N776R — G3X log review4 Oct 2026

This pass

3 flights on 4 Oct 2026, all with Emily: KCLW to Leesburg, Leesburg to Winter Haven, and home, with formation flying alongside N767MM on the last two legs. They are the first flights since N776R spent 28 Sep – 4 Oct at Bender Aviation, and they read lower oil pressure than any earlier flight on this page. Bender found the oil cooler door not opening all the way, and construction-type plastic in the cooler's air inlet hose(s), in the cooler, and in the cylinder fins. Jeff Bender's view is that it was FOD from construction on the field at KCLW, drawn in through the cowling. They removed it, ran the engine up, and found nothing wrong with the oil pressure relief valve or anything else. The oil was not changed or topped up.

1. Withdrawn"The oil change explains the post-annual oil-pressure rise" — §8 The same oil is still in the engine, and the rise is gone: 67.6–69.2 psi normalized, against 72.0–73.9 between the annual and Bender. Every reading is inside the limits. What moved it is a question for Bender and Ray, not something the telemetry can settle.
2. Still openLow fuel pressure at hot idle — §3 No dip this pass (lowest ground reading 2.0 psi). The question for Ray stands.
3. ConfoundedOil-temperature residual and right-sender noise — §2, §5 Formation flying keeps the throttle and attitude moving. The Winter Haven legs scored higher on both: an oil residual of +9.8 °F and 7.70% sender noise. Both are inside what this airplane has logged before, and neither leg is a fair test of the cooler fix.
4. HoldingCHT #4 probe and cylinder 3 — §4, §6 Both stay closed. The new flights add nothing against either.
5. Unchanged2000–2250 RPM band — §9 Oversquare time is still down 17× since 30 Jul. Time in the 2000–2250 band hasn't come down, and this pass the formation leg home added 701 s to it.

A data-source note. Twelve logs from 29 May to 25 Jul went missing from the Drive log folder during the September re-filing. They were restored on 27 Sep from the 11 Sep SD-card archive (log_20260911_N776R.zip), byte-identical, and the folder is complete again.

1Standing limits screen

The check the 8 Sep pass admitted it had never built, now run on every flight. Limits are from the Lycoming O-360 operator's manual (60297-25) in the aircraft folder. N776R has a Superior O-360-EXP, so whether these limits apply is open and belongs with Superior or Pinnacle. They are used here as a screen, not as an airworthiness determination: CHT 500 °F max (435 °F advised for service life; 450 °F is the G3X caution used throughout), oil 245 °F max, oil pressure 60–90 psi normal and 15 psi at idle, carburetor inlet fuel pressure 0.5 psi minimum, 2,700 rpm rated.

Last pass's ground run (27 Sep, after a hot restart) is still the only log since 8 Sep to cross a limit: fuel pressure fell to 0.4 psi, just under the 0.5 psi minimum. That's the §3 hot-idle signature. That day's first flight, which felt down on power, measured normal on full-throttle MAP, rpm, fuel flow, EGT balance and takeoff acceleration.

2Fuel senders — the step change retracted, a layup onset in its place

On 8 Sep I reported that the right sender's noise had stepped up across the annual: 0.71% at 6–12 gal in late August against 2.77% on 5–7 Sep, "ratio 3.9×, z = 6.6". The test was three or four more flights in that band. The 27 Sep pass supplied five, and they landed between the two figures I compared, which was neither outcome I had set up. The 2 flights this pass with enough samples in the band read 7.03% (1,166 samples, 4.69–7.7% by flight). The formation leg home supplies 73% of the samples and the higher figure. Formation flying keeps throttle and pitch moving, and the level-cruise filter doesn't remove that. Each flight sits inside the per-flight range logged since 29 Jun (up to 12.30%).

What the level-controlled comparison does show is a clear change at a different point. Before the June layup, both channels behaved alike: right 0.50%, left 0.81% (24,420 and 18,123 samples). From the first flight after it, the right has read 2.14% (38,340 samples) while the left has logged 2 deviations in 19,795. The left gauge is not simply stuck: it falls 5–6 gal on flights that draw from the left tank, in line with integrated fuel flow. G3X software is 9.33 in every log, so a software change doesn't explain it.

Watch item, not a squawk, with a narrower question. A 1–3% flicker rate means the right gauge occasionally jumps 2 gal or more for a moment. That's an annoyance at mid-tank, not a hazard, and nothing suggests it is getting worse. The useful observation is the timing. The right channel's change and the CHT #4 fault both begin on the first flight after the month-long June layup. Two unrelated sensors changing at the same moment is worth one question to Ray: was anything done behind the panel, to the EIS or sender wiring, or to the fuel-quantity calibration during the June layup? That is a hypothesis, not a finding. The CHT #4 fault was cured by replacing the probe, which argues against a shared cause.

3Low fuel pressure at hot idle — reopened

On 8 Sep I wrote that 4 Sep's 0.9 psi taxi dip stood "against 2.0–3.5 psi on every prior flight", and closed it when five flights passed without a repeat. Both halves were wrong. Scanning every flight with the same statistic, the minimum taxi fuel pressure, finds 14 flights with readings below 1.5 psi. 11 of them predate the annual, going back to 15 Feb 2026, and 3 went below the 0.5 psi carburetor-inlet minimum in the manual. On 26 Sep, departing Leesburg after about two hours on the ramp, it spent 44 seconds below 1.5 psi, bottoming at 0.6. The next evening's ground run after a hot restart went to 0.4 psi. This pass had no dip: the lowest ground reading was 2.0 psi.

The shape is consistent, and it points at the boost pump. On 26 Sep and on 25 May (the deepest recent case, 0.1 psi), fuel pressure reads 5–6 psi just after start, sags to 0.5–1 psi over the next one to two minutes at idle, then steps back up by 2.2–2.5 psi at constant rpm about four minutes before takeoff. The checklist has the boost pump on for start, off for taxi, and on again for runup and takeoff. The log has no boost-pump channel, so this is inferred, not measured: the dips look like the engine-driven pump alone at hot idle. Rpm and fuel flow stayed steady through every dip, and nothing starved.

4CHT #4 probe — fixed, and staying fixed

20 flights since the probe change, 0 episodes, 7,576 warm-idle seconds with 0 flagged. The deepest dip at idle on any of them is +4 °F, so the channel has never once read below cylinders 1–3 while idling. Before the change, the last six flights flagged 39% of idle time. Counting each 60-second idle block as one independent trial, which is deliberately conservative, the chance of this clean a run from the old probe is below 10⁻²⁰.

5Oil temperature — closed, and a method fix

No excursion in 53 flights since 3 Jul. Peaks run 182–215 °F, inside the ordinary range and far from the 245 °F limit. The vernatherm passed its bath test at the annual.

One change to how the residual is measured. The first cut of the 27 Sep pass showed new-flight residuals of −17 to +14 °F, against the ±6 °F scatter reported before. All of them came from the 19 Sep circuits, where each leg had only four to six minutes of "cruise" and the oil never reached equilibrium. Below 300 s of cruise the residual scatters with a standard deviation of 9.6 °F; above 600 s it's 4.7 °F. The residual is now reported only for flights with at least 10 minutes of cruise (58 flights; 41 excluded).

This pass. The one ordinary cruise leg, KCLW to Leesburg, read −2.8 °F. The Leesburg–Winter Haven leg read +9.8 °F, now the second-highest on the page. That leg was flown partly in formation at about 16" MAP, slower and with the throttle moving: less air through the cooler for the heat being made, and no steady state for the fit to describe. That is inferred, not measured. Bender found the oil cooler door not opening fully and debris in the cooler's air inlet. The record shows no sign of it in oil temperature: the median residual was +2.6 °F between the annual and Bender, against +1.1 °F before, with a scatter of 4.7 °F. Ordinary cruise flights will show whether the fix is visible at all.

6Cylinders and CHT

The cylinder-3 squawk stays retracted: a 5 °F lead, a 10 °F spread, and 79/80 compression. After the fix, cylinder 4 holds the in-flight peak on all 20 of 20 flights, because it heats fastest in the climb. In level cruise the hottest cylinder was #1 on 6, #3 on 2 and #4 on 11 of 19 flights. Spread runs 6–27 °F and peaks 377–438 °F. Which cylinder tops the list is a matter of a few degrees and shifts with altitude and power.

7RPM

No flight this pass went above 2,800. The highest was 2,740. Across all 104 flights, 89 touch 2,700 and 45 touch 2,750, nearly all in the normal takeoff overshoot. 3 exceed 2,800, and every one of those peaked in a descent with the throttle moving. That's governor lag with the prop unloaded, and the cure is easing the throttle in or letting the speed come back first. Peak rpm against engine hours runs −5.6 rpm per 25 h, r = -0.16: no trend.

8Trends over 80.0 engine hours: oil pressure stepped twice

Withdrawn, 4 Oct: "the oil change explains it." The 27 Sep review said oil pressure rose after the annual because the annual put in fresh 20W50, so the rise wasn't wear. The rise has now reversed with that same oil still in the engine: Bender Aviation didn't change or add oil. A change of oil can't explain a step that undoes itself while the oil stays the same. The cause was assigned because the timing coincided, without a second witness. Holding the oil constant was the test, and the Bender visit ran it by accident.

What was measured. Normalized to 190 °F and 2,380 rpm, for flights with at least 10 minutes of cruise: before the annual, median 69.8 psi (48 flights, 68.1–71.7), and flat across all 80 pre-annual flights with a cruise reading. From the annual to Bender, 72.8 psi (7 flights, 72.0–73.9). Since Bender, 68.0 psi (3 flights, 67.6–69.2). The lowest, 67.6, is the lowest of the 58 flights on this page. It's still inside every limit: the lowest at power this pass was 65 psi, against the 60 psi screen. These are three flights from one hot day, so this is a first look, not a settled level.

What the record can't say. Both steps line up with time in a shop: up at the 4 Sep annual, down across Bender's 28 Sep – 4 Oct visit. The debris was on the cooling-air side (the cooler's air inlet and the cylinder fins). Because this comparison already holds oil temperature constant, a blocked air inlet on its own shouldn't move pressure at a given oil temperature. That is inference, not measurement. The timing doesn't fit the FOD either: the rise was already there on 5 Sep (72.4 psi or more), flying home from the annual at Kaolin, before the airplane was back at KCLW. The G3X has no channel for oil quantity, oil grade or relief-valve position, and it can't tell a real pressure change from an offset in the sender. The question for Bender and Ray: at the annual or at Bender, was the relief valve removed, reshimmed or adjusted, or the oil-pressure sender or its wiring disturbed? That decides it, not the telemetry.

9Propeller placard bands

The Hartzell HC-C2YR-1BFP / F7497 manual carries two placard blocks, for an ECI (I)OX-360 and a Lycoming O-360-A1A with Lightspeed ignition. Neither is N776R's engine, and whether either applies is open with Hartzell and Pinnacle. What the logs can do is measure exposure: 91.0 h running and 65.9 h airborne across the flight set.

The 2000–2250 band has not come down, and it is now the larger exposure. Before 30 Jul it was 5.59% of running time and after it 7.61%, each with its largest flight removed (20 Aug: 2,196 s on one leg). Across everything, 59% of band time is level flight and 58% is below 1,500 ft. This pass put 14.7% of running time in the band, and 13% of that was below 1,500 ft. Most of it (701 of 1,168 s) was the formation leg home. Unlike the oversquare band, no technique change has touched this one. If the Lycoming placard ever turns out to apply, this is the live item, and setting low-power cruise and downwind at about 2,300 rpm would clear it at little cost.

10Performance charts

91 stabilized cruise segments, 5.7 hours with rpm steady within 60, MAP within 0.8", airspeed within 4 kt and vertical speed under 100 fpm for at least two minutes. They cover 1,253–9,428 ft and 36–70% power. The 11 Sep test card added the low-power points. That card also validated the airspeed system: position error is +0.7 kt, constant within ±1 kt from 80 to 145 KIAS, and the logged OAT is static temperature. So every TAS figure here is good to about a knot.

Specific range runs 13.7–20.8 nm/gal. At 7,000–8,200 ft (43 segments) the best segment made 168.3 kt on 8.1 gph and the worst 159.5 kt on 11.18 gph: 5% slower for 38% more fuel. On 31 usable gallons (36 indicated less a 5 gal reserve; confirm against the build documentation) that is 645 nm against 425 nm of still-air range. None of this recommends leaning differently. Mixture is Superior's guidance and your call; the data just puts a price on the choice. A deliberate mixture sweep at a fixed altitude and power would say whether the four cylinders peak close enough together for lean-of-peak to be an option at all.

11Marking it in the cockpit — carried forward

Unchanged from 8 Sep. The 2000–2250 band and the 2,600 takeoff-only limit can be expressed as tach arcs or G3X alerts. The oversquare rule relates two instruments, so no arc can express it, and it needs a placard. Until Hartzell says whether these placards bind a Superior O-360-EXP, the honest sequence is an advisory placard on the quadrant plus a G3X checklist page now, with arcs and alerts only after the answer.

12Retractions, and the pattern

Thirteen findings have now been walked back. This pass's is a cause, not a measurement: the oil-pressure step was real, and its explanation failed.

The 27 Sep pass's two failures came from the same gap. A number from recent flights was compared against a convenient slice instead of the whole record. The fix is structural, not a matter of care: the script now computes every figure over every flight with one definition, and states each claim as a test it must pass before the page can be built. This pass's withdrawal is checklist item 7: a step that coincided with a maintenance event was credited to one thing done at that event, with no independent witness.