Skip to content

Spot the mistake in this design: built exercise #3 (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: made up on purpose, ONE planted mistake, never presented as an AI's answer. It tests a design rule (where receiver leakage really comes from in a circulator T/R module), proved by a MagicON result.

Post 1 (midweek): the puzzle

Spot the mistake in this design. 🔎

A made-up teaching example. A 9.4 GHz marine radar shares ONE antenna between transmitter and receiver through a circulator, with a limiter protecting the receiver. Below are the figures and the student's leakage check. Take every part figure as given. What did the student miss?

Answer on Friday.

Image (figures and the check):

Item Figure
Transmitter output into the circulator 41.15 dBm peak (10% duty)
Circulator JQL JCSM0800T1200S20 isolation 20 dB, loss 0.5 dB
Antenna 1.5:1 VSWR (13.98 dB return loss)
Limiter Qorvo TGL2209-SM damage limit 40 dBm (CW rating)

Student's check: "Leakage into the limiter = 41.15 dBm − 20 dB circulator isolation = 21.15 dBm." (planted mistake: only the direct path through the circulator is counted)

Post 2 (Friday): the answer

The student counted only the leak THROUGH the circulator. The bigger one comes back from the antenna: a 1.5:1 antenna reflects part of the transmit pulse, and the circulator sends that reflection straight to the receive port. MagicON's budget: direct path 21.15 dBm, reflected path 26.17 dBm, and the reflected path is the one that sets the leakage.

The limiter still survives (13.83 dB below its 40 dBm limit), but the real leakage is higher than the student's number, and a worse antenna makes it higher still.

Lesson: in a circulator T/R module, the antenna match, not the circulator's isolation, usually decides how much transmit power reaches the receiver.

Image: the two paths side by side, each with its level (figures below).

Proof (MagicON result, for the answer image)

MagicON design_tr_module, run 2026-10-08 through the MagicON AI connector (the live public server). Inputs: frequencyGhz 9.4, targetOutputPowerDbm 40, transmitDutyCyclePercent 10, application marine_radar, topology circulator_limiter; grounding caller_asserted (a made-up brief). budgets.isolation:

  • directPathDbm 21.15, reflectedPathDbm 26.17, leakageDbm 26.17, bindingPath antenna_mismatch, headroomDb 13.83, rxMaxSafeDbm 40 (basis input_power_max_cw_dbm), verdict safe.
  • Disclosures: the 1.5:1 VSWR (13.98 dB return loss) is the tool's ASSUMPTION because no antenna return loss was given; the CW rating is used on purpose as the conservative damage threshold.
  • Transmit asBuiltOutputDbm 41.15; shared part circulator JCSM0800T1200S20 (isolation 20, insertion loss 0.5, max input 43 dBm); limiter TGL2209-SM.

The planted change: none of the figures; the student's check simply leaves out the reflected path. Every number shown is the engine's.

Other things the result says (not used in the post)

  • The input drive was ASSUMED at 9.5 dBm (not stated), so the brief does not mention the exciter.
  • The LNA (BFP840ESD) is driven into compression during each pulse ("deaf, not damaged"); the circulator's peak is "marginal" against its CW rating (no pulsed rating declared); the 12 V board cannot make the PA's 24 V rail ("UNMET RAIL" in the BOM). A later exercise could use any of these.

Before posting

  • âš  On hold (2026-10-09, issue 361): the circulator this exercise names, JQL JCSM0800T1200S20, is filed in our catalog as 8–12 GHz with no source, but JQL part names give the band in MHz, so it reads as an 800–1200 MHz part. A fair claude.ai chat (trap kit question 4) pointed this out. Do not post #3 until the part's band is confirmed or the catalog is fixed and the design re-run; an expert reader would catch an L-band circulator in an X-band radar.

  • Not checked against datasheets: the part figures (the engine's). The post says "take every part figure as given".

  • Images: spot-the-mistake-built-03-puzzle.png, spot-the-mistake-built-03-answer.png, built by docs/scripts/make-spot-the-mistake-built-03.py.
  • Check posted-log.md first; log both posts after posting.