A 6 GHz carrier showing up in a 3.7–4.2 GHz receive chain is not a mystery. It is an isolation budget that came out negative. The uplink and downlink bands are separate allocations — C-band uplink sits at 5.925–6.425 GHz and downlink at 3.7–4.2 GHz (Anritsu, Resolving Interference Issues at Satellite Ground Stations) — so any 6 GHz energy arriving at the LNB input is leakage across a shared path, not a legitimate downlink signal. The useful question is not “which component is bad?” but “how much isolation does this path actually have, and where does the measured level exceed what the budget allows?”
This article lays out a budget-first method: enumerate every coupling path from the transmit port to the receive port, sum them against the receive chain’s overload threshold, then use the receive chain’s own non-linearity as the leak detector. The finding is the gap between predicted and measured isolation. Everything else is guessing with extra steps.
Why the two bands share a path at all
In a typical C-band earth station the feed network is a duplexed assembly. The orthomode transducer (OMT) is a waveguide component that “serves as a polarized duplexer” and is used “to separate transmit and receive signals so they propagate through different ports” (Anritsu). That separation is finite. So is every other junction between the BUC output and the LNA input.
Where a single antenna serves both directions through a shared waveguide run, a circulator may be part of the routing. A waveguide circulator is a passive, non-reciprocal three-port ferrite device: signal entering any port exits almost entirely from the next port in a fixed rotational direction, “while the port behind it sees very little of that same signal” (Mi-Wave, What Is a Waveguide Circulator?). “Very little” is the operative phrase. Typical circulator isolation lands between 15 and 20+ dB minimum (Mi-Wave). That is a component specification, not a system guarantee, and it is one term in a sum.
The architecture is correct. The budget is what determines whether correct architecture is sufficient.
The isolation budget, term by term
Write the budget as a sum of coupling paths from the transmit port to the receive port, each expressed as an isolation in dB. The paths that matter in a C-band feed:
- OMT port-to-port isolation. The duplexer’s transmit-to-receive rejection at the operating frequency. This is the first and usually largest term.
- Circulator reverse isolation, if present. The backward coupling through the ferrite junction, specified per device. If two circulators are cascaded, add their isolations in dB — but only if the geometry actually cascades them.
- Waveguide flange joints. Every mating surface between the transmit-side hardware and the receive-side hardware is a potential discontinuity. Mi-Wave notes that flange mating surface flatness “directly affects insertion loss and VSWR at the joint” and that “any gap or misalignment introduces a discontinuity the signal has to cross.” A discontinuity is bidirectional: it can radiate outward and admit energy inward.
- Feed radiated coupling. Energy launched from the transmit horn that reaches the receive horn by paths other than the intended duplexed route — spillover, strut scattering, reflector backlobe. This term is geometry-dependent and is not a datasheet number.
- Cabinet, cable, and connector leakage. The IFL runs, the BUC-to-waveguide transition, the LNB input connector. Small terms individually; they add.
Sum the isolations as a cascade: the total TX-to-RX isolation is the parallel combination of all paths, which for dB values means converting each to a linear power ratio, summing the ratios, and converting back. In practice, the dominant path sets the floor and the rest contribute a fraction of a dB. State the uncertainty explicitly: a flange joint whose condition you have not inspected is not a known quantity, and a cable calibration from last month’s sweep carries its own tolerance, plus or minus 0.5 dB or whatever the sweep certificate says. Do not round the total in the direction that flatters the result.
The receive chain’s overload threshold is the other half of the budget
Isolation alone tells you nothing. You need the level the receive chain can tolerate before it misbehaves, and that threshold is lower than most people assume.
Anritsu’s application note states that satellite LNA/LNBs are optimized for very low-level signals and that “typically, the LNA/LNB will be saturated with a total input power of approximately -50 dBm, depending on equipment used,” with non-linear behavior beginning “at about -60 dBm input into the receiver system.” The same note cites ITU-R S.2199-0, which “recommends a maximum power into the LNB of no higher than -60 dBm.”
So the budget has a hard ceiling: total power arriving at the LNB input from all sources, wanted and unwanted, should stay at or below roughly -60 dBm. That is the number to design against, and it is a recommendation from a cited ITU-R study, not a rule of thumb. Your LNB’s actual compression point may differ; measure it or get it from the manufacturer, and carry the uncertainty.
Now the arithmetic. Suppose the BUC delivers +40 dBm at the waveguide flange. Suppose the summed TX-to-RX isolation across all paths is 90 dB. The leakage at the LNB input is +40 − 90 = −50 dBm. That is above the −60 dBm recommendation by 10 dB, and at the saturation figure Anritsu quotes. The budget is negative before anyone has touched a spectrum analyzer. If the summed isolation is 110 dB, leakage is −70 dBm and the path is comfortable with 10 dB of margin. The difference between those two stations is not a component — it is the sum of joints, orientations, and geometry.
Why the leak often shows up as distortion, not a tone
A leaking carrier 20 dB below the wanted downlink may be invisible as a discrete spectral line on a wide span. It is not invisible to the LNA. Because the receive chain begins non-linear behavior well below saturation, a leaking 6 GHz carrier can compress the wanted signal, shift the noise floor, or generate intermodulation products before it is obvious as a tone. Anritsu describes exactly this mechanism for out-of-band interferers: signals from nearby transmitters “can overload the LNA at the ground station,” and because LNAs are optimized for gain, “satellite signals can be easily compressed by the LNA due to other signals present in the LNA bandwidth.”
This is the diagnostic lever. The receive chain is a more sensitive leak detector than a spectrum search, because it responds to total power, not just to resolvable lines. A C/N0 degradation on the wanted carrier that tracks the transmit carrier’s on/off state is a measurement of leakage, even when the leakage itself is below the analyzer’s noise floor at that resolution bandwidth.
The bisection sequence
Use the budget as a prior, then falsify it segment by segment. The order matters because each step narrows the plausible location.
- Measure at the LNB input. Terminate or bypass as needed and measure the 6 GHz energy present at the receive port with the transmit carrier keyed and unkeyed. The difference is the leakage level reaching that point. Compare it to the budget prediction. If measured exceeds predicted, a path in the budget is worse than assumed.
- Test the feed first if the OMT term alone can account for the measured level. If the predicted leakage from the duplexer’s specified isolation matches the measurement, the feed is the first place to verify — but “matches” means within the stated uncertainty, not within a convenient rounding.
- If the OMT cannot account for it, move downstream. The leak is in the waveguide joints, the circulator orientation, or the receive-side hardware. Inspect each flange for flatness and full mating. Mi-Wave’s point about flange surface flatness is not cosmetic: a gap is a discontinuity, and a discontinuity is a coupling path.
- Check non-reciprocal components for orientation. A circulator’s ports are not interchangeable, and “installing it backward routes signal the wrong way rather than simply reducing performance” (Mi-Wave). A reversed circulator does not degrade gracefully; it creates a path that the schematic says cannot exist.
- Check BPF retrofits. If a bandpass filter was inserted into the receive path, its mating is suspect. Anritsu notes that inserting a BPF “can cause interference issues if the connections are not properly mated,” and that problems occur “when the materials used by the BPF and LNB do not provide a good seal.” Legacy LNBs in particular “may not have a standard waveguide flange which is fully metallic thus causing radiation egress and ingress.” A non-metallic flange is a leak path that no amount of torque will close.
Each step either confirms the budget or localizes the excess. The sequence terminates when the measured leakage is explained by the sum of the paths you have verified — not when you have found a component that looks suspicious.
Installation faults that break a correct budget
Three failure modes account for most cases where the schematic is right and the station is wrong:
- Reversed non-reciprocal components. Covered above. The budget assumed a direction; the hardware has the other one.
- Poorly mated flanges. A joint that is mechanically tight but not flat, or that has a gasket or coating between mating surfaces, is a discontinuity. The insertion loss penalty is the least of the problems; the coupling path is the problem.
- Non-metallic or non-standard flanges on legacy hardware. The BPF retrofit case is the common one, but any legacy LNB with a non-metallic flange is a candidate. Radiation egress and ingress are the same path in both directions.
None of these are performance degradations in the ordinary sense. They are budget-breaking faults, and they are invisible to anyone reading the block diagram.
Reconciliation discipline
The finding is the gap between predicted and measured isolation. A leak that matches the budget is expected coupling — it is the design working as specified, and the fix is to change the design, not to hunt for a fault. A leak that exceeds the budget is the fault, and the bisection sequence localizes it. Keep the two categories separate. The temptation, when a 6 GHz carrier appears where it should not, is to treat all leakage as abnormal and start replacing components. That approach finds nothing reliably and costs a lot of waveguide hardware.
Two bookkeeping rules keep the reconciliation honest. First, carry the uncertainty on every term: the flange condition you have not inspected, the cable calibration from last month’s sweep, the manufacturer’s isolation figure at a frequency other than your carrier. Second, never round the total in the direction that makes the margin look better. A budget that closes by 0.3 dB after favorable rounding does not close.
The receive chain told you the carrier was there. The budget tells you how much isolation you have. The measurement tells you which of those two numbers is wrong. That is the whole method, and it works because it treats the symptom as a number to be reconciled rather than a fault to be guessed at.
Frequently asked questions
Is a 6 GHz carrier in a 4 GHz receive chain always leakage?
Yes, in the sense that matters here. C-band uplink and downlink are separate allocations (5.925–6.425 GHz and 3.7–4.2 GHz per Anritsu), so 6 GHz energy at the LNB input cannot be a legitimate downlink signal. It arrived by a coupling path from the transmit side or from an external source. The budget method distinguishes the two by keying the transmit carrier and observing the difference.
What isolation figure should I design for?
There is no single number, and any source that gives you one without your BUC output power and LNB threshold is guessing. Work backward from the −60 dBm recommendation cited by Anritsu (from ITU-R S.2199-0): required isolation equals transmit power at the flange minus −60 dBm, plus whatever margin your uncertainty analysis supports. A +40 dBm BUC needs at least 100 dB of summed isolation to reach −60 dBm with no margin at all.
Can a spectrum analyzer see the leak directly?
Sometimes, but do not rely on it. A leak 20 dB below the wanted carrier may be below the analyzer’s noise floor at the resolution bandwidth you are using, while still compressing the LNA. Measure the wanted carrier’s C/N0 with the transmit carrier keyed and unkeyed. The difference is a more sensitive indicator than a spectral line search.
What does ITU-R SM.329 have to do with this?
SM.329 is the ITU-R recommendation on unwanted emissions in the spurious domain, in force as SM.329-13 (09/2024) per the ITU publication page. It governs transmitter emission limits, not the isolation of a receive path. It is relevant context for what the transmitter is allowed to emit, but it does not set the TX-to-RX isolation budget for your station. Do not cite it as if it did.
Where do intermodulation products fit in?
They are a separate mechanism that can mimic leakage. Anritsu notes that when more than one signal is transmitted by a power amplifier or by multiple transmitters, mixing produces intermodulation products whose power depends on the relative level of each signal. A product that lands in the receive band is not leakage across the duplexer — it is generated in the transmit chain. Keying tests that change the number of carriers, not just their power, distinguish the two.