Shift Debugging

Troubleshooting shift system issues — missed shifts, false neutrals, compressor faults, paddle problems, and GCU board diagnostics.

Troubleshooting steps for techs and crew diagnosing shift-system issues. For pin-level electrical reference (voltages, pull-ups, fail-safes), see Shift Harness and GCU. For firmware updates and the serial console, see Updating GCU Firmware.

The page is organized by symptom. Pick the closest match from the index below and jump in — each branch ends at either a fix or a deeper sub-test.

Where to Start

Shifts

• Both paddles dead and compressor won't run → Nothing Works: No Shifts, No Compressor

• One paddle direction dead, the other works → One Paddle Direction Doesn't Respond

• Both directions miss intermittently → Missed or Incomplete Shifts (Both Directions)

• One direction noticeably weaker (different sound, larger pressure drop) → Asymmetric Shifts: One Direction Weaker

• Can't find neutral with the engine running → Can't Find Neutral with the Engine Running

Compressor

• Runs continuously, can't reach 90 psi → Compressor runs continuously

• Never runs at all → Compressor never runs

• Short-cycles rapidly → Compressor short-cycles

Intermittent / power

• GCU shifts fine for a while, then dies; full car restart recovers it → Nothing Works step 3 (power-supply cap solder joints)

• No LEDs at all, Flasher can't open the USB serial port → Nothing Works step 5 (Mega 2560 swap)

• FALLBACK: Using safe default timing in the serial log → Pressure Sensor Fail-safe

Reference

• LED Reference · GCU Board Reference · Reading the Serial Console

• Shift Harness and GCU (pin-level electrical) · GCU Firmware (flashing, console access)

LED Reference

The GCU has three diagnostic LEDs on the board plus the onboard heartbeat LED. Their meaning depends on state:

LED	Idle (no shift in progress)	During a shift
Red	Off	Solid during downshift attempt
Yellow	Flashes ~2 Hz while compressor running	Solid during neutral attempt
Green	Off	Solid during upshift attempt
Heartbeat (D13)	Pulses continuously	Pulses continuously

On firmware 3.0.5 and later, red → yellow → green flash in sequence at boot (~100 ms each) as a power-on self-test.

During a neutral attempt, the yellow LED lights solid together with the direction LED (green for NT+up, red for NT+down), so yellow-plus-green means "neutral upshift attempt."

If the heartbeat LED is frozen, the firmware has hung. A cold power cycle is always the first step before chasing electrical faults.

In the rest of this page, Gen1 means KLine / throttle-cable cars and Gen2 means CAN / drive-by-wire cars (see GCU Firmware). A Gen2 GCU will run in a Gen1 car well enough to debug paddles and compressor behavior, but it won't drive the Gen1 throttle cut or blip — treat any cross-generation substitution as a paddock-only diagnostic, not a track-ready swap.

GCU Board Reference

GCU 2.0 DBW board, annotated. GCU 2.5 units have a single top board with two black connectors; component positions otherwise match.

Quick map of what lives where on a GCU 2.0 DBW unit (Gen2, pictured):

• Top green board — MOSFETs. Actuator drivers; the terminals along the bottom of this board carry valve and compressor relay signals.

• Middle board — Arduino Mega 2560 with the intermediate board (also called the proto-shield) mounted on top. The intermediate board carries:

  • Three black paddle-input filter capacitors (up, down, neutral) between pins 9/10/8 on the input side and pins 19/18/20-or-3 on the logic side.

  • A large blue electrolytic capacitor (labeled P4 35V) filtering the 12 V power input. This is the "power-supply cap" referenced throughout the troubleshooting sections below.

  • Resistors and smaller caps forming the PWM → analog voltage network for the gearshift-sensor mimic (Gen2 only; unused on Gen1).

• Bottom board — pin breakout that sits on top of the proto-shield and brings the Mega's digital and analog pins out to labeled connectors.

Nothing Works: No Shifts, No Compressor

Both paddles dead and the compressor won't kick on even with air demand. The fault is upstream of the per-direction hardware. Bisect in this order:

1. Drain the air tank and watch the compressor. The GCU commands the compressor on below 70 psi via the relay on pin 11. An empty tank should trigger the compressor within a second or two. If it doesn't, the GCU isn't making the relay call.

2. Check the heartbeat LED. No pulse = firmware not running. Signal fuse, brownout, or a board-level failure — check the signal-fuse feed first (see Shift Harness and GCU).

3. Suspect the GCU power-supply electrolytic capacitor (the large blue cap labeled P4 35V on the intermediate board; see the board reference). A failed power-supply cap produces exactly this pattern: no shifts, compressor won't run even with an empty tank. The rails sag under any load, so every subsystem misbehaves simultaneously. Visual check for bulging, leaking, or discolored electrolytic caps; replace if suspect. This is the most common cause once the fuse and heartbeat LED are ruled out. A related failure on the three-board Gen1 / Gen2.0 stack is a loose solder joint at the same cap — that presents as a GCU that shifts cleanly for a while, then drops out completely and requires a full car power cycle to recover. Reflow the joints if you see that intermittent pattern.

4. Pressure sensor stuck-high fail-safe. If the sensor wire is open or shorted to 5 V, the GCU reads pressure as high and keeps the compressor off on purpose (safe failure mode). See Pressure Sensor Fail-safe below — the drain-tank test already rules this in or out, because a stuck-high sensor means the compressor won't come on even with the tank drained.

5. No LEDs at all and the Flasher can't open the USB serial port. The Mega 2560 itself is dead; the surrounding boards can't be the cause if the Mega doesn't enumerate over USB. The Mega is interchangeable between Gen1 and Gen2 GCUs, and a stock Arduino or Elegoo Mega 2560 clone (cheaper, fully compatible, easy to source on Amazon) drops in directly. Lift the top boards off, unscrew the Mega from the chassis, fit the new board, and reinstall the top boards.

Can't Find Neutral with the Engine Running

If you can get into neutral with the engine off but not while the engine is running, the issue is almost always clutch-related rather than a GCU fault. A neutral shift fires a short half-length pulse to the gearbox (25 ms up-to-neutral, 28 ms down-to-neutral) and, unlike a regular shift, applies no throttle cut and no blip. If the clutch isn't fully disengaged, the dogs stay loaded and can't slide through the half-position.

Two things to check:

1. Push the clutch pedal all the way to the floor when attempting to reach neutral. Any partial release keeps enough drive on the gearbox to prevent neutral engagement.

2. Adjust your clutch cable to pull the clutch higher. If the cable is too loose, even a fully-depressed pedal won't fully disengage the clutch. See Adjusting Throttle & Clutch Cable.

Also confirm input technique: the GCU fires a neutral shift only when the neutral paddle is held while you pull the upshift or downshift paddle. Tapping neutral by itself does nothing. The neutral input is polled (not interrupt-driven), so a momentary tap that's simultaneous with a direction paddle can also miss — press and hold neutral first, then pull the direction paddle.

Car Won't Start — Dash Shows N but Engine Won't Crank

A common variant: the dash reads N but the starter refuses to engage. The gear-position sensor is reporting neutral while the gearbox is still partially in 1st or 2nd (dogs not fully cleared). The starter interlock sees "not neutral" and inhibits cranking.

To free it:

1. Wiggle the car to unload the gearbox. Even an inch of roll usually lets the dogs settle into the neutral detent. Easier with a friend pushing while you watch the dash.

2. Push the clutch and tap the upshift or downshift paddle to settle the position, then back to neutral.

3. Check the gear-position sensor wiring if the dash never updates regardless of car motion — see Shift Harness and GCU.

One Paddle Direction Doesn't Respond

When only one direction is dead and the other works normally, the firmware is fine (both sides share the same code path) and the fault is somewhere in the paddle → wheel → coil cable → harness → GCU chain. Work from the outside in with swaps; each swap isolates a different layer:

1. Swap the up and down signal wires at the coil cable or the GCU-side connector. If the fault follows the wire to the other direction, the problem is in that wire / pin. If the fault stays on the original direction, the problem is upstream (switch, wheel-side wiring) or downstream (GCU itself).

2. Ground the paddle pin directly at the GCU connector. Jumpering pin 18 (up) or 19 (down) to ground simulates a paddle press. If this produces a shift, everything from the GCU inward is fine — the fault is in the switch, wheel, or coil cable.

3. Bypass the filter capacitor (up/down only). Up and down paddle signals pass through filter caps on the intermediate board — pin 10 into 18 (up), pin 9 into 19 (down). Jumpering the input pin directly to the logic pin (10→18 or 9→19) bypasses the cap. If the direction comes back, the cap is the culprit. This test is not documented for the neutral line; see Shift Harness and GCU for the reference procedure.

4. Swap the entire wheel and coil cable. If a spare known-good wheel is on hand, this isolates everything from the paddle down to the GCU's harness-side connector in a single test. Often the fastest way to confirm a wheel-side fault.

Filter Capacitor Failure Modes

The three black caps on the intermediate board (see the board reference) sit between paddle input (pins 8, 9, 10) and logic input (pins 3-or-20, 19, 18) and reject transient noise. The up/down bypass procedure above is the fastest test; failure signatures are:

• Open cap: logic pin sees unfiltered noise. Spurious shifts or missed presses, often worse on rough tracks where vibration-induced chatter gets through.

• Shorted cap: logic pin stuck near 0 V. GCU sees a continuous paddle press; the first shift fires, then further presses are suppressed until the internal state resets. Feels like "one shift then nothing."

• Leaky / degraded cap: slow charge/discharge. Presses register with variable delay or get dropped.

The bypass test (step 3 above) is the fastest confirmation.

Asymmetric Shifts: One Direction Weaker

One direction misses shifts noticeably more often than the other — and, importantly, sounds different from the cockpit. The fault is in the valve block or the air path feeding one valve, not in the electronics. Firmware runs both directions through identical code.

Diagnostic steps:

1. Listen. A healthy shift is a sharp crack from the valve block; a weak one sounds muffled or extended. A persistent hiss after the shift indicates that valve is venting past its seat.

2. Compare the pressure drop per shift. A normal 125 ms shift at ~80 psi consumes roughly 4 psi of tank air (about 3 psi at high-pressure / 90 ms shifts on firmware 3.0.6+). Read the drop off the AiM dash or analog gauge. Much larger drop on one direction = that valve is venting too much; much smaller drop = that valve isn't fully opening.

3. Inspect the valve block Amphenol connector before pulling the block — see the next section.

4. Replace the valve block if the Amphenol is clean and the asymmetry persists. The valve block is a single unit with three solenoid valves (upshift, downshift, throttle blip — the blip valve is unused on drive-by-wire cars), air in, two or three outputs, and shared electronics.

Inspecting the Valve Block Amphenol Connector

Before pulling a valve block, check the small Amphenol connector on the block itself. We've seen a pin sit unseated inside the connector housing — the shift was intermittent, but nothing on the wiring side looked wrong. Procedure:

1. Disconnect the Amphenol connector from the side of the valve block.

2. Visually inspect the pins on the block-side half of the connector. An unseated pin sits proud of its housing instead of flush with its neighbors.

3. If a pin is proud, push it fully home with a small blunt tool until it seats flush with the others.

4. Reconnect and test. If shifts return, done.

If all pins are seated correctly and the block still misbehaves per the asymmetric-shifts symptoms, the valve itself is the likely cause.

Missed or Incomplete Shifts (Both Directions)

Both directions occasionally misfire, typically under load or at low pressure. If the misfires are clearly biased to one direction, skip to the asymmetric-shifts section above instead. Likely causes in rough order:

1. Low tank pressure. Firmware 3.0.6 adapts shift timing to tank pressure (90 ms above 150 psi, 125 ms in the normal 70–90 band, up to 175 ms below 30 psi). But a tank stuck well below 70 psi still produces weak shifts. Watch the yellow LED after a shift burst — if the compressor can't catch up, you're shifting faster than the tank can refill (~1 psi/sec in the 70–90 band). Points back to compressor sizing, an air leak, or an asymmetric valve.

2. Countershift disabled or mistuned. Countershift (a brief reverse pulse after each shift to seat the dog ring) is on by default in both directions as of 3.0.4, and configurable per direction in 3.1.0. If countershift was turned off experimentally, turn it back on before chasing other causes. Contact Rush Auto Works before changing countershift settings.

3. Shift cylinder / gearbox. If air is healthy, countershift is on, and asymmetric-shift checks come back clean, the mechanical side is next — worn cylinder seals, stuck piston, or a shift fork issue. Inspection requires pulling the shifter.

Compressor Issues

The yellow LED flashes ~2 Hz while the compressor is running (firmware 3.0.5+). Hysteresis is hard-coded: compressor on below 70 psi, off above 90 psi.

Compressor runs continuously

The tank can't reach 90 psi. Quick math for context: the Kleinn 6270C compressor fills at about 1 psi/sec in the 70–90 band, and a single 125 ms shift at 80 psi consumes roughly 4 psi, so recovery after one shift takes about 4 seconds. If the compressor runs indefinitely with no shifts happening, look for leaks first — spray soapy water on fittings, the valve block seams, hose crimps, and cylinder seals under pressure and watch for bubbles.

If the air system is tight and the compressor still won't cycle off, a MOSFET stuck open on the GCU can hold the compressor relay closed regardless of pressure or GCU command. Track-side workaround: redirect compressor control from MOSFET 5 to MOSFET 6 — move pin 11 off the board to MOSFET 6 control, and move the orange and black wires on MOSFET 5 output across to MOSFET 6 output. The rest of the GCU stays functional until the board can be repaired.

Compressor never runs

Yellow LED never flashes: either the GCU thinks pressure is already above 90 psi (stuck-high sensor, see Pressure Sensor Fail-safe), the compressor circuit is dead, or the GCU itself isn't running (see Nothing Works: No Shifts, No Compressor above). Drain the tank to force the compressor demand — that one test separates sensor vs. board vs. compressor. If the yellow LED is flashing but the compressor stays silent the GCU is making the call, and the fault sits on the power side. Work cheapest to most involved:

• Fuses. Check the compressor / ABS fuse before assuming a board fault.

• Compressor supply harness. The harness feeding power to the compressor can work loose or go intermittent — reseat the connector and check for vibration damage at the crimps.

• Black 4-wire relay. We've seen this relay fail. The four terminals sit at N, S, E, W: N = red, E = red/black, S = red, W = black/black. N and S are interchangeable. See the shift harness wiring diagram for the full circuit.

Compressor short-cycles

Rapid on/off typically means a slow leak drops pressure below 70 psi shortly after each fill, or the pressure sensor is noisy near the thresholds.

Finding an air leak

If the compressor is working often even when you're not shifting, the system is leaking somewhere. Standard approach:

1. Pressurize the system, then turn the car off and listen — a loud leak is audible.

2. Spray soapy water on every fitting, valve block seam, hose crimp, cylinder seal, and Schrader valve. Bubbles betray the leak.

3. Purge the air tank a few times (car off!) by pulling the purge valve — it has a small pull that looks like a key ring. Purging clears moisture or contamination that can hold valve seats open.

Pressure Sensor Fail-safe

The GCU validates every pressure reading before using it for timing. Readings rejected: NaN, infinite, negative, or outside roughly −10 to 250 psi. Invalid readings cause dynamic timing to be bypassed; the GCU falls back to the fixed 125 ms shift delay used before 3.0.6. Shifts still work; they just don't adapt to pressure. Repeated FALLBACK: Using safe default timing due to pressure sensor failure lines in the serial log point at the A6 sensor wiring.

A sensor that reads stuck-high pulls the compressor off (safe failure — you can't overpressure the system). A sensor stuck low would drive the compressor continuously; the pull-up on A6 prevents this on a clean disconnect, but a pinched wire that shorts to ground can still produce that symptom.

Reading the Serial Console

Connect at 115200 baud via the GCU Flasher's console (see GCU Firmware). At boot you'll see:

Processor is working in main loop.
Firmware 3.1.0
Running in CAN mode.

(KLine mode on Gen1 cars.) During normal operation the GCU prints a line for every paddle press and every completed shift. Useful patterns:

• upshift attempt / downshift attempt / NT upshift attempt / NT downshift attempt — the GCU detected the paddle input. If you pull a paddle and see nothing, the switch, wheel wiring, coil cable, or filter cap is at fault, not the valve block.

• Shift completed in X ms. Pressure: Y PSI. Shift Delay: Z ms... — a full shift ran to completion. If you see this line but no gear change, the mechanical side (valve block, cylinder, shift fork) didn't follow through.

• Periodic tank_psi = N — sanity check on the pressure reading. Values stuck at 0 or near the sensor max (~150) indicate a wiring fault; compare to the analog gauge or AiM dash.

• Writing X.XXV to DAC pin (CAN cars) or Writing X.XXV to GSS pin — the GCU is driving the gearshift-sensor mimic to request blip/cut from the ECU. Absence of this line during shifts on a Gen2 car means the QuickShift request isn't reaching the ECU.

• FALLBACK: Using safe default timing... — pressure sensor reading rejected as invalid; see the section above.

• reset state — normal housekeeping at the end of each shift.

Leave the console connected during a test drive around the paddock to capture intermittent issues in context.