hexandboost

home / articles / a 140 hp automatic under turbo torque

Build series · Gearbox

Teaching a 140 hp automatic to survive turbo torque

The 5-speed automatic in the Civic Rebirth was designed for a 140 hp engine with no turbo. Mine now shifts at full throttle under boost. Getting there took more reversals than any other part of the car.

The problem in one sentence

During an upshift, a clutch has to absorb the engine's momentum while the engine is still pushing, and the harder the engine pushes, the more the clutch slips and heats.

Honda's answer is to close the throttle for a moment. On a naturally aspirated engine that works. On a turbo engine it does not, because the air under pressure is already past the throttle. I checked 18 boosted upshifts in my logs. In all 18, manifold pressure when the shift finished was at least as high as when it started.

The wastegate cannot help either. It takes about a fifth of a second to move, and the critical part of the shift lasts little more than a tenth.

Attempt one: hold the spark retard

Ignition timing acts on the very next firing. Honda already pulls a little timing during a shift. From 6 October 2026 I held that pull until the gears were almost synchronised and then eased it back in over about a sixth of a second. This part worked, and it has stayed on the car ever since.

Attempt two: more of everything

The same week I added more: fuel cut during the shift, a rich mixture floor, and a faster ratio change on the boosted upper gears. On paper it was thorough.

On the road it was worse. The fuel cut shortened the second-to-third shift by about 40 %, which sounds good and felt like a harder bang at the end. My own notes as the driver, 17 October: the clutch disengages, the car loses momentum, then it starts again. And on 29 October: shifts harsher.

On 2 November I made a ruling and wrote it down: keep the spark retard, keep the shift points, give the gearbox the right torque figure, and put everything else back to stock.

Most of attempt two came off the car. Not everything I build survives, and it should not.

What the gearbox did not know

That ruling contained the real fix. The gearbox decides how hard to clamp its clutches from the engine's torque, and its torque figure came from a model that stopped at atmospheric pressure. It was clamping for an engine that did not exist any more.

From 3 November the gearbox's torque estimate continues into the boost range. Boosted shifts now see 23 to 54 % more torque than they used to, and clutch pressure rises to match. In a logged boosted third-to-fourth, the gearbox saw about double the torque of a gentle one and asked for 23 to 30 % more pressure on the incoming clutch.

Attempt three: skip the sparks

From 6 November, on boosted upshifts the ECU skips whole firing events for a moment, rotated across all four cylinders so no cylinder is singled out. The throttle stays open, so the boost is still there when the next gear arrives.

It took three tries to get the amount right:

  • The first drive skipped 53 firing events across 12 boosted shifts. I wanted less shove as first-to-second landed, so I lengthened the cut.
  • I then made a skipped cylinder get no fuel either, so unburnt fuel was not going down the exhaust. That removed the pops, and with the pops gone the longer cut felt like too much.
  • So I shortened it again. It now covers the first 45 % of the ratio change.

It also bit me once. After a run on the rev limiter the car set a coil fault and ran on three cylinders. The ECU's coil monitor had counted too many missed firings on one coil. That led to a guard that keeps every coil firing often enough.

In EXTREME mode the cut is spark only, with the fuel left on. It is there for the sound and nothing else.

Other changes that stayed

  • Lock-up only in fourth and fifth. The torque converter no longer locks in the lower gears.
  • Shift points. I moved the gentle first-to-second shift down to 9 km/h. That night I decided it was too early and set it to 15. The stock figure is 16.

What the logs say

  • Across 70 open-throttle upshifts, 19 of them under real boost, there is no flare. The ratio never wandered more than 1.5 %.
  • A boosted first-to-second ratio change takes about 235 ms.
  • On 10 November, a second-to-third skipped 7 firing events and a third-to-fourth skipped 6.
  • Four full-throttle shifts at up to 12.7 psi are written up in full-throttle upshifts under boost.

What is still wrong

  • Boosted third-to-fourth falls flat. Right after the shift, the boost limiter closes the throttle to about half while my foot is still down. Fixes are built and not yet on the car.
  • The shift "hole" is bigger than before boost. The dip in acceleration during a shift has roughly tripled, in step with the torque.
  • Downshifts can jerk, probably because the turbo spools during the shift. I have not proven that from a log.
  • A low-speed stumble when you touch the pedal while the car is coasting down into first. The fix is in the current file. I have not yet measured whether it worked.

An automatic is a negotiation between an engine and a set of clutches, and I changed one side of it a great deal. The gearbox is alive and shifting cleanly at full boost. It is not yet as smooth as Honda made it at 140 hp, and I would rather say so.

By Haris, Hex and Boost.

Have a car the usual tools cannot touch?

Tell me the car, the engine, the gearbox and what you want it to do. I reply to every serious enquiry.