If you’ve ever stood on a windswept rooftop staring at a constellation diagram and felt that sinking ‘something isn’t right’ feeling, you’re in good company. I’ve spent more than twenty years aligning dishes and chasing down elusive link problems, and I can tell you this: the real difference between a button-pushing installer and a proper satellite engineer comes down to how well they read a spectrum. It’s not about letting the auto-setup routine do the work. It’s about seeing what the carrier is actually doing before you even try to lock a modem onto it.
Most field techs I’ve trained lean way too hard on automated routines. They’ll nudge a dish around until a green light blinks on the modem, then pack up and leave. Six months later, the link collapses in a downpour because nobody ever checked the cross-pol isolation. A solid spectrum analysis catches those hidden margins. It’s a physical layer investigation, plain and simple. Let me walk you through the exact field process I use, from hooking up the gear to making sense of the noise floor.
Setting Up Your Measurement Chain
Before you look at a single trace, make sure your measurement setup isn’t feeding you lies. I’ve watched technicians chase phantom faults for hours because of a dodgy adapter or an overdriven input. A spectrum analyzer is a sensitive beast, and its front end is easy to damage or saturate.
Power Budget and DC Blocking
Start with a basic power budget. If you’re tapping straight into the LNB output, you’re dealing with an IF signal, usually in the L-band (950–2150 MHz). Total power here can be surprisingly high on a multi-carrier trunk. Most analyzers have a maximum safe input around +30 dBm (1 Watt), but the mixer’s damage threshold is often lower—closer to +20 dBm. For clean measurements, though, you’ll want total integrated power below -10 dBm to steer clear of compression.
Always stick a DC block in line if you’re not 100% sure the line isn’t powering the LNB. A lot of modern analyzers have an internal DC block option, but a physical external one is cheap peace of mind. I keep a Mini-Circuits BLK-89-S+ in my kit for exactly this. It lets the IF through cleanly while stopping that 13/18V DC from cooking the analyzer’s input stage.
Calibration and Reference Levels
Don’t take the factory calibration as gospel. Before you connect to the dish, hook the analyzer’s tracking generator (if it has one) or a known reference signal to your test cable. Normalize out the cable loss. A 20-metre run of RG6 can easily eat 4 dB at 2150 MHz. If you don’t subtract that, your C/N numbers will look worse than they are, and you might end up repointing a dish that was fine all along.
Set your reference level with a bit of thought. I usually start with a wide span—say 500 MHz—and a reference level of -20 dBm. If the trace looks flat-topped against the top of the screen, you’re overloading the input. Back off the input attenuation or bump up the reference level until the real shape of the carriers emerges.
Capturing the Wideband Transponder View
With the analyzer connected to the LNB output, your first scan is a wideband view. Think of it as the satellite’s fingerprint. You’re looking for the characteristic slope of the transponders, the noise floor between them, and any hint of interference or spillover from an adjacent bird.
Identifying the Correct Orbital Slot
Don’t just assume you’re locked onto the right satellite. Each orbital slot has its own spectral signature. Astra 28.2°E, for instance, has a distinct mix of narrow and wide transponders across the band. Compare what you see on the analyzer to a known reference plot. I keep a library of screenshots on my phone for the slots I visit regularly. If the pattern is shifted by 20 MHz or the transponder spacing looks off, you’re probably sitting on the adjacent satellite. A few degrees off-axis makes a massive difference, especially with smaller dishes.
Checking Transponder Power Flatness
Once you’ve confirmed the satellite, look at the amplitude flatness across the transponder. A healthy one should have a reasonably flat top, with sharp roll-off at the edges. If you spot a significant tilt, it might be a mispointed dish, a flaky LNB with poor gain flatness, or even a transponder fault. Don’t jump to conclusions—rotate the feed slightly and see if the tilt changes. If it does, the problem is likely your feed alignment or polarization skew.
Zooming In: Carrier Under Test
Now narrow the span to the specific carrier you’re commissioning or troubleshooting. Set the span to about twice the symbol rate. For a 27.5 Msym DVB-S2 carrier, a 50–60 MHz span is a good starting point. Set the resolution bandwidth (RBW) to 100 kHz or lower, and the video bandwidth (VBW) to 1 kHz or less to smooth the trace without burying the details.
Measuring Occupied Bandwidth and Shoulders
The occupied bandwidth should match the carrier’s nominal bandwidth. A 27.5 Msym carrier with a roll-off factor of 0.20 should occupy roughly 33 MHz. Use the analyzer’s occupied bandwidth measurement function, or just use markers. Look at the shoulders of the carrier—the transition from the flat top to the noise floor. They should be symmetrical and steep. If one shoulder slopes more gradually, you might have a filter issue in the transmission chain or, more likely, a non-linear amplifier creating spectral regrowth.
I once burned an entire afternoon chasing a high BER on a link, only to discover the BUC (Block Upconverter) on the transmit side was being driven 3 dB into compression. The spectrum looked okay at a glance, but the shoulders were soft. Backing off the drive power cleaned up the shoulders and the errors disappeared. The spectrum analyzer told the story long before the modem could.
Measuring C/N and Eb/No
Modern analyzers often have built-in routines for carrier-to-noise ratio (C/N) or even Eb/No estimates. I still prefer to do it manually. It forces you to understand what you’re measuring and keeps you from blindly trusting a number that might be calculated oddly.
The Manual Method
Set your analyzer to a span that shows the whole carrier and some clear noise floor on both sides. Place a marker on the carrier peak and a delta marker on the noise floor. Make sure the noise measurement is taken far enough from the carrier to avoid the shoulders. Then apply a correction factor for the noise bandwidth of the resolution bandwidth filter. Most analyzers have a noise marker function that does this automatically, normalising the noise power to a 1 Hz bandwidth. The difference between the carrier peak and the normalised noise floor is your C/No (carrier-to-noise density ratio).
To get C/N, you need the noise bandwidth of the carrier, which is roughly equal to the symbol rate. So C/N = C/No – 10*log(symbol rate). For a 27.5 Msym carrier, that’s C/No – 74.4 dB. If your analyzer shows a C/No of 85 dBHz, your C/N is 10.6 dB. That’s a tight link, and you’ll need a sturdy FEC to close it.
Polarization and Cross-Pol Isolation
This is where most field techs get lazy. They peak the co-pol signal and call it done. But if you skip the cross-pol isolation measurement, you’re leaving capacity on the table and inviting interference. The procedure is simple but demands a steady hand.
Nulling the Cross-Pol Signal
First, find a continuous carrier on the opposite polarization. If you’re aligning for a vertical carrier, pick a horizontal one. Adjust the feed rotation (skew) while watching the amplitude of that cross-pol carrier. You’re not trying to peak it; you’re trying to null it. Rotate the feed until the cross-pol signal hits its absolute minimum. Then, and only then, check your co-pol carrier. It should be near maximum. A well-aligned dish should give you at least 27 dB of cross-pol isolation. I’ve seen 35 dB on a perfectly tuned 1.8m dish.
If you can’t get decent isolation, look for mechanical problems. A warped reflector, a feed that’s not centred in the focal point, or even a dented feedhorn can wreck cross-pol performance. No amount of electronic equalisation can fix a physically distorted antenna.
Interference Hunting
Interference is the curse of satellite communications. It can be deliberate, accidental, or just a misconfigured carrier on an adjacent transponder. The spectrum analyzer is your main weapon for spotting and classifying it.
Using the Spectrogram
A spectrogram, or waterfall display, is worth its weight in gold. It shows the history of the spectrum over time, with amplitude represented by colour. Set the analyzer to a zero-span mode centred on the carrier frequency, or use a narrow span with the spectrogram enabled. Watch for intermittent spikes, periodic bursts, or a noise floor that slowly creeps upward. I once tracked a 2 AM interference spike to a faulty microwave oven in a break room near the teleport. The spectrogram showed a clear 50 Hz pattern from the oven’s magnetron power supply.
Identifying External Interference
If you see a carrier where it shouldn’t be, don’t immediately point a finger at the adjacent satellite operator. First, check if the signal vanishes when you block the feed. If it sticks around, it’s probably a local interference source coupling into your IF cabling. I’ve seen everything from DECT phones to poorly shielded HDMI cables radiate into satellite IF. Use a small handheld spectrum analyzer with a near-field probe to sniff around the cabling and connectors.
Documenting the Analysis
A measurement without documentation is just a hobby. For every site, I save a set of traces: the full-band view, a zoomed-in carrier view with markers, and a cross-pol null trace. I also note the analyzer settings—RBW, VBW, span, and reference level—so anyone reviewing the data later understands the measurement context. A screenshot without settings is often useless for comparison down the road.
I also record the make and model of the LNB, the dish size, the cable type and length, and the weather conditions. Rain attenuation can skew your measurements by several dB, especially at Ku-band. If you’re measuring a link margin, do it under clear sky and then calculate the expected fade margin based on your location and availability target.
Common Pitfalls and Practical Tips
Over the years, I’ve built a mental checklist of the most frequent blunders. Here are a few that can ruin your day:
- Ignoring the noise figure of the LNB. A cheap LNB with a 1.5 dB noise figure will clobber your G/T compared to a 0.3 dB unit. Always measure the system noise floor with the LNB connected and compare it to the expected value.
- Using too much input attenuation. It protects the analyzer, sure, but excessive attenuation raises the measurement noise floor and hides low-level spurs. Use the minimum attenuation needed to keep total power below the compression point.
- Forgetting to re-tighten connectors. A loose F-connector can cause intermittent ingress that looks like a wandering noise floor. After adjusting the feed, always torque the connectors properly.
- Measuring a carrier that’s in a deep fade. If you’re troubleshooting a link, check the weather at the uplink site. A rain fade at the teleport can masquerade as a local problem.
FAQ
What’s the difference between a spectrum analyzer and a signal analyzer?
A spectrum analyzer measures amplitude versus frequency, giving you the power spectral density. A signal analyzer, or vector signal analyzer, captures IQ data and can demodulate the signal to show constellation diagrams, error vector magnitude, and other modulation quality metrics. For satellite work, a spectrum analyzer is your primary tool for physical layer checks, while a signal analyzer is used for deeper modulation analysis.
How do I know if my LNB is causing a gain slope?
Connect the LNB to a known flat noise source, like a noise diode, and observe the spectrum. If the slope remains, the LNB is the culprit. Alternatively, compare the transponder power levels across the band to a reference measurement taken with a calibrated LNB. A consistent tilt across all transponders points to the LNB or the cabling, not the satellite.
Why does my carrier look wider than expected on the analyzer?
This is often down to the resolution bandwidth (RBW) setting. If the RBW is too wide relative to the symbol rate, the analyzer will smear the carrier shape. Set the RBW to about 1% of the symbol rate for a clear view. If the carrier still appears wider, you might be seeing spectral regrowth from amplifier compression or excessive phase noise on the uplink.
Can I measure group delay with a spectrum analyzer?
Not directly. A spectrum analyzer measures magnitude only. To measure group delay, you need a vector network analyzer or a specialised group delay measurement setup. However, you can infer group delay issues by looking at the constellation diagram on a signal analyzer—a smeared or curved constellation often indicates group delay distortion.
Proper spectrum analysis is a discipline. It takes patience, a methodical approach, and a healthy dose of scepticism about what the instrument is telling you. The analyzer is just a tool; your grasp of the satellite link budget and the physics of the signal path is what makes the measurement meaningful. Next time you’re on a roof, spend an extra ten minutes running a full spectral check. Your bit error rate will thank you.