Most engineers working in satellite communications treat phase noise as a datasheet line item. You check the dBc/Hz number at 10 kHz offset, see if it meets a requirement, and move on. After years of designing, testing, and troubleshooting SATCOM links—from L-band terminals in the dirt to Ka-band gateways baking on a rooftop—I’ve learned that phase noise is rarely just a number. It worms its way into system performance in ways that spec sheets don’t capture. When you’re pushing data rates, clawing for margin, or hunting down interference, phase noise can make or break your link.
This article walks through what phase noise actually does inside a SATCOM receiver, why modern waveforms make it more critical than ever, and how to think about it without the marketing fluff—whether you’re specifying, measuring, or debugging a system.
Phase Noise Is Not Just an Oscillator Problem
When we talk about phase noise, the conversation usually starts with the local oscillator. That makes sense—the LO is the first place where short-term frequency instability sneaks into the signal chain. But stopping there misses half the story. In a real SATCOM terminal, phase noise contributions come from synthesizers, clock distribution, power supply ripple coupling, and even mechanical vibration in airborne or mobile platforms. I’ve seen cases where a perfectly clean benchtop measurement fell apart once the unit got bolted to a vibrating antenna mount. Literally shook the performance right out of it.
What matters is the composite phase noise at the point of downconversion. A low-noise OCXO feeding a PLL can still spit out a noisy signal if the loop filter is too wide or the reference spur suppression is poor. In block upconverters (BUCs) and low-noise block downconverters (LNBs), the internal oscillators are often free-running DROs or PLLs with simplified filtering to shave cost and power. The phase noise you measure at the L-band IF port is the sum of all these contributions, plus whatever the modem’s own clock recovery adds. It’s a messy stew, and you don’t get to pick out the ingredients.
Where Phase Noise Hides in the Chain
- Reference oscillators: Even a 10 MHz GPS-disciplined reference can have elevated close-in noise if the disciplining loop is poorly tuned. I’ve seen a GPSDO with worse phase noise at 100 Hz than a free-running OCXO—so much for “disciplined.”
- Synthesizer spurs: Fractional-N spurs that fall inside the signal bandwidth are effectively phase noise at discrete offsets. They’re sneaky because they don’t look like noise on a spectrum analyzer, but the modem feels them all the same.
- Power supplies: Switching regulators create sidebands that modulate the LO, especially in compact outdoor units where shielding is minimal. A 100 mV ripple on a VCO supply can turn into a phase noise bump you’ll chase for days.
- Vibration-induced noise: Crystal oscillators are accelerometer-sensitive; on helicopters, UAVs, or fast-moving vehicles, vibration spectra translate directly to phase noise sidebands. If your platform vibrates, your LO sings along.
Modern Waveforms Raise the Stakes
Twenty years ago, most SATCOM links used QPSK or BPSK with relatively low symbol rates and wide carrier spacing. Phase noise was present, but the modulation formats were forgiving. You could get away with a lot. Today, the push for spectral efficiency means 16APSK, 32APSK, and even 64APSK are common in DVB-S2X and proprietary waveforms. These higher-order modulations pack more bits per symbol by shrinking the distance between constellation points. When phase noise rotates those points, the receiver’s decision boundaries blur, and bit errors climb. Suddenly, that cheap LNB doesn’t look so cheap anymore.
The problem gets worse with low symbol rates. A 100 ksps carrier’s bandwidth is narrow enough that close-in phase noise—within a few kilohertz of the carrier—dominates the signal’s phase trajectory. I’ve seen links that meet their Eb/No margin on paper fail in the field because the phase noise at 1 kHz and 10 kHz offsets was underestimated. The modem’s carrier recovery loop can track slow phase wander, but fast noise components pass straight through to the symbol decision stage. It’s like trying to read a sign while someone jiggles your glasses.
Phase Noise and Carrier Recovery
All coherent modems use a carrier recovery loop—typically a PLL or a feedforward estimator—to remove phase rotation from the received signal. The loop bandwidth sets a boundary: phase noise inside the loop bandwidth is tracked and cancelled; phase noise outside it is not. This means the system’s sensitivity to phase noise is shaped by the loop filter. A narrow loop bandwidth helps suppress phase noise but slows acquisition and makes the link vulnerable to Doppler in mobile scenarios. A wide loop tracks dynamics but lets more oscillator noise through. It’s a trade-off with no free lunch.
Design becomes a balancing act. In fixed satellite services, you can often use narrow loop bandwidths and rely on stable frequency references. In LEO or MEO constellations, the Doppler rate forces a wider loop, which demands lower oscillator phase noise to keep the residual jitter acceptable. I’ve had to reject otherwise good LNBs because their phase noise at 100 Hz offset—often not even plotted on typical datasheets—was too high for a low-symbol-rate MEO link. The vendor was not happy. Neither was I.
How Much Phase Noise Is Too Much?
There’s no universal number. The allowable phase noise depends on the modulation order, symbol rate, FEC coding rate, and target BER. A practical approach starts with the integrated phase noise over the relevant bandwidth. For a given carrier recovery loop bandwidth, you integrate the single-sideband phase noise from the loop bandwidth out to the symbol rate (or the noise bandwidth of the receiver, whichever is narrower). The result, expressed in degrees or radians RMS, tells you how much residual phase jitter the decision circuit sees.
For QPSK, a few degrees of RMS phase error is manageable—modern FEC handles it without much degradation. For 32APSK with a high code rate, you might need less than one degree RMS. The DVB-S2X implementation guidelines give some reference values, but they assume a particular phase noise mask that may not match your hardware. I always recommend measuring the actual phase noise of the complete downconversion chain and running it through a simulation with the intended modem parameters. Guessing is expensive.
A Field Measurement Trap
One mistake I see repeatedly is measuring phase noise with a spectrum analyzer that has insufficient dynamic range. A typical analyzer’s own phase noise can mask the DUT’s performance, especially at close offsets. For serious work, use a phase noise test set or a cross-correlation method. If you must use a spectrum analyzer, verify its phase noise floor first with a known clean source. I’ve seen an engineer chase a “noisy” LNB for days before realizing the analyzer’s internal LO was the problem. He was not smiling when he figured it out.
Phase Noise in the Frequency Domain: More Than Just Jitter
Phase noise also acts as a reciprocal mixer in the presence of strong interferers. The LO’s phase noise sidebands mix with a nearby carrier, translating that carrier’s energy into the IF passband. This reciprocal mixing can desensitize the receiver even if the interferer is far enough away that its main lobe does not overlap the desired signal. In crowded satellite transponders where multiple carriers share the same polarization, this effect limits the achievable carrier-to-interference ratio. You can’t power your way out of it.
This is why the phase noise mask for SATCOM LNBs often includes tight specifications at offsets corresponding to typical adjacent carrier spacing—say 1 to 10 MHz. A low-cost LNB might have excellent noise figure and gain flatness but fail in a multi-carrier scenario because its phase noise at 5 MHz offset is 10 dB worse than a higher-grade unit. When you’re trying to fit more carriers into a transponder, that 10 dB directly reduces capacity. Someone else gets the slot you thought was yours.
Practical Steps for Better Phase Noise Performance
Improving phase noise starts with component selection, but it doesn’t end there. System integration often undoes the careful work done at the oscillator level. Some steps I’ve found effective in fielded designs:
- Buffer the reference: A high-quality OCXO can be degraded by loading from multiple synthesizers. Use distribution amplifiers with high reverse isolation. One bad load shouldn’t spoil the whole bus.
- Mind the PLL loop bandwidth: A wider loop suppresses VCO noise but passes more reference noise. Model the total output noise and pick the crossover point where the two noise sources are equal. Don’t just copy a reference design—run the numbers for your parts.
- Clean power: Linear regulators for oscillator and PLL supplies are worth the power penalty. If you must use switchers, synchronize them to a frequency outside the loop bandwidth and use ferrite beads extensively. A little heat is better than a lot of noise.
- Vibration isolation: In mobile terminals, mount oscillators on damped standoffs and avoid rigid PCB clamping near the crystal. Let the board flex a little. Your phase noise plot will thank you.
- Test end-to-end: Measure phase noise at the IF output with the full chain connected, including the LNB/BUC, cables, and modem reference. Don’t trust a standalone oscillator measurement. The system is the spec—not the component.
The Modem’s Role
Modern modems aren’t passive victims of phase noise. Adaptive equalizers and pilot symbols help, but they have limits. Some SDR-based modems can estimate and cancel phase noise adaptively, especially if the phase noise has a predictable structure (e.g., power supply ripple at known frequencies). If you have control over the modem’s firmware, consider feeding it a phase noise profile to optimize the carrier recovery loop. This is not yet standard practice in commercial terminals, but it makes a measurable difference in marginal links. Sometimes a few lines of code save a hardware redesign.
When to Spend Money on Phase Noise
Not every link needs an expensive ultra-low-phase-noise reference. For a wideband carrier with strong FEC, the integrated phase noise budget may be generous. The cases where spending more pays off include: low symbol rates, high-order modulations, multi-carrier receivers, and platforms with high Doppler dynamics. If you’re designing a terminal that needs to work across multiple scenarios—as many service providers now demand—the safest path is to specify phase noise for the worst-case waveform and let the easier cases take care of themselves. Build it for the hard stuff, and the easy stuff just works.
In practice, a mid-grade OCXO with a well-designed PLL synthesizer can meet most fixed-service requirements. For SOTM (SATCOM on the move) or LEO tracking, a low-noise OCXO or even a rubidium reference might be needed, depending on the Doppler rate and the modem’s capability. Don’t guess—run the link budget with the actual phase noise numbers and a realistic carrier recovery model. Spreadsheets don’t lie, but assumptions do.
FAQ: Phase Noise in SATCOM
What is the difference between phase noise and jitter?
Phase noise is the frequency-domain representation of short-term oscillator instability, expressed as dBc/Hz at a given offset from the carrier. Jitter is the time-domain deviation of a signal’s edges from their ideal positions. In SATCOM, phase noise is more directly useful because it maps to the spectral regrowth and reciprocal mixing effects that affect link performance. However, integrated phase noise over a given bandwidth can be converted to RMS phase jitter in degrees or radians, which is what the modem’s decision circuit sees. They’re two sides of the same coin—pick the one that fits your measurement.
How does phase noise affect the bit error rate?
Phase noise adds random phase rotation to the received symbols. In the constellation diagram, this spreads each ideal point into an arc-shaped cloud. The receiver’s slicer sees a reduced distance between adjacent points, effectively lowering the signal-to-noise ratio for a given BER. Forward error correction can compensate to some extent, but once the RMS phase error exceeds a threshold that depends on the modulation order, the BER floor rises sharply and cannot be improved by increasing signal power. You can’t just shout louder—the noise is inside the room.
Can I improve phase noise by filtering after downconversion?
No. Once the signal is downconverted, the phase noise from the LO is already imprinted on the IF signal. Filtering in the IF or baseband can remove amplitude noise and out-of-band interference, but it does not remove phase modulation. The only way to reduce the effect of phase noise is in the modem’s carrier recovery loop, which can track and cancel some portion of it. That’s why the oscillator and synthesizer design is the primary defense. Fix it at the source, because no filter downstream will save you.
What phase noise level is acceptable for DVB-S2X 32APSK?
There is no single number, but as a practical starting point, the integrated phase noise from the carrier recovery loop bandwidth (often a few hundred kilohertz for typical symbol rates) should be below about 0.8 degrees RMS. This requires a relatively clean LO, typically better than -80 dBc/Hz at 1 kHz offset and -95 dBc/Hz at 10 kHz offset for a Ku-band LNB, depending on the loop bandwidth. Always simulate with the actual modem parameters, because the interaction between FEC, pilot patterns, and phase noise is complex. Rules of thumb are fine for a sanity check, but simulation is what keeps you out of trouble.