Scope: everything other than the known right-tank sender noise. Three items are worth a wrench; the rest of the engine and electrical picture is clean and, where it can be normalized for temperature and power, flat over 56 engine hours.
Starting 29 June 2026 — the first flight after the ~1-month layup — the CHT #4 channel intermittently reads 85–170 °F below cylinders 1–3. It has recurred on 10 flights since, in episodes from 11 seconds to 23 minutes, and it was absent from all 29 flights between 7 Feb and 29 May.
This is the instrument, not the cylinder. During every episode EGT #4 tracks the other three normally — a cylinder actually running 140 °F cool would show it on EGT too. The channel also freezes: on the 11 Jul departure CHT #4 sat at exactly 244 °F for 25 seconds while 1–3 crept up degree by degree, then rose only 7 °F through a takeoff roll that moved the other three up 36 °F. That signature — stuck, damped, offset low, recovering on its own — reads as a bayonet probe backed out of its well or a loose/chafed lead, not a hot-side problem.
| Date | Window (local) | Duration | CHT4 low by | EGT4 delta |
|---|---|---|---|---|
| 29 Jun | 19:11:06–19:14:17 | 2m 37s | 148 °F | −5 °F |
| 29 Jun | 19:15:26–19:15:51 | 26 s | 86 °F | +28 °F |
| 30 Jun | 08:12:15–08:18:57 | 6m 43s | 134 °F | −46 °F |
| 3 Jul | 12:55:09–12:55:24 | 16 s | 130 °F | −2 °F |
| 3 Jul | 12:56:29–13:14:21 | 17m 53s | 139 °F | −103 °F |
| 11 Jul | 08:30:05–08:30:25 | 21 s | 91 °F | −42 °F |
| 11 Jul | 08:54:22–08:54:34 | 13 s | 106 °F | +21 °F |
| 11 Jul | 08:58:04–09:21:28 | 23m 09s | 132 °F | −114 °F |
| 25 Jul | 12:46:19–12:46:34 | 11 s | 104 °F | −22 °F |
| 8 Aug | 09:18:56–09:19:31 | 33 s | 89 °F | +18 °F |
| 11 Aug | 08:13:19–08:13:50 | 30 s | 107 °F | — |
Peak oil temperature ran 230 °F (29 Jun), 220 °F (30 Jun) and 228 °F (3 Jul) — the three highest readings in the whole dataset, all in the same 5-day window that the CHT #4 fault appeared. Every other flight since February peaks between 176 and 218 °F.
On the 3 Jul morning leg the temperature climbed steadily and then sat pinned at exactly 228 °F for 12+ minutes in level cruise while CHTs were an unremarkable 370 °F and oil pressure sagged to 63 psi. Normal cylinder temps with hot oil points at the oil side rather than cooling airflow — a vernatherm not seating (bypassing the cooler) or a partially blocked / air-starved oil cooler are the usual causes. It has not recurred since 3 Jul, which fits an intermittently-sticking thermostatic valve.
Cylinder 3 records the flight's hottest CHT on 61 of 67 flights. In stable cruise it only leads by 4–10 °F, which is normal, but in low-airspeed climbs it separates and drives the three CHT TEMP cautions in the dataset: 454 °F (16 May), 470 °F (29 Jun), 466 °F (30 Jun). All three were climbs at 98–107 KIAS with 12–12.6 gph and OAT 27–32 °C — a lean, slow, hot climb.
Nothing here says a cylinder is sick: EGT 3 sits in the middle of the pack and compression-adjacent indicators (fuel flow, EGT stability, mag drops) are all normal. It reads as a cooling-distribution item. Worth having the inter-cylinder baffles, #3 baffle seal and the cylinder-3 air ramp looked at while the cowl is off.
Quantifying the known issue so the shop has a number: in level cruise the right sender now sits ≥2 gal away from its own 60-second median on 7.6% of samples, up from 3.9% in Feb–May. It is real and it is getting worse.
The thing that may not be on the squawk list: the left sender does the same thing, just less often — 1.7% of samples recently, and it spiked to 22.9% on 16 May and 11.4% on 10 Jul. Both channels are 1-gallon-quantized and both saturate at 18 gal, so a full tank hides left-side noise entirely. If the right sender is coming out anyway, it's worth pulling and checking the left float/wiring in the same visit rather than doing this twice.
The totalizer itself is fine: fuel burned by flow integration matches the summed sender delta with a mean error of +0.35 gal across 38 flights. Left and right tanks drain evenly (2.97 vs 2.99 gal per flight average), so there is no cross-feed or tank-selector asymmetry hiding under the noise.
| System | Result over 56 engine hours |
|---|---|
| Oil pressure | Normalized for oil temp and RPM: 70.0 → 70.1 psi, slope +0.01 psi/25 h, r = 0.01. Flat. The raw downward drift is fully explained by hotter summer oil. |
| Oil temp (baseline) | Normalized for OAT and CHT, +2 °F/25 h with r = 0.20 — noise, once the three June/July outliers are set aside. |
| Electrical | Cruise bus 14.19 V every single flight; alternator 16–17 A steady; no low-voltage events with the engine running. Zero trend. |
| Mag checks | Drops of 60–190 RPM with a 40–130 °F EGT rise on all four cylinders each time. No dead or lazy plug signature anywhere in the dataset. |
| Ignition / combustion stability | Cruise fuel-flow jitter 0.15–0.68 gph and EGT jitter 3–19 °F, with no single cylinder standing out and no trend. |
| Cruise efficiency | Fuel flow at matched RPM/MAP/altitude: 9.0 → 8.9 gph. No power loss. |
| Fuel pressure | Cruise mean 4.4 psi, slope −0.3 psi/25 h — inside the flight-to-flight scatter. All 34 FUEL PRESS alerts but six occurred with the engine stopped; the six exceptions are single-sample dropouts at engine start, not real pressure loss. |
| Carbon monoxide | 0 ppm on every sample of every flight. Sensor is reporting (not null) — no exhaust leak indication. |
| Autopilot / trim | Roll torque 1–8%, pitch torque 2.2 → 3.5% average. The pitch creep is mild and could just be loading; worth a glance at the pitch servo and trim, but nothing is out of family. |
One housekeeping note: the 75 samples where several engine parameters go implausible at once all land within a second or two of engine start on a hot restart, one cluster per flight. That is the GEA settling, not a data-bus fault.
Derived from the raw G3X CSV logs in N776R\g3x_data_log. All thresholds and windows used are stated inline so any finding can be re-checked against the source files.