Two Transmitters, One Mast, and a Dead Receiver at 455 MHz
A water-tower site carries two 25 W PMR base transmitters on 450.000 MHz and 452.500 MHz, feeding separate vertically polarized colinear antennas 2 m apart on the same mast, plus a utility telemetry receiver on 455.000 MHz with a 12.5 kHz bandwidth. Since the second transmitter was commissioned, the telemetry link reports a raised noise floor and corrupted frames whenever both transmitters are keyed; keying either one alone causes no problem. The site has clear line of sight and the receive antenna is on the same mast, roughly 3 m below the transmitters. With the receiver disconnected, a measurement at the receiver antenna port shows -12 dBm at 450.000 MHz and -14 dBm at 452.500 MHz. The receiver front end has no preselector, an input third-order intercept of +5 dBm and an input 1 dB compression point of -20 dBm, and the wanted telemetry signal arrives at about -90 dBm.
1. Identify the interference mechanism and show why it is present only when both transmitters are keyed.
Read the worked answer
Frequency arithmetic. The third-order product 2f2 - f1 = 2(452.500) - 450.000 = 455.000 MHz, exactly on the telemetry channel. This is a two-transmitter product: with only one carrier present there is nothing to mix against, which is precisely why single-transmitter keying is clean. The single-carrier tests are themselves the diagnostic that separates intermodulation from blocking, spurious emission or a receiver fault.
Other products in the same family. 2f1 - f2 = 447.500 MHz; 3f1 - 2f2 = 445.000 MHz; 3f2 - 2f1 = 457.500 MHz. None of the transmitter harmonics (900, 905, 1350, 1357.5 MHz) falls near 455 MHz, and the products listed are the ones to check against other receivers on the site.
Bandwidth. A third-order product is wider than its parents: roughly 2 x B2 + B1, so two 12.5 kHz signals produce a product about 37 kHz wide, which fully covers the 12.5 kHz telemetry channel rather than clipping its edge. That is consistent with constant corruption rather than occasional hits.
Where the mixing happens. Three candidates: the transmitter output stages or the combining network, the receiver front end, or a nonlinear junction on the mast (corroded hardware, loose clamps, dissimilar metals). A dummy-load substitution test separates them: replace the transmitter antennas with loads; if the interference at the receiver disappears, the mixing is external to the transmitters, in the mast hardware or the shared near field. If it persists, it is inside the transmitter chain or the receiver front end.
2. Quantify the intermodulation level at the receiver input and explain why front-end filtering is so effective here.
Read the worked answer
Third-order level. Using P_IMD = P1 + P2 + min(P1, P2) - 2 x IIP3 with the carriers at the receiver port: P_IMD = (-12) + (-14) + (-14) - 2 x (+5) = -50 dBm.
Compare with the wanted signal. The telemetry arrives at about -90 dBm, so the product sits 40 dB above the wanted carrier. No amount of baseband processing recovers that; the receiver is being operated far outside its linear range even before compression is considered. The carriers themselves are also only 6 dB below the -20 dBm input compression point, so gain compression and cross-modulation are adding to the damage.
Why filtering wins. Third-order products fall 3 dB for every 1 dB reduction in carrier level. Reducing both carriers by 20 dB (a cavity filter plus better antenna separation) lowers the product by 60 dB, to about -110 dBm, which is 20 dB below the wanted signal. A 40 dB reduction gives 120 dB of improvement, i.e. the product disappears entirely. This cubic slope is why the correct fix is to attack the coupling, not to buy a more linear receiver: an IIP3 improvement large enough to give 40 dB would be far more expensive and would not remove the compression problem.
The lesson to record in the site file: co-sited transmitters must be assessed for third-order products at the planning stage, because the product frequency is fixed by the two transmit frequencies and cannot be filtered out at the receiver once it is generated in the shared near field.
3. Recommend and rank mitigation measures, and state how you would verify each one.
Read the worked answer
1. Reduce coupling first (cheapest, largest effect). Increase separation between the transmit and receive antennas - both vertical (several metres, ideally more than a wavelength-scale multiple at 455 MHz) and horizontal, and use cross-polarization between transmit and receive paths. Verify by repeating the port-level measurement: a 20 dB drop is a 60 dB drop in the product.
2. Filter at the receiver input. A cavity bandpass on 455 MHz with a notch or high rejection at 450-452.5 MHz protects the front end. Filtering after the first amplifier is too late. Verify by measuring the carrier levels at the receiver port and re-checking the telemetry with both transmitters keyed.
3. Isolate the transmitters. Circulators plus cavity filters on each transmitter output, or a proper transmit combiner, prevent transmitter-generated products and reduce the energy leaving the mast.
4. Repair external nonlinearities. If the dummy-load test showed external mixing, clean and replace corroded hardware, use stainless or bimetallic-safe fittings, and re-torque clamps.
5. Change the receive frequency as a last resort. Moving the telemetry channel by 100 kHz clears a 37 kHz-wide product, but the product position moves with the transmit frequencies, so any future retune must repeat the calculation, and the new channel must be coordinated.
Verification. After each step, run a two-tone test into the antenna system at controlled levels, measure the product at 455 MHz with a spectrum analyser, and confirm with live traffic. Record the carrier levels, the product level and the resulting C/I in the site file so later changes can be assessed against a baseline.