.

The Mismatch with Amplifier Behavior

An SSPA's distortion isn't static. It changes based on the signal's history, a phenomenon known as memory effects. These are caused by thermal effects in the transistor and the frequency response of the bias network⁵. The amplifier's current output state depends on the inputs from moments ago, not just the present input.

APD circuits, however, are typically "memoryless." They apply a fixed, static non-linearity that is the inverse of the amplifier's static AM-AM curve. This can offer some correction for slow-varying signals but completely fails for wideband, dynamic signals where memory effects dominate.

The table below contrasts the nature of the problem with the limitations of the APD solution.

Because APD can't address memory effects, its ability to improve ACPR is fundamentally capped. We saw this firsthand; a 3-5 dB improvement was the best we could achieve. For modern standards, that simply isn't enough.

How Does DPD Overcome SSPA Memory Effects?

After the limited success with APD, we switched our focus to Digital Pre-Distortion (DPD). The difference was night and day. DPD doesn't just approximate; it models the amplifier's real behavior.

DPD overcomes memory effects by using advanced digital models, like the memory polynomial. This model captures not only the amplifier's static non-linearity but also its dynamic, time-dependent behavior. By accurately predicting the total distortion, DPD can generate a precise inverse signal to cancel it out.

Deeper Dive: Modeling and Canceling Distortion

The magic of DPD happens in the digital domain, usually within an FPGA or DSP, long before the signal ever reaches the SSPA. It's a two-step process: learning and correcting.

Understanding the DPD Correction Process

First, the DPD system "learns" the SSPA's unique distortion fingerprint. It sends a known signal through the amplifier and captures the distorted output. A feedback path brings this output signal back to the digital processor. Here, an adaptation algorithm compares the original signal to the distorted one. This comparison allows it to build a highly accurate mathematical model of the amplifier's behavior, including its memory effects.

The most common model used is the Generalized Memory Polynomial (GMP)⁶. This model is powerful because it includes "taps" that look at previous signal samples.

• y(n): The predicted distorted output at time n.
• x(n-q): The input signal at previous time steps q.
• a_kq: Coefficients that represent the amplifier's distortion characteristics.

The model uses these coefficients to predict exactly how the SSPA will distort any given input.

Creating the "Anti-Distortion" Signal

Once the model is built, the DPD system enters its correction phase. For any new input signal x(n) that needs to be transmitted, the DPD block uses the GMP model to pre-emptively calculate the inverse of the distortion the SSPA will add. It warps the signal in the opposite direction of the amplifier's distortion. When this pre-distorted signal passes through the SSPA, the amplifier's own distortion effectively cancels out the pre-distortion, resulting in a clean, linear output. This process is continuous, allowing the DPD to adapt to changes in temperature or device aging. This is how we can achieve ACPR improvements of 15 dB or more⁷, something impossible with APD.

What Are the Practical Benefits of Using DPD with SSPAs?

Implementing DPD isn't just an academic exercise. For us and our customers, it translates directly into better performance, lower costs, and a more competitive product.

The main benefit of DPD is that it allows SSPAs to run much closer to their saturation point while maintaining excellent linearity. This dramatically improves power efficiency⁸, reduces heat and operational costs, and enables the use of smaller, less expensive amplifiers for the same output power.

Deeper Dive: Translating Technical Gains into Business Value

When we can run an SSPA closer to its peak power, a cascade of positive effects occurs. On that satellite project I mentioned, implementing DPD meant we could meet the strict out-of-band emissions requirements without having to back off the power significantly. This had huge implications for the client.

Key Business and System Advantages

The improvement in ACPR is the technical enabler, but the real-world value is what matters to system designers and business owners. The ability to operate at a higher power-added efficiency (PAE) is probably the most significant benefit.

• Reduced Operational Costs (OPEX): A more efficient amplifier consumes less DC power and generates less waste heat. This leads to lower electricity bills and reduced cooling requirements, which are major operational expenses in data centers and base stations.
• Reduced Capital Costs (CAPEX): Since DPD allows you to get more linear power out of a given amplifier, you can often use a smaller, less expensive SSPA to do the job of a much larger, un-linearized one. For example, a DPD-enabled 100W SSPA might deliver the same linear output power as a 200W SSPA operating with a large back-off⁹.
• Enhanced System Performance: The clean signal produced by a DPD-linearized SSPA supports higher-order modulation schemes. This translates to higher data throughput and better spectral efficiency, allowing operators to serve more users or transmit more data in the same amount of bandwidth.

The following table summarizes these tangible benefits.

At Safari Microwave, integrating advanced linearization is key to our high-power products, like our 3000W SSPAs. It allows us to deliver on our promise of High Power and Premium-Efficiency.

Conclusion

DPD solves the core SSPA trade-off. It delivers high power and high linearity at the same time, leading to more efficient, cost-effective, and higher-performing RF systems.