PULSCAR ANALYSIS OF NHTSA PUBLIC DATA — ENGINE, MOTOR AND TRANSMISSION MOUNTS Published September 2026. License: CC BY 4.0 (credit "PulsCar analysis of NHTSA public data, September 2026"). ------------------------------------------------------------------------------- 1. SOURCE FILES ------------------------------------------------------------------------------- NHTSA public ODI flat files, downloaded 17 September 2026 from https://static.nhtsa.gov/odi/ffdd/ (field definitions: cmpl/CMPL.txt, rcl/RCL.txt, ftsb/FLAT_TSBS.txt), loaded into a local SQLite database. complaints 777,040 rows, received 1 January 2015 - 8 September 2026 recalls one row per make/model/year per campaign (campaigns counted DISTINCT) tsbs 187,827 distinct manufacturer communications Owner complaints are self-reported, unverified and have no denominator. Nothing here is a failure rate, and no "most reliable" or "least reliable" claim is made from these counts. ------------------------------------------------------------------------------- 2. HOW THE COMPLAINT SET WAS BUILT ------------------------------------------------------------------------------- ROUND 1 (rejected). Every complaint whose narrative contains ENGINE MOUNT, MOTOR MOUNT, TRANSMISSION MOUNT, TORQUE STRUT, TORQUE ROD, DOG BONE or DOGBONE, minus motorcycles: 1,296. Reading 30 at random showed that about half were long multi-fault narratives in which a mount was one item on a list, or in which the phrase "engine mounted" described the location of a fuel filter or an oil cooler hose rather than a mount. Observed false-positive rate: 14/30 = 47%. We did not publish anything from that set. ROUND 2 (published). Same starting phrases, plus: a) the phrase "ENGINE MOUNTED" / "ENGINE-MOUNTED" is excluded, so the location sense of the word is dropped b) AND within 55 characters of the mount phrase there is a failure or repair cue: BROKE, BROKEN, SNAP, SEPARAT, CRACK, COLLAPS, TORN, TORE, SHEAR, FRACTUR, FAIL, DEFECT, BAD, WORN, WEAR, RUST, LOOSE, SAG, DETACH, FELL, SHOT, REPLAC, NEEDED, LEAK or DAMAG c) motorcycles excluded by the model pattern used throughout our work Result: 1,012 complaints, received 1 January 2015 - 8 September 2026. SAMPLE READING OF THE PUBLISHED SET. 30 drawn at random (random.seed(9)) and read in full: 27 were genuinely about a mount — including three where a shop had ruled the mounts out, which is on-topic. Three were not: two long multi-fault narratives, one collision. Observed false-positive rate 3/30 = 10%. The independent reading of 110 complaints in section 4 gives 7/110 = 6.4% off target, which is consistent. The tightening matters for one figure in particular. Requiring a failure cue near the mount phrase, instead of anywhere in the narrative, is what removes the "they also replaced the mounts" laundry lists — and those lists are exactly what would have inflated any claim about how often a mount is the answer. ------------------------------------------------------------------------------- 3. WHAT THE 1,012 SAY ------------------------------------------------------------------------------- something broke, sheared, separated or collapsed 327 32.3% a vibration or shake is described 246 24.3% a bolt, fastener, stud or screw is named 84 8.3% rust or corrosion is named 119 11.8% a clunk, knock, thud or bang is described 77 7.6% the symptom is at idle or at a stop 80 7.9% the symptom is under acceleration 136 13.4% a second or later replacement is described 205 20.3% collateral damage is named (CV joint, half shaft, exhaust, coolant or radiator hose, wiring harness, fuel line) 69 6.8% told it was normal wear or not covered 59 5.8% the shop could not find or duplicate it 20 2.0% towed 46 4.5% (baseline 5.1%) crash reported 28 2.8% (baseline 4.4%) fire reported 8 0.8% (baseline 1.8%) Failure mileage: 608 of the 1,012 report a usable figure (0 and >= 400,000 dropped). Median 85,000 miles; 25th percentile 55,000; 75th percentile 121,000. 46.5% under 80,000 miles and 60.7% under 100,000. Only 14 of the 1,012 mention neutral at all, which is why we do not publish any percentage about the neutral-versus-drive test. We read all 14 instead; what they show is described in section 5. By make: mount-complaints-by-make.csv (makes with >= 6 complaints). By model: mount-top-models.csv (models with >= 5). Two patterns in those tables are sharp enough to name: BMW: 73 complaints, 63.0% describe something broken or sheared and 47.9% name a bolt, bracket, stud or screw. No other make in the table is above 29%. Chrysler: 97 complaints, 38.1% name rust or corrosion, driven by the Pacifica (77 complaints, 48.1% naming rust, median reported mileage 114,000, concentrated in model years 2004-2006). Toyota's 44.4% rust share comes from frame corrosion on older trucks and SUVs. These are complaint counts, not failure rates. A make with more vehicles on the road, an older fleet or a more engaged owner base will produce more complaints regardless of how its mounts are built. ------------------------------------------------------------------------------- 4. 110 COMPLAINTS READ IN FULL — WHAT IT TURNED OUT TO BE ------------------------------------------------------------------------------- 110 drawn at random from the 1,012 (random.seed(2026)) and read in full, one primary outcome each: a mount failed and is named as the fault 46 41.8% a mount was replaced and the symptom stayed, or the mount was named and something else turned out to be at fault 17 15.5% the mount was collateral damage from something else — harsh shifting, a pothole, a collision, an engine replacement, a defective replacement part 13 11.8% bolts, brackets, studs or screws sheared rather than rubber wearing out 10 9.1% the frame, subframe or engine cradle the mount bolts to had rusted 8 7.3% a second or later replacement of the same mount 6 5.5% no diagnosis: nothing found, or the owner is guessing 3 2.7% off target 7 6.4% Derived figures used in the article: - 70 of 110 (64%) are genuine mount failures of some kind. - 17 of 110 (15%) are the counterweight: the mount was replaced and the car still did it, or the mount was a wrong answer for a transmission, a torque converter or a software problem. Four of those 17 name a transmission as the eventual finding. - 18 of 110 (16%) are structural rather than rubber: sheared fasteners (10) or a rusted frame, subframe or cradle (8). - 13 of 110 (12%) had the mount destroyed by another fault, which is the evidence behind the article's "fix what broke it" advice. Published as mount-outcomes-sample110.csv: complaint id, make, model, model year, reported failure mileage, assigned category. No free text, no VIN, no dealer name, no city. ------------------------------------------------------------------------------- 5. THE 14 COMPLAINTS THAT MENTION NEUTRAL ------------------------------------------------------------------------------- Read in full because the article previously stated an absolute rule: that shaking which is identical in neutral and in gear is never a mount. The complaints do not support an absolute: - A 2007 Honda Civic at about 46,000 miles: the owner reports vibration "in park, neutral and when backing up", and writes that the dealership diagnosed a collapsed motor mount. A collapsed mount can therefore be present in a car that shakes in neutral too. - A 2015 Chrysler 200, a 2014 Honda CR-V, a 2010 Ford Escape and a 2022 Volkswagen Tiguan all report the textbook positive result — vibration in drive that disappears in neutral. In the first three, the mounts were replaced and the vibration remained. - A 2015 Toyota Corolla reports vibration in drive at idle that goes away in neutral, with an idle speed that dips to roughly 500 rpm: an idle-control symptom that passes the mount test. - A Jeep Patriot complaint in the wider round-1 set (not in the published 1,012, because its mount phrase carries no failure cue) reports the same pattern with the mounts inspected and found good. Fourteen complaints is a small number and we say so in the article. What it is enough to do is retire a rule stated as certainty. ------------------------------------------------------------------------------- 6. RECALLS AND MANUFACTURER COMMUNICATIONS ------------------------------------------------------------------------------- RECALLS. Campaigns whose defect description or component names an engine, motor or transmission mount: 10 in the entire file, and most are not passenger-car mount failures at all (two Daimler diesel fuel line campaigns that mention an engine-mounted filter, a seat motor mount, a transit-bus traction motor mount). The passenger-vehicle ones: - 20V-788, Ford, December 2020, 1,405 vehicles: 2020-2021 Explorer and Lincoln Aviator, "The right hand motor mount fasteners may have been incorrectly tightened, and may loosen over time." - 25V-434, Volkswagen, June 2025, 546 vehicles: 2023-2025 Atlas, Atlas Cross Sport and 2023-2024 ID.4, bolts on the chassis, brake system and suspension including engine mount bolts possibly not correctly tightened. - 17V-105, Isuzu, February 2017, 21,000 trucks: burrs on the frame crossmember may cause the transmission mounting studs to break. Mount failure is overwhelmingly handled as wear, not as a safety campaign. MANUFACTURER COMMUNICATIONS: 314 distinct bulletins name an engine, motor or transmission mount, a torque strut, a torque rod or a dogbone mount. 27.4% describe a vibration, 27.4% describe a noise, 11.8% name a torque rod, torque strut, dogbone, upper or side mount, 9.6% concern an active or electronically controlled mount, and 8.3% name a bolt, bracket or screw. Published in mount-documents.csv together with the recalls. ------------------------------------------------------------------------------- 7. PRIMARY DOCUMENTS OPENED ------------------------------------------------------------------------------- - GM PIP5211F, "rough idle and/or vibration at idle in Drive", May 2017 (summary read in the manufacturer communications file, record 10099468; no PDF served). - GM PIP5348D, vibration at idle in gear caused by debris wedged in the transmission mount, May 2022 (record 10211767). - MINI SI M22 02 17, "Knocking noise from the engine compartment", June 2017, F54/F55/F56/F57. MC-10146727-9999.pdf, read in full. - BMW SI B22 04 15, "Service Action: Replace the Motor Mount Screws", October 2016, i3 and i3 REx built 1 October 2013 to 6 November 2015. MC-10146886-9999.pdf, read in full. - Nissan NTB18-002a, Voluntary Service Campaign PC601, 1 March 2018, 2016-2017 Rogue, left hand engine mounting insulator. MC-10134854-9999.pdf, 22 pages, read in full. - Recall texts 20V-788, 25V-434 and 17V-105, read in the recall file. ------------------------------------------------------------------------------- 8. WHAT WE DID NOT DO ------------------------------------------------------------------------------- - No normalization by vehicles on the road. No reliability ranking. - No estimate of how often the neutral test is right. 14 complaints mention neutral; that is a reading, not a rate, and it is presented as one. - No price re-verification. The repair price ranges in the article are prior published estimates and are labelled as such; consumer estimator sites block automated access and we do not work around that. ------------------------------------------------------------------------------ CASE SENSITIVITY — SECTION ADDED 19 SEPTEMBER 2026 ------------------------------------------------------------------------------ NHTSA stores complaint narratives in ALL CAPS for records filed through roughly 2020 and in mixed case for records filed from 2021 on. A case-SENSITIVE search of the narrative field loses about 92% of the complaints filed after 2020, so this is the single largest way an analysis of this database can go wrong while still looking finished. Every query behind this file is case-insensitive. SQLite's LIKE and its FTS5 full-text index both ignore case by default, and no query in this work uses GLOB or any case-sensitive comparison. This was not taken on trust. On 19 September 2026 every published folder was re-checked against the same database, comparing the figures here with what a case-sensitive query would have returned. The result: the complaint set published here (1,012) is LARGER than the absolute ceiling a case-sensitive search of the same terms could have returned (462 narratives). A case-sensitive query cannot produce a bigger set than itself, so the bug is ruled out for this folder by arithmetic, not by assurance.