Estimated Output

The “Estimated output” tool has two independent halves that share one graph:

1. Estimate from the binary — reads the calibration maps out of the currently loaded firmware and plots what the ECU would allow at WOT. Math ported from T5Suite’s “Show dyno graph” and T7/T8Suite’s “Airmass result viewer”.
2. Street dyno from log — takes a recorded log, finds the WOT pull in it, and computes what the car actually did from dv/dt.

Both draw onto the same RPM axis (the axis comes from the binary), so a measured pull can be laid straight on top of the estimate. Every curve is toggleable in the legend above the graph; the summary line under the graph reports the peak of the curves flagged as peak-worthy plus the RPM spans where a limiter was active.

Which ECU family you get is decided by the ECU selected in txlogger — T5, T7 and T8 have estimators; anything else shows “No output estimator for ECU …”.

An option is greyed out when the loaded binary does not contain the map that option needs. The “Refresh” button re-runs everything against the binary as it is right now — use it after editing a map so the estimate follows the change.

The source lives in pkg/widgets/estimatedoutput.

Section 1 — Estimate from the binary

How it works

T7 / T8 follow the ECU’s own request chain:

pedal position ──► pedal request map ──► [T8: torque → airmass] ──► torque limiters
                                                                 ──► airmass (knock) limiter
                                                                 ──► turbo speed limiter (T7)
                                                                 ──► fuel-cut limiter
                                                                 ──► limited airmass (mg/c)
limited airmass ──► torque ──► power, injector DC, target lambda, fuel flow, EGT (T8)

The whole pedal × RPM grid is evaluated, not just WOT: the Table tab shows the resulting limited airmass for every cell (highest pedal position on the top row, like the suites), with a coloured corner marker naming the limiter that clipped that cell. The Graph tab plots the WOT column only.

On T7 the pedal map already holds airmass, so the request goes straight into the limiter chain. On T8 the pedal map holds torque in 0.1 Nm, which is converted to airmass through TrqMastCal.m_AirTorqMap first.

Torque from airmass is either the simple model airmass / 3.1, ×1.07 for E85, times a fixed high-RPM roll-off (1.00 below 5060 rpm, 0.99, 0.95, 0.94, 0.85 at ≥6000 rpm), or the binary’s nominal torque map if you tick that option. Power is torque × rpm / 7024 in whole hp, using integer division exactly like the original suites, so the curve steps in 1 hp increments. Note 7024 is the metric hp (PS) constant (60·735.5 W/2π); the curves are labeled plain “hp” accordingly — multiply by 0.986 if you want true brake hp.

T5 has no torque model in the binary, so it works off boost instead: it reads the WOT column of the boost request map (Tryck_mat! or Tryck_mat_a!), converts it to gauge bar, and looks torque up in a fixed torque-vs-boost table per turbo family. The Table tab is the boost request in bar over throttle × RPM. Injector DC replays the T5 firmware’s own injection-time math at a fixed ~20 °C intake temperature and 13.0 V battery.

If a required symbol is missing you get one error naming every missing symbol at once, rather than a silent wrong answer.

T7 inputs

T8 inputs

Same as T7 except as noted:

T8 has no ambient-pressure input — there is no turbo speed limiter in the T8 chain.

T5 inputs

Output curves

Limiters

When any limiter clips a cell it is named in the table’s corner marker and, for the WOT column, in the summary line as an RPM span. Possible limiters: engine torque, E85 torque, E85 auto torque, gear torque, overboost, airmass (knock), turbo speed, fuel cut. When both a torque limiter and an airmass limiter apply, the reported one is whichever clipped last (airmass wins).

Section 2 — Street dyno from log

Load a recorded log (.t5l, .t7l, .t8l, .csv, .bpl) and the tool extracts the WOT pull from it and adds measured curves to the graph.

How the pull is found

Signals are picked per ECU:

On T7/T8 In.v_Vehicle2 is preferred when it is present in the log, because In.v_Vehicle is the left front (driven) wheel — wheelspin corrupts dv/dt there, while the rear wheel does not spin. If you can, log In.v_Vehicle2. The channel is sanity-checked first: if it never varies or disagrees with In.v_Vehicle by more than 20 % (dead rear ABS ring), the tool falls back to In.v_Vehicle instead of plotting a ~0 hp curve off a dead signal.

The tool then scans the log for every segment where the engine keeps pulling. When the throttle channel is in the log, “pulling” means at (near) full pedal — at least 85 % of the log’s own top throttle level, so the per-ECU scaling doesn’t matter — with a smoothed RPM slope above 120 rpm/s, which keeps slow 4th/5th-gear pulls. Without a throttle channel the slope threshold alone has to separate pulls from brisk driving, and is stricter: 250 rpm/s (which rejects tall-gear pulls — log the throttle signal). In both cases dropouts of up to 0.5 s are tolerated and there is a hard break whenever RPM falls more than 300 — so a gear change or a lift is never bridged. A segment must span at least 1000 rpm or it is dropped; if none survives you get “no WOT pull found”. Every found pull appears in the selector under the buttons, e.g. Pull 1: 2400-5900 rpm, 8.4 s, along with an Average of N pulls entry when there is more than one.

Practical consequence: one clean pull in one gear, full throttle, several seconds long, on a flat road. The longer and higher the gear, the better — more time per RPM point means less noise in the derivative. Everything else in the log is ignored, so it is fine to record a whole drive.

Inputs

The estimate side’s E85 fuel and Torque efficiency (%) options also feed the log’s airmass curves (see below); nothing else from section 1 affects the measured curves.

The math

For each RPM point on the estimator’s axis, every log sample within ±250 rpm of that point is collected (at least 3 needed, else the point stays blank). A least-squares fit of speed against time over that window gives acceleration a and mean speed v̄, then:

P_wheel  = mass·a·v̄  +  ½·1.225·Cd·A·v̄³  +  Crr·mass·9.81·v̄
P_crank  = P_wheel / (1 − loss/100)
torque   = P_crank · 60 / (2π · rpm)
power    = torque · rpm / 7024          (hp, metric)

RPM points the pull does not cover stay NaN and are simply not drawn.

Output curves

The two air curves are not a measurement of acceleration — they push the logged airmass through exactly the same torque model the estimate uses (airmass / 3.1, ×1.07 for E85, × the high-RPM roll-off, × your efficiency factor). They answer a different question: “given the air the engine actually got, what should it have made?” Comparing them against the dv/dt curves separates an airflow problem from a spark/fuel problem — if air torque matches the estimate but measured torque is well below both, the engine got the air and did nothing with it.

Because these curves reuse the efficiency factor, the useful workflow is: get one good pull, then adjust Torque efficiency (%) until Log air torque lines up with Log torque. From then on the binary estimate is calibrated to your car.

Accuracy, honestly

This is a coastdown-free inertia dyno with hand-entered constants, so:

• Absolute numbers are only as good as mass, drivetrain loss and Cd × frontal area. Two of the three are guesses.
• Relative numbers are good. Same car, same road, same gear, same inputs, before and after a change — that comparison is meaningful, which is what most tuning actually needs.
• Any road grade shows up directly as power. Use a flat road, or run both directions.
• Wheelspin, a slipping clutch, a torque-converter automatic, or a gear change inside the window all corrupt dv/dt. The pull detector rejects gear changes; it cannot see slip.
• The model has no rotational inertia term, which makes low gears read low. Use the tallest gear the road allows.