Reciprocal Mixing

You have a high-quality HF transceiver with a noise figure of 15 dB and an impressive IP3 specification. You also have an inexpensive software-defined radio with a noise figure of only 5 dB. On a quiet Tuesday morning with a clear band, the SDR hears weak signals better — its lower noise figure means a lower noise floor. But on a crowded Saturday during a major contest, with dozens of strong signals filling the 40-meter band, the expensive transceiver suddenly performs dramatically better. Weak DX that the SDR can barely detect is solid copy on the transceiver. What's happening?

The answer is reciprocal mixing — a phenomenon caused by phase noise in the receiver's local oscillator. It is one of the most important and least-understood limits on real-world receiver performance, and it explains why the Sherwood Engineering receiver rankings list phase noise as one of the two most critical receiver parameters (alongside IP3). Understanding reciprocal mixing will fundamentally change how you think about receiver specifications.

What Is Reciprocal Mixing?

In a superheterodyne receiver, the job of the first mixer is to multiply the incoming RF signal by a local oscillator (LO) signal to produce a difference frequency — the intermediate frequency (IF). If you are tuned to 14.200 MHz with a 9 MHz IF, the LO runs at 23.200 MHz. The mixer output contains 14.200 × 23.200 = 9.000 MHz (the IF) and other products that are rejected by the IF filter.

In an ideal receiver, the LO would be a perfectly clean sine wave — a single spectral line at exactly 23.200 MHz with no energy at any other frequency. Real oscillators are never that clean. They have phase noise — tiny, random fluctuations in the instantaneous frequency of the oscillator that spread energy into sidebands around the LO frequency. On a spectrum analyzer, a real oscillator looks like a peak with a noise skirt spreading out on both sides.

Reciprocal mixing occurs when a strong off-channel signal mixes with the LO's phase noise rather than with the clean LO carrier. The result is that the strong signal's energy, spread by the LO phase noise, appears as a broad noise pedestal in the IF passband — right where your desired weak signal lives. The strong signal does not need to be in your passband. It just needs to be at a frequency offset from your desired signal that matches a region of significant LO phase noise.

Here is the key insight: your desired signal is typically 50–80 dB weaker than the strong interferer. The LO phase noise at the offset of the strong signal might be −120 to −130 dBc/Hz. But if the strong signal is at −30 dBm and the phase noise at that offset is −120 dBc/Hz, the noise pedestal generated in the IF is at −30 + (−120) = −150 dBm/Hz. In a 2.4 kHz SSB bandwidth (which is 33.8 dB·Hz), this becomes −150 + 33.8 = −116 dBm. That may be well above your noise floor, completely swamping the weak signal you were trying to copy.

Left: a clean LO translates signals correctly — the strong signal becomes a clean IF spike. Right: a noisy LO spreads the strong signal's energy into a noise pedestal across the IF passband, burying the desired weak signal.

View Larger

Phase Noise — What It Is and Why It Exists

Every oscillator — whether it is a free-running LC circuit, a crystal oscillator, a voltage-controlled oscillator (VCO), or a phase-locked loop (PLL) — has some degree of phase noise. Phase noise arises from fundamental physical processes that cannot be entirely eliminated:

• Thermal noise (Johnson noise) in the oscillator's components — particularly the resistive losses in the tank circuit or crystal resonator — creates random fluctuations in the oscillator's energy that translate into phase fluctuations.
• Flicker noise (1/f noise) in the active device (transistor or amplifier) modulates the oscillator's phase at low offset frequencies.
• Reference oscillator quality in a PLL — the phase noise of the reference crystal is multiplied up by 20×log10(N) dB when the PLL multiplies the reference frequency by N.
• Loop filter bandwidth in a PLL — a narrow loop filter passes less reference phase noise but allows more VCO free-running phase noise to appear at larger offsets.

Phase noise is specified as a power spectral density (noise power per unit bandwidth) normalized to the carrier power. The unit is dBc/Hz (decibels below carrier per hertz). A specification of "−120 dBc/Hz at 1 kHz offset" means that the noise power in a 1 Hz bandwidth, measured 1 kHz away from the carrier, is 120 dB below the carrier power. Lower (more negative) numbers are better — they mean less noise in the sidebands.

Phase noise is measured at specific offset frequencies from the carrier. The phase noise floor typically improves (becomes lower) as the offset from the carrier increases, following a profile that includes regions dominated by flicker noise close in, thermal noise floor further out, and a flat thermal noise floor at very large offsets. For HF amateur radio, the critical offset range is 1 kHz to 30 kHz — the range over which strong adjacent-channel signals will cause reciprocal mixing problems.

The Reciprocal Mixing Mechanism Step by Step

To understand reciprocal mixing, you need to think about what the mixer actually does. The mixer's output is the mathematical product of its two input signals. If the RF input contains both your desired signal and a strong nearby signal, and the LO contains both the clean LO carrier and its phase noise sidebands, then the mixer output contains all possible products of these components.

Consider the four products that matter:

1. Desired signal × clean LO → desired signal translated to the IF at the correct frequency. This is the signal you want.
2. Strong interferer × clean LO → strong interferer translated to the IF at a frequency determined by the offset from the LO. If the IF filter rejects this frequency, no problem.
3. Desired signal × LO phase noise → a small noise skirt around the desired signal in the IF. Usually small enough to ignore.
4. Strong interferer × LO phase noise → this is the problem. The strong interferer mixes with every sideband component of the LO phase noise, spreading the strong signal's energy across a range of IF frequencies. If any of this noise pedestal falls within the IF passband, it raises the noise floor at the desired signal's IF frequency.

The critical point is in step 4. The strong interferer is, say, 50 dB stronger than the desired signal. The LO phase noise at the offset equal to the spacing between the two signals is, say, −120 dBc/Hz. The noise pedestal from the strong interferer, expressed per unit bandwidth, is:

This formula shows that the reciprocal mixing noise floor is determined by both the strength of the interfering signal and the phase noise at the relevant offset. A stronger interferer or worse phase noise both make the problem worse.

Calculating the Reciprocal Mixing Noise Floor

With the formula established, let's work through a realistic example that shows exactly how much reciprocal mixing can degrade receiver performance.

This example is not hypothetical. It describes the real-world performance difference between a low-cost SDR and a high-quality HF transceiver, or between a budget radio and a high-performance contest receiver. The thermal noise figure advantage of the cheaper receiver becomes irrelevant on a crowded band because reciprocal mixing from the many strong signals dominates the effective noise floor.

The formula can also be rearranged to find the minimum spacing at which a given strong signal will no longer cause problems, or to determine what phase noise specification is needed to keep reciprocal mixing below the thermal noise floor for a worst-case input signal. For most HF amateur radio applications, a phase noise specification of better than −130 dBc/Hz at 1 kHz offset and −145 dBc/Hz at 10 kHz offset is needed for good performance on crowded bands.

Phase Noise vs Noise Figure — Which Matters More?

Both phase noise and noise figure limit receiver performance, but in different operating conditions. Understanding which one dominates in a given situation helps you make better equipment decisions.

Noise figure dominates when:

• The operating band is quiet (no strong nearby signals)
• You are listening for extremely weak signals (EME, satellite, meteor scatter)
• You are operating in a remote location away from other amateurs and broadcast stations
• The receiving antenna is small and the thermal noise of the environment is low

Phase noise dominates when:

• You are operating on a crowded HF band (40m or 80m during contests)
• There are strong nearby stations — neighboring amateurs, broadcast stations, or beacon transmitters
• You are at a multi-operator site (Field Day, DXpedition, contest station)
• You are using a software-defined radio with a relatively poor local oscillator

The crossover point — where reciprocal mixing noise equals thermal noise floor noise — depends on the phase noise of the receiver and the level of the strong nearby signal. For most modern HF transceivers on a moderately busy 40-meter band, reciprocal mixing is the dominant noise source once there are more than a handful of strong signals on the band.

A receiver with a noise figure of 12–15 dB but excellent phase noise of −145 dBc/Hz at 1 kHz will perform dramatically better than a receiver with NF = 3 dB but phase noise of −115 dBc/Hz at 1 kHz on a crowded band. This is why high-performance DX receivers often sacrifice some sensitivity (higher NF) in order to achieve outstanding phase noise performance — and why their operators consistently report superior performance on crowded bands even when a budget SDR appears to show lower noise on an empty band.

Reciprocal Mixing vs IMD — Similar Effects, Different Causes

Both reciprocal mixing and intermodulation distortion (IMD) raise the apparent noise floor when strong signals are present, and both can cause a desired weak signal to become unreadable. They are distinct phenomena with different causes, and telling them apart is important because the remedies differ.

In practice, on a crowded HF band, both mechanisms are often operating simultaneously. High-performance receivers reduce both by combining excellent phase noise with high IP3. Budget receivers and SDRs often suffer from both simultaneously, which is why their real-world performance on busy bands is much worse than their headline NF and sensitivity figures might suggest.

A quick test to determine which mechanism is dominant: tune to a strong known signal and note the apparent noise floor at a known offset. Then back off the strong signal by 10 dB and observe the change. If the noise floor drops by approximately 10 dB, reciprocal mixing dominates (linear relationship). If it drops by approximately 30 dB, IMD dominates (cubic relationship). If it drops by something in between, both are contributing.

SDR Receivers and Phase Noise

Software-defined radio receivers based on low-cost chips like the RTL2832U (used in RTL-SDRs) have become popular tools in amateur radio for satellite work, spectrum monitoring, and wide-band signal scanning. They offer remarkable capability at low cost. However, their local oscillators are significantly noisier than those in quality HF transceivers, which limits their performance on crowded HF bands.

The original RTL-SDR dongles used a simple crystal-based LO reference with no temperature compensation, resulting in phase noise in the range of −100 to −110 dBc/Hz at 1 kHz offset. This is 30–40 dB worse than a good HF transceiver. On a crowded 40-meter evening with dozens of strong signals, the RTL-SDR's waterfall display can show a dramatically elevated noise floor as each strong signal contributes a noise pedestal through reciprocal mixing.

The RTL-SDR V3 and similar improved models include a temperature-compensated crystal oscillator (TCXO), which improves phase noise to approximately −110 to −115 dBc/Hz at 1 kHz — still significantly worse than a quality transceiver but meaningfully better than the original design. Higher-end SDR platforms (SDRplay, Airspy, Hermes-Lite) use better oscillator designs and typically achieve −120 to −130 dBc/Hz, approaching the performance of mid-range HF transceivers.

Adding an external upconverter ahead of an RTL-SDR for HF operation can improve or worsen phase noise depending on the design of the upconverter's local oscillator. A high-quality upconverter with a low-phase-noise LO can actually improve the overall system phase noise compared to operating the SDR directly at HF frequencies.

Receiver Specifications and Phase Noise

Most receiver manufacturers specify phase noise at the antenna terminals — that is, the input-referred reciprocal mixing performance as seen from the antenna. This is the most useful form of the specification because it tells you directly what level of strong signal at a given offset will cause reciprocal mixing problems at a given level.

The Sherwood Engineering receiver ranking database (available online) is one of the most referenced sources for real-world receiver performance. It rates receivers primarily on two parameters: IP3 (large-signal dynamic range) and phase noise (reciprocal mixing dynamic range). Both are measured at 2 kHz and/or 5 kHz offsets. The combination of these two parameters — called the "dynamic range" in Sherwood's terminology — is a much better predictor of real-world performance on crowded HF bands than the manufacturer's noise figure specification.

When evaluating a receiver for purchase, check both the Sherwood phase noise measurement (if available) and the manufacturer's IP3 specification. A receiver that excels in both parameters will perform well across all operating environments. A receiver that trades one for the other will show its weakness when operating conditions are not in its favor.