PulsCar analysis of NHTSA public data, September 2026 Alternators and charging systems: what 3,608 owner complaints, 123 recall campaigns and 637 manufacturer bulletins say Licensed CC BY 4.0. Queries run 18 September 2026. -------------------------------------------------------------------------- SOURCE FILES -------------------------------------------------------------------------- NHTSA ODI complaint flat files COMPLAINTS_RECEIVED_2015-2019, _2020-2024, _2025-2026 -- https://static.nhtsa.gov/odi/ffdd/cmpl/ (field dictionary CMPL.txt). 777,040 distinct vehicle complaints (ODINO, product type V) received 1 January 2015 to 8 September 2026. NHTSA recall flat file FLAT_RCL.zip -- campaign, make, model, model year, component, potentially affected, defect, consequence, remedy. NHTSA manufacturer communication (TSB) flat files TSBS_RECEIVED_2015-2019, _2020-2024, _2025-2026 -- https://static.nhtsa.gov/odi/ffdd/tsbs/ (field dictionary TSBS.txt). NHTSA investigation flat file FLAT_INV.zip. -------------------------------------------------------------------------- CASE SENSITIVITY -- READ THIS BEFORE REUSING ANY QUERY BELOW -------------------------------------------------------------------------- NHTSA complaint narratives are written in ALL CAPS through the end of 2020 and in mixed case from 2021 onward. We measured this on the file itself: of 58,465 complaints received in 2020, 58,386 have a narrative identical to its own uppercase form; of 62,775 received in 2023, only 939 do. Consequence: any uppercase-only regular expression, and any uppercase SQLite GLOB pattern, silently drops most complaints filed after 2020 -- a trap that does not announce itself, because the query still returns thousands of rows. Every query on this page is case-insensitive. Python regular expressions were compiled with re.IGNORECASE; SQL text matching used LIKE with lower(), which is case-insensitive in SQLite for ASCII. Counts here therefore cover the whole 2015-2026 window on equal terms. -------------------------------------------------------------------------- WHICH COMPLAINTS COUNT (the alternator set: 3,608) -------------------------------------------------------------------------- Step 1. Every complaint whose narrative contains the string "alternator", case-insensitively: 5,887. Step 2. Split each narrative into sentences at . ; ! ? Keep only the sentences that contain "alternator". Step 3. Keep the complaint if any such sentence matches a FAILURE pattern (case-insensitive): (bad|faulty|failed|failing|defective|new|replacement|weak|dead|burnt| burned|fried|seized|smoking|overheated|melted|shorted|leaking) alternator OR alternator <=80 chars> (fail|failed|failure|failing|went out|went bad| gone bad|goes out|died|dead|quit|stopped charging|not charging|wasn't charging|isn't charging|no longer charging|undercharg*|overcharg*|seiz*| burn*|fried|melt*|smok*|short*|replac*|defect*|faulty|bad|overheat*| detach*|fell off|caught fire) OR (replac*|install*|chang*|put in|needed|needs|need|diagnos*|condemn*) <=60 chars> alternator Step 4. Drop the sentence if it matches a NEGATION pattern: alternator ... (was|were|is|are|tested|checked|check|test|read|reads) ... (fine|good|ok|okay|normal|healthy|working|fully charged|no problem| no issue|no fault) OR (not|isn't|wasn't|never) the alternator OR rul* out ... alternator OR alternator ... (ruled out|not the problem|not the issue|not the cause) OR no (problem|issue|fault) ... alternator OR (battery and alternator|alternator and battery) ... (are|were|is|was) (fine|good|ok|okay|new) Result: 3,608 kept, 2,279 dropped. The definition this produces, stated plainly: a complaint counts when the owner says an alternator failed, was diagnosed as failed, or was replaced. It is NOT "complaints caused by the alternator" and it is NOT a failure rate. Complaints are self-reported, unverified, and filed by a self-selected minority of owners. No denominator of vehicles on the road exists in this file, so nothing here ranks any vehicle as reliable or unreliable. -------------------------------------------------------------------------- SAMPLE READING (the false-positive rate we found) -------------------------------------------------------------------------- Round 1: 30 random complaints drawn from the 5,887 raw "alternator" mentions and read in full. Roughly a quarter were not alternator failures at all -- alternators that had been checked and found good, alternators a friend speculated about, and one narrative about a missing "radio isotope electric generator" that is not about cars. That round is what produced the failure and negation patterns above. Round 2: 30 random complaints drawn from the 3,608 kept set and read in full. One clear false positive: a 2022 narrative in which the dealer reported the alternator "normal", missed by the negation filter because the text used curly quotation marks that had been mangled into multi-byte sequences. That is a measured false-positive rate of 1 in 30, about 3.3%. Also counted in that round, and reported here because it changes how the counts should be read: 6 of the 30 were laundry lists of unrelated failures in which an alternator replacement is one line among many ("bad struts, bad alternator, bad battery, bad brakes"). Those are in scope under the stated definition but they are not narratives about a charging-system fault. We then tightened the negation pattern to cope with mangled quotation marks. The kept total did not change (3,608 before and after), so the 3.3% stands as measured. -------------------------------------------------------------------------- MILEAGE -------------------------------------------------------------------------- Mileage figures use the complaint's own failure-mileage field, excluding rows at or below 0 and at or above 400,000. 2,468 of the 3,608 complaints carry a usable mileage; 1,140 do not, and those are simply absent from every mileage figure rather than counted as zero. median 81,495 miles first quartile 55,403 third quartile 116,875 at or under 36,000 miles: 277 of 2,468 (11.2%) at or under 60,000 miles: 739 of 2,468 (29.9%) at or under 100,000 miles: 1,625 of 2,468 (65.8%) at or under 150,000 miles: 2,220 of 2,468 (90.0%) All-complaint baseline for the same file: median 70,000 miles (392,935 rows with usable mileage). -------------------------------------------------------------------------- BASELINES USED FOR COMPARISON -------------------------------------------------------------------------- Computed over all 777,040 complaints, from the structured fields, not from narrative text: towed 39,739 (5.11%) crash 33,940 (4.37%) fire 14,121 (1.82%) The alternator set: towed 11.67%, fire 9.67%, crash 1.39%, 29 injuries, 0 deaths. Split by manufacturer group, because one cluster dominates the fire figure: Stellantis brands (Dodge, Chrysler, Jeep, Ram) 1,755 rows (48.6%) fire 14.59% towed 13.56% every other make 1,853 rows (51.4%) fire 5.02% towed 9.88% -------------------------------------------------------------------------- NARRATIVE PATTERNS (alternator-context-and-outcomes.csv) -------------------------------------------------------------------------- All case-insensitive, all searched over the whole narrative, all overlapping -- a complaint can and usually does match several. These are counts of what owners WROTE, not of what was mechanically true. lost power / stalled / shut down while driving 1,471 40.77% names a recall or campaign number 1,646 45.62% battery replaced as well 1,200 33.26% smoke, burning smell, melting or flame 863 23.92% battery or charging warning light appeared 503 13.94% says there was no warning at all 312 8.65% the repair did not hold / failure returned 455 12.61% alternator replaced more than once 293 8.12% power steering or brake assist went heavy 208 5.76% shop could not find or duplicate the fault 106 2.94% oil found on or leaking onto the alternator 203 5.63% belt, tensioner or pulley named 198 5.49% alternator-parts-owners-name.csv counts the other parts named anywhere in the same narratives. A mention is not a diagnosis: "the battery appears in 57.82% of these narratives" means the word is there, not that the battery was at fault. Published as vocabulary, not as a cause distribution. -------------------------------------------------------------------------- RECALLS (alternator-recall-campaigns.csv) -------------------------------------------------------------------------- Query: recalls whose component field contains "alternator" or "generator", or whose defect or remedy text contains "alternator" (all case-insensitive). The flat file carries one row per make/model/model year per campaign, so campaigns are counted DISTINCT by campaign number and "potentially affected" is taken once per campaign. 123 distinct campaigns, 2,241,656 vehicles 80 of 123 (65%) name fire in the consequence field 31 of 123 name a stall or a loss of motive power Filtering out recreational-vehicle, heavy-truck, bus, motorcycle and component-supplier manufacturers by name leaves: 48 car and light-truck campaigns, 1,868,908 vehicles 36 of 48 name fire, 18 name a stall The name filter is a blunt instrument -- it matches on manufacturer name, so a passenger-car campaign filed by a company whose name reads like a truck builder would be excluded. The full 123-row file is published so the split can be redone differently. -------------------------------------------------------------------------- MANUFACTURER BULLETINS -------------------------------------------------------------------------- Query: manufacturer communications whose summary contains "alternator" or "charging system", case-insensitively, counted DISTINCT by NHTSA record id. 790 bulletins. Bulletins filed by stationary-generator, coach and heavy-duty manufacturers (Cummins, Prevost, Temsa, Navistar, Paccar and similar) were removed by manufacturer name because "generator set" bulletins have nothing to do with car alternators, leaving 637. software or reprogramming named 173 27.2% battery named 259 40.7% belt, tensioner, pulley or decoupler named 54 8.5% a noise named 40 6.3% -------------------------------------------------------------------------- LIMITATIONS -------------------------------------------------------------------------- 1. Complaints are self-reported and unverified. They are a record of what owners told NHTSA, nothing more. 2. There is no denominator. Nothing here supports "brand X alternators fail more often than brand Y". A make with many complaints may simply have many vehicles on the road, an active recall drawing owners to the complaint form, or both -- and 45.62% of this set names a recall. 3. The Stellantis cluster is a documented recall population (14V-634 and 17V-435, together 877,295 vehicles, the same 2011-2014 platforms). It distorts every aggregate figure, which is why the fire share is published both with and without it. 4. Narrative pattern counts measure words, not mechanics. "Belt named in 5.49%" does not mean belts caused 5.49% of failures; it means owners wrote the word. 5. Mileage is owner-entered and missing in 1,140 of 3,608 rows. 6. The complaint file ends 8 September 2026. Recent model years are under-represented simply because their cars are young. Contact: support@pulscar.io