Up-shift & down-shift analysis: shift points, shift time, blips, over-rev
Optimal upshift rpm maximises the power used between post-shift and shift rpm; shift time is read from the longitudinal-g dip; down-shifts need a correct blip and must be protected against over-rev.
Up-shift & down-shift analysis: shift points, shift time, blips, over-rev: L1 Foundation explanation
Shift point. Shifting at the rev limiter is not automatically best. After an upshift the engine drops to . What matters is the power the engine delivers over the rpm range it actually works in — from to . Segers' F3 example: with ≈ 1200 rpm drop per shift, the area under the power curve between post-shift and shift rpm was maximised at 6500 rpm, not at 7000. A simulated 1 km acceleration test was 0.12 s quicker with 6 km/h higher top speed when shifting 500 rpm earlier.
Exact rule (which Segers' area method approximates): upshift when the power at the current rpm equals the power at the post-shift rpm — at the same road speed, force = P/v, so beyond that point the next gear gives more force. The area method gives the same answer when the rpm drop is a fixed number of rpm, as in Segers' example. When the drop scales with the ratio (), use the mean power over , not the raw sum: with a proportional drop the raw sum biases the shift late, often all the way to the limiter.
Check it in data: look at longitudinal g across an upshift. After the dip, acceleration should resume on the same downward-sloping trend line as before. If it resumes lower (a step down), the shift rpm was too low; if acceleration was already sagging before the shift, it was too high.
Shift time. Time in a shift is time without drive. Read it from the longitudinal-g dip — from where drive drops off to where it recovers. Segers' reference values:
| Car / system (Segers Table 6.3, early-2000s data) | Upshift time |
|---|---|
| F3 car | 0.15 s |
| Porsche 911 GT2 Turbo, synchronised H-pattern | 0.35 s |
| Dodge Viper GTS-R, synchronised H-pattern | 0.32 s |
| Viper GTS-R, sequential, no power-shift | 0.23 s |
| Viper GTS-R, sequential + power-shift (ignition cut) | 0.18 s |
| LMP1 prototype, sequential, electronic paddle-shift | 0.10 s |
Modern paddle-shift GT3 and formula cars typically cut drive for ≈ 40–80 ms. Dual-clutch boxes (PDK/DCT in GT4 and some one-make cars) hand torque over between two clutches, with almost no gap. You need ≥ 50 Hz (better 100 Hz) logging of long g to measure this.
Down-shifts. The engine must be sped up to match the lower gear — the driver (or auto-blip) blips the throttle. Too little blip: the rear wheels have to accelerate the engine — a braking torque spike on the rear axle mid-trail-braking → snap oversteer. Too much: over-rev. Most over-revs happen on downshifts — from shifting down too early (rpm too high) or over-blipping.
Up-shift & down-shift analysis: shift points, shift time, blips, over-rev: L2 Practitioner explanation
Finding the optimum shift rpm per gear. With dyno power in 100–200 rpm steps:
- For each candidate shift rpm compute .
- Compute the mean power over : the sum of the table points divided by the number of points (the raw sum, i.e. the area, grows with the window and biases the shift late).
- Pick with the largest mean, then check it against the equal-power crossover, which is the exact answer. Repeat per gear (steps differ!). Shift-light settings per gear follow from this; set the light ahead of the target by the rpm the engine gains during the driver's reaction time (≈ 0.1–0.2 s). Compute it as with [rpm/s]: for a GT3 ≈ 200–400 rpm in 1st, ≈ 70–250 rpm in 2nd–3rd and only 10–20 rpm in top. Paddle-shift cars may use an automatic upshift light with gear-dependent rpm.
Comparing drivers. Overlay speed and rpm on a straight: Segers' example shows one driver consistently shifting higher and ending 3 km/h slower at the end of the straight. Always compare several laps (wind, tow).
Shift-time analysis as a reliability tool. Calculate the average upshift time per gear per stint: a rising trend (slower, double-shifts, missed shifts) indicates actuator pressure, dog wear or clutch problems. In endurance racing with several drivers it also reveals who is abusing the gearbox (e.g. upshifting without lift on H-pattern, or "pre-loading" the lever on dog boxes).
Down-shift analysis:
- Blip quality: total-ratio channel spikes (see gear chart lesson), rear wheel speed dipping below front wheel speed at each downshift = rear-axle locking from engine braking.
- Over-rev: predicted rpm after a downshift is . If a driver downshifts 3→2 at 7000 rpm with a 1.27 step, the engine would see 8900 rpm — a likely engine-killer. Modern ECUs/gear-control units refuse the downshift above a calculated threshold (downshift lock-out), and drivers learn to "queue" downshifts. Check the log for refused-shift events: frequent refusals mean the driver is downshifting too early — lost time and potential frustration.
- Over-rev counters (time above limit, max rpm) are often read by the engine supplier; know them before they do.
Auto-blip calibration (paddle cars): blip throttle %, duration and ignition settings per gear. Insufficient blip in low gears (larger rpm jump) is the classic cause of rear instability in hairpin entries.
Up-shift & down-shift analysis: shift points, shift time, blips, over-rev: L3 Advanced explanation
Shift-point theory with shift time and inertia. The simple "equal power" rule assumes instantaneous shifts. With a finite gap the car has no drive and decelerates under drag and rolling resistance, so it re-engages at a lower speed than it shifted at:
The time-optimal criterion becomes force just before the cut = force just after re-engagement, . If the post-shift rpm sits on the rising part of the torque curve, landing lower costs force and the optimum moves slightly later than the instantaneous crossover. Size of the effect: a GT3 at 250 km/h with ≈ 3 kN of drag and rolling resistance loses ≈ 0.14 m/s in a 60 ms gap, ≈ 0.2 % of speed (≈ 10–15 rpm). Even a 300 ms H-pattern shift only gives ≈ 1 % (≈ 60–80 rpm). So the shift point barely moves. What shift time really changes is whether to shift at all: near the end of a straight, upshift only if the extra force in the next gear over the remaining distance outweighs the gap. Otherwise hold the lower gear on the limiter into the braking zone, which is the rational choice in the last 50–100 m with a slow gearbox.
Strictly, compare accelerations, not forces: the lower gear has the larger (engine inertia × ), so the acceleration crossover comes slightly before the force crossover, most noticeably on the 1→2 shift.
The rotational energy released on upshift, , is a separate effect and is partly returned to the car as the engagement torque spike. With modern low-inertia engines it is small: – kg·m², 7500→6000 rpm releases ≈ 5.5–11 kJ, equivalent to ≈ 14–28 ms of 400 kW.
Restricted and turbo engines. Above restrictor choke rpm power is flat. If the plateau is wider than the gear step (), any shift rpm that keeps the post-shift rpm on the plateau gives the same force, so shift at the low end of that window to save fuel and engine life with no time loss. If the plateau is narrower than the step, the post-shift rpm always falls off the plateau, and the equal-power criterion pushes the shift to the limiter. Turbo engines with boost tapering at high rpm often have an optimum well below the limiter; boost maps by BoP can move it event-to-event.
Torque interruption and chassis. An upshift cut on a corner exit (car still at combined slip) unloads the rear tyres for 40–80 ms and then re-applies torque in a step. If the rear is near the limit this torque step can trigger oversteer — another reason to avoid shifts on loaded exits and to tune the torque ramp-in after the cut. On downshifts, the blip/engine-braking torque passes through the diff on coast — interacting with coast locking (see differential lessons).
Quantifying shift-time value. A gap removes the drive force for (the car coasts against drag instead of accelerating), leaving a speed deficit , at least . With the first-order exit-gain result from the time–distance lesson, , the loss per shift is
The acceleration level cancels out; only the drag share remains (≈ 1.1 at 110 km/h, ≈ 2 near top speed for a GT3). An early shift (1→2 at 110 km/h, braking at 265 km/h) costs ≈ 0.6 , ≈ 35–40 ms for a 60 ms gap; a 5→6 shift at 240 km/h costs only ≈ 0.2 . Summed over 15–25 upshifts per lap, a 60 ms system beats a 150 ms system by roughly 0.3–0.6 s/lap. That number justifies gearbox maintenance and actuator investment, and explains why seamless-shift dog boxes (F1 since the mid-2000s) are banned in most categories on cost grounds.
Dual-clutch transmissions. DCT/PDK/DSG boxes (Porsche 718 GT4 RS Clubsport, Ferrari 296 Challenge, some TCR cars) pre-select the next gear on the second shaft and hand torque over from one clutch to the other. Wheel torque never drops to zero; the shift is a torque phase (clutch-to-clutch overlap) followed by an inertia phase (clutch slip pulls the engine down to the new speed). Analysis changes accordingly: shift time comes from the TCU, the long-g trace shows a shallow sag rather than a dip, and the engineer watches clutch temperature, clutch slip during the inertia phase and launch behaviour. Against a dog box with 60–80 ms gaps, a DCT gains a measurable amount per shift on the straights, which BoP between DCT and dog-box cars has to absorb.
Up-shift & down-shift analysis: shift points, shift time, blips, over-rev: L4 Expert explanation
Upshift on a loaded exit: torque ramp-in vs combined-slip margin. During the cut the rear tyres go from drive slip to near-zero slip. Their lateral capacity rises (friction-ellipse effect), and the car tightens slightly. Then torque returns. If it returns as a step, the rear tyre builds its drive slip again within a short longitudinal relaxation length (a few ms at exit speeds), so the tyre passes on the torque step almost instantly and the force overshoots through driveline shuffle. Near the friction limit that overshoot is what removes the lateral margin. Model it as follows:
- available lateral margin before the shift ;
- peak drive force after re-engagement , with overshoot from the shuffle damping (often 20–50 % for a step input, much less for a 50–100 ms ramp). Calibrate the ramp-in per gear so the peak never exceeds the drive force the tyre can carry at the lateral g of that corner exit. In practice the low gears (high wheel torque, low shuffle frequency) need the longest ramps; top gear on a straight can take a near-step. The cost is a few ms of lost drive per shift, small next to a single snap.
Auto-blip under engine-braking maps. On a downshift the GCU opens the throttle (or sets a torque request) to raise the engine to within the dog-engagement window, then hands back to the engine-braking map. Two calibrations interact:
- Blip amplitude/duration: in low gears the rpm jump is large (≈ 1500–2000 rpm), the engine must accelerate at ≈ 20 000–40 000 rpm/s, and the blip torque must cover engine friction plus the inertia term . Under-blip shows as a rear braking spike; over-blip as a drive pulse on entry.
- Engine-braking map: after engagement the drag torque reaches the driven axle through the diff's coast ramp. Too much drag at high rpm on a short gear means rear lock-up or oversteer in braking. Many ECUs therefore use gear- and rpm-dependent engine-braking targets (a small throttle opening on overrun) that limit drag torque in the low gears. Adjust them together while watching rear-vs-front wheel speed and the total-ratio channel at each downshift, not separately.
Limits of the shift-point model.
- The equal-power criterion uses steady-state dyno power. In a transient (turbo spool after the cut, inlet-pressure dynamics, ignition retard during ramp-in) the post-shift power is lower for the first 0.1–0.5 s. With turbo engines this pushes the real optimum later than the static one.
- It ignores where the shift happens. A shift during a kink, or during the high-lateral-g end of an exit, costs more than its gap (torque step, line disturbance). Sometimes the fastest shift point is the one that moves the shift onto a straight.
- Gear-dependent cut times, and fuel and energy limits (Le Mans fuel stints, hybrid energy caps), add constraints that can make a short-shift strategy faster over a stint even though it is slower over one lap.
Key formulas
Trackside takeaways
Shift where power at current rpm equals power at post-shift rpm — often below the limiter.
After an upshift, long g should resume on the pre-shift trend; a step down means the shift was too early.
Measure shift time from the long-g dip; log long g at ≥ 50 Hz.
Insufficient blip = rear braking spike on entry; excessive blip/early downshift = over-rev.
Downshift lock-out events in the log mean the driver is downshifting too early.
Rising shift times over a stint are an early gearbox/actuator failure indicator.
On your car
Pneumatic or electric paddle shift with auto-blip and ignition cut; cut/blip calibrations largely homologated. Engineers set shift lights per gear and monitor shift times and refused downshifts.
Check yourself
After each upshift the longitudinal-g trace recovers to a value clearly below the extrapolated pre-shift trend. What does this tell you?
- Segers – Analysis Techniques for Racecar Data Acquisition — §6.1.1 Shifting Point · area under power curve, F3 example, long-g trend check
- Segers – Analysis Techniques for Racecar Data Acquisition — §6.1.2 Shift Duration · shift time from long-g, typical durations, sampling rate
- Segers – Analysis Techniques for Racecar Data Acquisition — §6.2 Down-shifting · blipping, over-rev causes