Spot the mistake in this design: built exercise #2 (draft)¶
Drafted 8 Oct 2026 for issue #227. Nothing has been posted. A BUILT exercise under the "Built exercises" rule in marketing-block-1.md: the design is made up on purpose, with exactly ONE planted mistake, and must never be presented as an AI's answer. The owner judged a first version (an out-of-band LNA) too easy, so this one tests a design rule.
Post 1 (midweek): the puzzle¶
Spot the mistake in this design. 🔎
A made-up teaching example. A student designs the receiver for a 9.4 GHz (X-band) marine radar, 10 MHz bandwidth, and needs 12 dB signal-to-noise. Below is the chain, stage by stage, and the results the student reports. Take every part figure in the table as given. Something else is wrong. What?
Answer on Friday.
RFEngineering #ElectronicsEngineering #EngineeringStudents #Radar¶
Image (chain and claimed results):
| # | Stage | Part | Gain / loss | Noise figure |
|---|---|---|---|---|
| 1 | Limiter | Qorvo TGL2209-SM | −0.4 dB | 0.4 dB |
| 2 | Image filter | Knowles B095MB1S | −1.75 dB | 1.75 dB |
| 3 | Low-noise amplifier | Infineon BFP840ESD | +21.5 dB | 1.15 dB |
| 4 | Mixer | Qorvo CMD183C4 | −6 dB | 6 dB |
| 5 | IF filter | Mini-Circuits CBP2-1090C+ | −1.4 dB | 1.4 dB |
| 6 | IF amplifier | Mini-Circuits PHA-23LN+ | +21 dB | 1.2 dB |
| 7 | Step attenuator | Analog Devices HMC1119 | −1.3 dB | 1.3 dB |
Claimed results: receiver noise figure 1.78 dB; weakest signal (MDS) −102.22 dBm. (planted mistake: stage 2 sits in front of the LNA, but the results are those of the order with the LNA first)
Post 2 (Friday): the answer¶
Spot the mistake in this design: the answer 🔎
The results don't match the order. With the image filter IN FRONT of the low-noise amplifier, its 1.75 dB loss adds almost directly to the receiver's noise figure: the chain as drawn comes to 3.45 dB, not 1.78 dB, and the weakest signal it can hear rises from −102.22 to −100.55 dBm.
1.78 dB is what you get with the LNA first and the filter after it. Anything before the first amplifier costs you its full loss in noise figure; after the LNA's gain it barely matters.
The 12 dB signal-to-noise target is the student's own choice and plays no part in the mistake.
Lesson: put the LNA as early as you can, and recompute the noise figure whenever you move a stage.
RFEngineering #ElectronicsEngineering #EngineeringStudents #Radar¶
Image: the two orders side by side, each with its noise figure and MDS (figures below).
Proof (MagicON results, for the answer image)¶
MagicON run_link_budget (MagicON AI connector, live server, 2026-10-08), the same seven stages
and figures as the table, 10 MHz bandwidth, 12 dB required SNR, −80 dBm reference input:
- Order as drawn (filter, then LNA): cascadedNfDB 3.45, mdsDbm −100.55, sensitivityDbm −88.55. Cumulative NF after each stage: 0.4, 2.15, 3.3, 3.37, 3.4, 3.44, 3.45.
- Engine's order (LNA, then filter): cascadedNfDB 1.78, mdsDbm −102.22, sensitivityDbm
−90.22. Cumulative NF: 0.4, 1.55, 1.56, 1.67, 1.72, 1.78, 1.78. This reproduces the
design_rf_receiverresult below exactly, which is the control. - âš Both runs warn that the mixer's 6 dB noise figure is not marked single- or double-sideband (Maxim AN1131: they differ by 3.01 dB). The uncertainty is the same in both orders, so the comparison stands; the post quotes no claim about it.
The planted change: only the ORDER of stages 2 and 3 (the engine placed the LNA second, the image filter third); every part, part figure and claimed result is the engine's.
Source of the correct design (keep with the draft)¶
MagicON design_rf_receiver, run 2026-10-08 through the MagicON AI connector (the live public
server). Inputs: rfFrequencyHz 9.4e9, bandwidthHz 1e7, requiredSnrDb 12, application
marine_radar; grounding caller_asserted (a made-up brief, so no user's words were passed).
parts: limiter TGL2209-SM, LNA BFP840ESD, image filter B095MB1S, mixer CMD183C4, LO HMC530LP5, IF filter CBP2-1090C+, IF amp PHA-23LN+, step attenuator HMC1119;needsPartFlags[].stages(figures used in the table): limiter loss 0.4 / NF 0.4; LNA gain 21.5 / NF 1.15; image filter loss 1.75; mixer loss 6 / NF 6; IF filter loss 1.4; IF amp gain 21 / NF 1.2; HMC1119 −1.3 / NF 1.3 (its minimum setting).cascade: cascadedNfDB 1.78, mdsDbm −102.22, noiseFloorDbm −104, sensitivityDbm −90.22.frequencyPlan: ifHz 1128000000 (chosen by the engine), loHz 10528000000 (high side), imageHz 11656000000, imageRejectionDb 65.- ⚠The engine warned that BFP840ESD's 21.5 dB gain is the MIDPOINT of a stated 9.5–33.5 dB range, and that its nearest measured point (10 GHz) states 12.0–17.0 dB. The puzzle uses the engine's figure; the lesson (loss before the LNA) does not depend on the exact gain.
Before posting¶
- Not checked against datasheets: the part figures (they are the engine's). The post says "take every part figure as given" rather than "every figure is right", because of the BFP840ESD gain warning above: a reader could otherwise argue the LNA gain is a second mistake.
- Re-check 2026-10-12: MagicON
run_link_budgeton the table's stages gives 3.45 dB / −100.55 dBm as drawn and 1.78 dB / −102.22 dBm with the LNA first, as above. MagicONdesign_rf_receiverwith the same inputs picks the same eight parts and gives cascadedNfDB 1.78 and mdsDbm −102.22; its cumulative NF after the image filter reads 1.57 againstrun_link_budget's 1.56 (final figures identical; neither number is in the post). - Answer text checked with MagicON
check_grounding: grounded, 0 findings. On the puzzle table it flags −1.75, −6 and −1.4 dB: these are the engine's ownlossDB1.75 / 6 / 1.4 written as negative gains, a checker limit added to issue #477, not a wrong figure. - Image titles carry no number, because draft numbers do not match the posting order (this one, built #2, went out first: puzzle posted 2026-10-12, see posted-log.md).
- Ready-to-post folder (outside the repo):
Documents/MagicON-posts/puzzle-built-02-filter-before-lna/. - 2026-10-12, after the puzzle went out: the answer gained the line about the 12 dB target, so
readers do not take it for the mistake. The 12 dB only sets sensitivity (−90.22 dBm in the
engine result), which neither post shows; whether 12 dB suits a real marine radar is not
verified (no MagicON tool derives a required SNR;
run_radar_performancetakes it as an input). - Check posted-log.md first; log both posts after posting.