The fault-code dataset, free to download and reuse

Every code on this site in one file. 297 fault codes across 27 brands and 15 appliance types, each with its meaning, likely causes, checks, reset steps and the manufacturer documents it was compiled from. Licensed CC BY 4.0: use it in anything, commercially or not, as long as you credit where it came from.

Fault codes297
Brands27
Appliance types15
Records with a cited source297 of 297
Distinct documents referenced167
Last verified record2026-09-23
LicenceCreative Commons Attribution 4.0

Why this exists

No manufacturer publishes a cross-brand fault-code table, because none of them has a reason to. The codes live in a few hundred separate PDFs, each written for one product family, and they contradict each other more often than you would expect — the same two letters can mean different things on two fridges from the same maker. Assembling that into one comparable table was most of the work behind this site, and it is more useful to everyone as a file you can query than as pages you have to read one at a time.

It is deliberately boring data: no tracking, no sign-up, no API key, no rate limit. Two static files.

What is in each field

FieldMeaning
brand, brand_nameCanonical brand slug and its display name.
appliance, appliance_nameAppliance-type slug (refrigerator, furnace, generator …) and display name.
code, code_slugThe code as the appliance shows it, and the slug used in the URL.
urlCanonical page for this code on rvcodefix.com.
severityinfo · warning · danger — our reading of how urgent the fault is.
meaningWhat the code means, in our own words.
causesLikely causes, each tagged most-likely · common · less-common · rare.
diy_checksChecks an owner can safely make.
reset_stepsDocumented reset or clear procedure, where one exists.
affected_modelsModel numbers or families the code is documented for.
safety_note, call_tech_whenSafety warning, and when the job stops being a DIY job.
caveatsKnown limits: contradictions between manuals, model-year differences, corrections.
sourcesTitle — publisher — URL — manual reference, for every source used.
confidence0–1. How firmly the sources pin the meaning down.
verified, last_verifiedWhether a source confirmed it, and the date it was last checked.
url_de, meaning_deGerman page and translated meaning, for the codes that have one.

In the CSV, fields that hold several values (causes, checks, steps, models, sources) are joined with  | . The JSON keeps them as arrays and keeps each source as an object, so prefer JSON if you intend to process the provenance.

How to credit it

Attribution is the only condition, and a link is the easiest way to satisfy it:

Fault-code data: RVCodeFix (https://rvcodefix.com/data/), licensed CC BY 4.0.

If you build something with it — an app, a lookup tool, a workshop cheat sheet, a video — we would genuinely like to know. Write to info@rvcodefix.com.

What this dataset is not

How each record is verified →