How to read a dyno sheet without being fooled

A dyno sheet is the tuning industry's favourite piece of paper and its least reliable one. Not because dynos lie, but because the number at the top is the least informative thing on the page. Here is how to read one.

Wheel or crank

A chassis dyno measures torque at the tyres. The engine makes more than that; the gearbox, the differential, the driveshafts and the tyres eat some on the way. Sheets that quote "crank" figures have added an estimated drivetrain loss back on, and the estimate is a choice. Two shops can print the same wheel figure and different crank figures without either being dishonest. Compare wheel to wheel, and when a figure is not labelled, assume it is the flattering one.

Correction

Engines make more power in cool, dense air. Dyno software applies a standard correction to bring the day's result to a reference atmosphere, and there are several standards. Two sheets corrected to different standards differ by a few percent for the same pull. A sheet with no correction noted, or one that quotes an unusually favourable one, is a sheet to read with care.

Smoothing

Raw dyno data is noisy. Software smooths it, and the amount of smoothing changes the peak: heavy smoothing lowers spikes and raises the appearance of a flat, healthy curve. The setting is usually printed on the sheet. Comparing two sheets with different smoothing is comparing two drawings of the same thing.

The dyno itself

Dynos differ in type and in the load they apply. An inertia dyno lets the engine accelerate a heavy drum and infers power from how fast; a load-bearing dyno holds the engine at a set rate. They can read the same car differently, and both differ from the road. A number is only comparable to another number from the same dyno, on the same settings, ideally on the same day. This is why the honest way to present a tune is before-and-after on one dyno, in one session, and why a single peak figure from an unknown dyno is close to meaningless.

The curve

Look at the shape before the number. Where does torque come in, and how fast? Does it hold, or peak and fall? A Stage 1 on a turbo engine should raise the whole plateau and extend it, not add a spike. A curve with a dip in the middle is a boost control problem; one that falls off early is a turbo at its limit or timing being pulled at high rpm. A curve that gains at the top and nowhere else is a rev limit change dressed up as a tune. The peak is one point on a line, and the line is the tune.

What a sheet cannot show

Knock. Mixture. Timing pull. A car can make a strong number while the ECU is holding it back on one cylinder and the mixture is going lean at the top. Those live in the datalog, not on the sheet, and a tuner who shows a sheet without a log is showing the result and hiding the method.

The results page on this site

Has no customer sheets yet, and says so; what it shows is the simulator's own curves, marked simulated in the plot itself. When real sheets arrive they will be before-and-after, one dyno, one session, with the correction and smoothing stated, and a log beside each. That is the standard this article describes, and it would be strange to hold other people's sheets to it and not our own.