Front-End Ordering in a 162.5 MHz Base Receiver
A systems integrator is building a VHF land-mobile base receiver on 162.5 MHz that must work within 2 km of a 1 kW FM broadcast mast and a high-power paging transmitter on 152 MHz. The candidate front end is a four-cavity preselector with 3.5 dB insertion loss, a low-noise amplifier with 1.0 dB noise figure and 20 dB gain, an image filter with 1.5 dB loss, and a mixer with 7 dB noise figure. In one prototype the preselector sits ahead of the LNA; in the other the LNA is first. The LNA has 20 dB gain, an output 1 dB compression point of -10 dBm and an output third-order intercept of +25 dBm; the mixer's input compression point is -2 dBm. The integrator wants the best sensitivity that still survives the local broadcast and paging environment, and must defend the choice to a customer who will measure noise figure at the antenna port with a calibrated noise source.
1. Compute the cascaded noise figure for both front-end orderings and quantify the sensitivity difference.
Read the worked answer
Use Friis in linear terms: F = F1 + (F2-1)/G1 + (F3-1)/(G1 G2) + ...
Preselector first. F1 = 3.5 dB = 2.239, G1 = 0.447; LNA F2 = 1.259, G2 = 100; image filter F3 = 1.413, G3 = 0.708; mixer F4 = 5.012.
- Term 1: 2.239
- Term 2: (1.259-1)/0.447 = 0.579
- Term 3: (1.413-1)/(0.447 x 100) = 0.009
- Term 4: (5.012-1)/(0.447 x 100 x 0.708) = 0.127 F = 2.954, so NF = 4.7 dB.
LNA first. F1 = 1.259, G1 = 100; preselector F2 = 2.239; image filter F3 = 1.413; mixer F4 = 5.012.
- 1.259 + (2.239-1)/100 = 0.012 + (1.413-1)/44.7 = 0.009 + (5.012-1)/31.7 = 0.127 F = 1.408, so NF = 1.5 dB.
The ordering is worth about 3.2 dB of sensitivity, because in the preselector-first case the 3.5 dB of loss is injected ahead of all the gain and adds directly to the system noise figure. Note also that the mixer term, 0.13, is identical in both and is set by the LNA gain; raising LNA gain further would buy little.
2. Using the blocking environment described, decide whether the LNA-first configuration can be shipped as drawn, and propose a front end that keeps most of the noise-figure gain.
Read the worked answer
Blocker level at the LNA input. Free-space loss at 2 km, 100 MHz is 32.45 + 20log(100) + 20log(2) = 78.5 dB. A 1 kW ERP mast gives 60 dBm, so with a 0 dBi receive antenna at 100 MHz the LNA input sees about -18.5 dBm.
Compare with the LNA. The quoted output compression of -10 dBm with 20 dB gain means an input compression point of -30 dBm. A -18.5 dBm blocker is 11.5 dB above it: the LNA is driven hard into compression, its gain collapses for the wanted signal, the noise figure degrades by many dB, and cross-modulation transfers the broadcast modulation onto 162.5 MHz. The mixer's -2 dBm input compression would also be approached once 20 dB of LNA gain is applied to any residual. Shipping LNA-first as drawn is not defensible.
Split the filtering. Put a small, low-loss bandpass (about 1 dB, 20-25 dB rejection at 100 MHz) between the antenna and the LNA, and move the four-cavity preselector to after the LNA, where its 3.5 dB loss costs almost nothing because it is divided by the preceding gain.
Cascade for that arrangement: F = 1.259 + 0.259/0.794 + 1.239/79.4 + 0.413/35.5 + 4.012/25.1 = 1.772, i.e. NF = 2.5 dB, only 1 dB worse than the unprotected LNA-first figure but with the blocker reduced to about -43 dBm, roughly 13 dB below the LNA's input compression point. That is the engineering trade to present to the customer.
3. The built receiver measures 5.8 dB noise figure at the antenna port against the predicted 2.5 dB. List the checks you would run, in order, and what each result would tell you.
Read the worked answer
A 3.3 dB excess is too large to be measurement scatter, so work upstream-to-downstream and separate measurement error from hardware fault.
- Validate the measurement set-up first. Confirm the noise source ENR and calibration frequency, correct for the source's own match, and verify the analyser/meter is in noise-figure mode with the correct bandwidth and averaging. A wrong ENR entry or an uncorrected preamp in the meter can easily produce 3 dB of error. Cross-check with a gain method (Y-factor plus measured gain) before touching hardware.
- Measure each block in isolation. LNA alone should give 1.0 dB; the 1 dB input filter should give 1.0 dB; the four-cavity preselector should give 3.5 dB. Any block reading 2-3 dB high points at a damaged connector, a cracked cavity, moisture ingress, or a filter that has drifted.
- Check the assembly between antenna port and LNA. A damaged jumper, a wrong-length cable, an extra lightning arrestor, or a DC feed failure that leaves the LNA unpowered will each show up as loss ahead of the gain, which adds dB-for-dB to system noise figure. Measure the cable and connector loss at 162.5 MHz, not at DC.
- Check mismatch and bias. Verify LNA bias current and supply voltage at the device; verify input return loss. A poorly matched LNA in a mismatched chain reads higher noise figure than its datasheet.
- Check the environment. Never measure noise figure with the antenna connected; ambient and man-made noise at a busy site will inflate the reading. Terminate in 50 ohm.
When the fault is found, re-run the full cascade prediction against the measured per-block values; they should agree within about 0.5 dB.