I’ve spent more hours staring at a spectrum analyzer screen than I care to count, often in the back of a van with a lukewarm coffee balanced on a rack panel. The first thing you learn in the field is that a clean spectrum display doesn’t always mean a clean signal. The second thing you learn is that most people rush the process. They hook up the gear, see a carrier, and call it a day. That’s not analysis; that’s just looking. Proper satellite spectrum analysis is a methodical, physical-layer investigation. It’s about understanding what your dish is actually seeing, separating the signal from the noise—and the noise from the self-inflicted interference.
This guide is built for the engineer who has to align a dish on a rooftop in the rain, or troubleshoot a maritime VSAT link that keeps dropping out for no obvious reason. We’ll walk through the equipment setup, the key measurements, and the common pitfalls that turn a simple carrier check into a wild goose chase.
1. Start with the Physical Layer, Not the Software
Before you even power on the spectrum analyzer, you need to verify the physical signal path. I’ve seen too many technicians chase a “missing carrier” only to find a corroded F-connector or a waterlogged LNB. Satellite signals are incredibly weak after their 36,000 km journey. A 0.5 dB loss from a poorly crimped connector or a kinked cable can be the difference between a locked transponder and a blank screen.
Check every connection. Use a torque wrench for C-band or Ku-band waveguide flanges if you’re working on a larger terminal. For smaller consumer or VSAT systems, replace any compression connector that shows signs of oxidation. A simple DC resistance check from the indoor unit to the LNB can reveal a lot. You should see a short circuit at DC (the LNB’s internal regulator path) but a high impedance at RF. If you see a few hundred ohms, you’ve got moisture in the line.
2. Choosing and Configuring Your Spectrum Analyzer
You don’t need a lab-grade Rohde & Schwarz for field work, but you do need an instrument that covers the IF or L-band range you’re working with. For most Ku-band VSAT work, that’s 950–2150 MHz. A handheld unit like the Viavi or a reliable USB spectrum analyzer paired with a rugged laptop is the standard. The key specifications to check before you head out are the noise floor (you want at least -100 dBm typical) and the resolution bandwidth (RBW) filters. A 10 kHz RBW is the minimum for identifying narrow carriers; 1 kHz or 300 Hz is better for spotting low-symbol-rate beacons.
Set your reference level carefully. If you’re measuring directly at the LNB output, start with a reference level of -20 dBm and a scale of 10 dB/div. You can adjust down once you see the noise floor. The most common mistake is setting the reference level too low, which compresses the display and hides the true noise floor. The second most common mistake is leaving the input attenuator at 0 dB. A small amount of internal attenuation (5–10 dB) often improves the analyzer’s own noise figure and prevents front-end overload from strong terrestrial interference.
3. Understanding the Noise Floor and System Temperature
A spectrum analyzer shows you power, but what you really care about is carrier-to-noise ratio (C/N). To get that, you need a reliable noise floor reference. Point the dish at cold sky—an area with no known satellites—and observe the baseline. This is your system noise floor, which is a combination of the LNB noise figure, the cable loss, and the analyzer’s own noise. A typical Ku-band LNB with a 0.7 dB noise figure should show a noise floor around -95 to -100 dBm in a 10 kHz RBW, assuming minimal cable loss.
If your noise floor is sitting at -85 dBm, something is wrong. It could be a faulty LNB, terrestrial interference, or a dish pointing at a warm object like a tree or a building. Remember, the noise power is proportional to the system temperature. A tree has a physical temperature of about 290 Kelvin, which is much “hotter” than the cold sky (often below 30 Kelvin at Ku-band). Pointing at a tree can raise your noise floor by 10 dB or more, completely masking weak carriers.
4. Capturing a Transponder: More Than Just a Hump
Once you’ve verified a healthy noise floor, it’s time to find your satellite. Use a known active transponder frequency. For Ku-band, a good reference is a strong DTH broadcast transponder with a wide carrier. Slowly pan the dish across the orbital arc. You’re not just looking for a peak; you’re looking for the correct signature. A digital carrier has a characteristic flat-top shape with steep roll-off at the edges, matching the transponder’s bandwidth (typically 27, 33, 36, or 54 MHz). An analogue carrier or a beacon will look like a narrow spike.
If you see a hump that’s rounded and mushy, you might be looking at a noise source or a terrestrial interference signal. Cross-polarization can also cause a second, weaker copy of the same carrier to appear at a slightly different frequency if you’re not peaked correctly. Use the spectrum analyzer’s marker function to measure the carrier’s center frequency and bandwidth. Compare it to the known satellite parameters. A frequency error of more than a few hundred kHz at Ku-band usually means you’re on the wrong satellite or your LNB’s local oscillator has drifted.
4.1 Identifying Interference and Spurs
Interference hunting is where spectrum analysis gets interesting. You’ll often see narrow spikes that don’t belong to any known carrier. These can be internal to your setup (a switching power supply radiating noise) or external (a nearby radar, a faulty DECT phone base station, or even a microwave oven). To isolate the source, power down everything at your site except the spectrum analyzer and LNB. If the spike disappears, it’s self-inflicted. If it remains, use a directional antenna to triangulate. I once tracked a persistent L-band spike to a faulty LED driver in a building 200 meters away. The owner was not pleased when I knocked on his door, but the interference disappeared after he replaced the unit.
5. Measuring Carrier-to-Noise Ratio (C/N) Correctly
Modern spectrum analyzers have a built-in C/N measurement function, but it’s often misapplied. The analyzer measures the power in the carrier’s bandwidth and compares it to the power in a reference bandwidth. If you set the reference bandwidth incorrectly, your reading is meaningless. For a DVB-S2 carrier, the relevant noise bandwidth is the symbol rate, not the occupied bandwidth. A 27.5 Msym/s carrier occupies about 33 MHz, but the noise power should be measured in 27.5 MHz. The analyzer doesn’t know this unless you tell it.
Manually, you can measure C/N by using the marker function to read the carrier peak power, then measure the noise floor in a clear part of the spectrum. Apply the formula: C/N = Carrier Power – Noise Power + 10*log10(Signal Bandwidth / RBW). If your resolution bandwidth (RBW) is 10 kHz and your signal bandwidth is 27.5 MHz, the correction factor is 10*log10(27.5e6 / 10e3) = 34.4 dB. So if your carrier peak is -30 dBm and your noise floor is -80 dBm in 10 kHz RBW, your C/N is -30 – (-80) + 34.4 = 84.4 dB. That’s a healthy signal. If you forget the correction factor, you’ll think your C/N is only 50 dB and you’ll waste hours trying to “fix” a perfectly good link.
6. The Role of the Spectrum Analyzer’s RBW and VBW
Resolution bandwidth (RBW) and video bandwidth (VBW) are not just abstract settings; they directly affect what you see. A wide RBW will smear narrow carriers and hide low-level interference. A narrow RBW will slow down the sweep and can make a wideband carrier look like noise. For initial satellite identification, use a 100 kHz or 300 kHz RBW. For detailed transponder analysis, switch to 10 kHz or 30 kHz. For beacon acquisition, drop to 1 kHz or even 300 Hz.
VBW is a smoothing filter. Set it to 1/10th of the RBW for a clean trace, or set it equal to RBW for a faster sweep. When hunting for intermittent interference, set the analyzer to “max hold” and let it run for several minutes. This will capture transient spikes that you’d miss on a live sweep. I once found a radar pulse that appeared only every 12 seconds—the sweep time was 2 seconds, so it was invisible until I used max hold.
7. Polarization and Cross-Pol Isolation
Satellite transponders reuse frequencies by transmitting on orthogonal polarizations—horizontal and vertical, or left-hand and right-hand circular. Your system’s ability to reject the opposite polarization is called cross-polarization isolation (XPI). Poor XPI causes interference from the opposite polarization, which can look like a ghost carrier overlapping your desired signal. To measure XPI, tune to a known unmodulated carrier or a clear section of the transponder. Note the level on your desired polarization. Then switch the LNB to the opposite polarization (or rotate the feed if you’re using a motorized polarizer). The level should drop by at least 27 dB for a typical Ku-band system. If it’s less, your feed is misaligned, your polarizer is faulty, or the dish is warped.
On a spectrum analyzer, poor XPI often appears as a “bump” on the noise floor that rises and falls as you adjust the polarization. It’s a subtle effect, but once you’ve seen it, you’ll recognize it instantly. Correcting it can improve your C/N by several dB—often the difference between a link that works in clear sky and one that fades out in moderate rain.
8. Documenting the Spectrum
A spectrum capture without context is just a pretty picture. Every time you save a trace, annotate it with the site location, dish size, LNB type, cable length, weather conditions, and the satellite and transponder details. I use a simple naming convention: SiteID_Satellite_Polarization_DateTime. This makes it easy to compare traces over time. A gradual increase in the noise floor over weeks can indicate a failing LNB or water ingress. A sudden appearance of a new carrier could be a misaligned dish on a neighboring satellite.
If you’re using a PC-based analyzer, save the trace data in a format that allows offline analysis. Many engineers just take screenshots, but that’s like taking a photo of an oscilloscope—you lose the ability to zoom in and measure later. Export the raw data points whenever possible. I’ve solved more than one intermittent problem by loading a week’s worth of traces into a spreadsheet and plotting the noise floor over time.
9. Common Pitfalls and Field-Expedient Fixes
Here’s a short list of problems I’ve encountered more than once, and the quick checks that can save you hours:
- No signal, but the analyzer shows a healthy noise floor. Check the LNB power. Many spectrum analyzers don’t provide DC power to the LNB. You need a bias tee or a separate power inserter. Also, verify the LNB local oscillator frequency. A Universal Ku LNB uses 9.75 GHz or 10.6 GHz; if your analyzer is set for the wrong LO, the entire spectrum will be shifted by hundreds of MHz.
- Signal looks “smeared” or has a tilted noise floor. This is often a sign of a failing LNB or a poor connection causing an impedance mismatch. The LNB’s output return loss degrades, creating ripples in the passband. Replace the LNB or check the connectors.
- Intermittent spikes that move in frequency. These are almost always local interference. Use a battery-powered spectrum analyzer to rule out ground loops. If the spikes disappear when you run on battery, you have a grounding issue.
- Carrier level fluctuates rapidly. This could be scintillation (atmospheric effect, common at low elevation angles) or a dish that’s physically oscillating in the wind. Check the mount rigidity. On a calm day, scintillation at Ku-band is usually less than 1 dB peak-to-peak.
10. Beyond the Spectrum: Constellation and MER
Once you’ve optimized the physical layer and the spectrum looks clean, the next step is to look at the modulation quality. A spectrum analyzer shows you the aggregate power, but it doesn’t tell you if the phase noise is destroying your bit error rate. For that, you need a vector signal analyzer or a dedicated satellite modem that reports modulation error ratio (MER). A high C/N with a low MER indicates phase noise, usually from a poor-quality LNB or a drifting local oscillator. I’ve seen LNBs that produce a beautiful spectrum but have such bad phase noise that an 8PSK carrier won’t lock. The spectrum analyzer can’t see that; you need to look at the constellation.
If you don’t have a vector analyzer, use the modem’s built-in diagnostics. Most modern satellite routers report MER in real time. Compare the MER to the theoretical value for your C/N. For DVB-S2 8PSK with a 3/4 FEC rate, you need about 7.9 dB C/N for a quasi-error-free link, which corresponds to an MER of roughly 10 dB. If your C/N is 12 dB but your MER is only 9 dB, you’ve got an impairment that the spectrum analyzer isn’t showing you.
Frequently Asked Questions
Why does my spectrum analyzer show a carrier, but my modem won’t lock?
Several reasons. First, verify you’re on the correct satellite and polarization. A spectrum analyzer can’t decode the satellite’s network identification. Second, check the symbol rate and FEC settings in the modem. Third, look at the carrier’s shape. If it’s not a clean flat-top with steep edges, you might be seeing an analogue carrier or a data carrier with a different roll-off factor. Finally, phase noise or excessive jitter can prevent a lock even with a strong C/N. You’ll need a vector analyzer or the modem’s MER reading to diagnose that.
How do I know if my LNB’s local oscillator has drifted?
Tune to a known reference carrier with a precise frequency, such as a satellite beacon. Note the frequency your analyzer displays. If it’s off by more than a few hundred kHz, your LNB’s LO has likely drifted. Temperature changes can cause drift, so measure when the LNB is at its normal operating temperature. Some drift is correctable by adjusting the modem’s receive frequency, but excessive drift (over 1 MHz) usually means the LNB is failing and should be replaced.
Can I use a software-defined radio (SDR) instead of a spectrum analyzer?
Yes, with caveats. An SDR like an RTL-SDR or an Airspy can display the L-band spectrum, but its noise figure, dynamic range, and frequency accuracy are far inferior to a proper spectrum analyzer. It’s fine for identifying strong carriers and checking for gross interference, but don’t trust it for precise C/N measurements or for detecting low-level spurs. If you’re doing professional installations, invest in a calibrated instrument. The time you save in troubleshooting will pay for it quickly.