Evidence-graded diagnostic cases
Cases
A fault code tells you a threshold was crossed. A case shows what someone actually measured, what it did and did not prove, and whether the repair held. Every case is graded by how strong its evidence is — and nothing below the floor moves a published number.
The evidence ladder
Why it exists: a dataset that treats "somebody said so" the same as a before/after dyno becomes folklore with a percentage sign on it.
Published cases
No verified real-vehicle cases published yet. We would rather show the gap than pad it. Below is a worked demonstration on synthetic data, so you can see the shape a real one will take — then the method goes live with documented DMS cases, after technical review and consent.
Underboost — worked example
A requested-vs-actual boost trace that confirms an underboost in the data, carries the P0299 check list, and deliberately does not name the wastegate before the test that would. Synthetic samples — no real vehicle was analysed.
Example shape: actual builds but plateaus around 0.9 bar and never reaches the 1.40 bar request — the gap is the fault.
For workshops & DMS
Have a documented case worth publishing?
DMS Tuned seeds this library with real, instrumented cases (boost control, fuel trims, calibration validation). Pilot workshops can contribute their own — with evidence, consent and a technical review before anything goes live. We never publish a repair we cannot back.
Talk to the engineers →