2015 Subaru STI Emissions Pump Code and Idle RPM Issue
Having an issue with my 2015 sti. I usually have codes for the emissions pump failing causing a CEL and I’ll just reset them with the accessport and go on my way without the CEL for a few days. Did just that before leaving my house earlier and when I started my car it jumped to like 3k rpms cold idle for about a second or two and then started going down to normal ranges. Started driving and I was going into boost at like 5% throttle which is super weird, usually it’s around 13%+ throttle. Stopped off the road and noticed my idle boost was only like -3psi to -4psi. It slowly after driving a little bit and getting to Walmart would idle around -5psi to -6 psi. Never had anything like this happen before and I’m not even sure where to start. On the way home I went kinda hard into boost and hit 24psi which I never go over 22psi WOT. No codes for overboosting or anything, only code is P2431 which is a normal one for me. Anyone have any idea of what/where I should start looking? I’m thinking a vacuum leak of some kinda maybe? When changing gears the rpms seem to wanna linger high before dropping as well. Any help or info would be appreciated. Not even sure how to word this to google around cause it just brings up rough idle, etc. which I don’t have a rough idle.
Working hypothesis: you’ve got an unmetered air / boost control issue that showed up right after the AP reset, not an emissions pump problem. The low idle vacuum, early boost, higher peak boost, and hanging RPM all point more to a vacuum/boost leak or control offset than to the P2431. Runner‑up is a sensor/learning issue (MAP/MAF or boost control learning) that got whacked when you cleared codes.
Most likely culprits
- Vacuum / boost leak post‑MAF – Turbo inlet, BPV, manifold/TGV, or a hose off; gives low idle vacuum, early spool, weird throttle/boost relationship, and hanging revs.
- Boost control / wastegate control issue – Stuck/changed WG duty behavior after reset, or a line/solenoid issue letting it overshoot to 24 psi.
- Sensor / learning offset (MAP/MAF / A/F learning) – AP reset wiped learned trims, or a MAP/MAF starting to skew; ECU is targeting wrong load/boost until it relearns.
How to narrow it down
-
Pull live data on the AP or scan tool, warm idle, no load.
- Watch MAF g/s, AFR (commanded vs actual), A/F Learning A/B, MAP (psi), RPM.
- On a healthy EJ257 at hot idle you should see: ~2.5–3.5 g/s MAF, AFR ~14.7, A/F Learning A within about ±5–8%, and manifold pressure around ‑9 to ‑11 psi (depending on baro).
- If MAP is only showing ‑3 to ‑6 psi but engine sounds smooth, that’s either real low vacuum (air leak) or a MAP offset.
-
Verify boost reading with a mechanical gauge if you have one.
- Tee a known‑good gauge into the manifold.
- If the gauge shows normal vacuum (‑9 to ‑11) but the AP/MAP says ‑3 to ‑6, you’ve got a MAP/sensor scaling issue, not a vacuum leak.
- If both show low vacuum, chase a real leak / mechanical issue.
-
Smoke test the intake from turbo inlet forward.
- Cap the turbo inlet or compressor inlet, smoke it at low pressure.
- Pay attention to:
- Turbo inlet pipe (they love to split or pull at the PCV/EVAP ports).
- BPV flange and hose.
- TGV to head and TGV to manifold gaskets.
- EVAP purge line to manifold and any small vac lines on the manifold.
- Any smoke out of those spots explains low vacuum, early boost, and hanging RPM.
-
Check boost control hardware quickly.
- Inspect wastegate actuator line, T‑fittings, restrictor pill (if still stock plumbing), and boost control solenoid hoses.
- Look for a line off, cracked, or someone previously “tuned” it.
- A line off the WG or wrong routing will give early boost and higher peak (24 psi) without necessarily throwing an overboost code right away.
-
Log a controlled pull, 3rd gear, from ~2500–5500 rpm.
- Log Throttle %, Boost target (if your AP map shows it), Actual boost, WG duty cycle, MAF g/s, AFR, Knock correction.
- If actual boost overshoots target with high WGDC, that’s a mechanical boost control issue (WG, lines, solenoid).
- If target itself is higher than you remember, your map/learning may have changed or you’re on a different AP map than you think.
-
Check A/F Learning after a few drive cycles.
- After some mixed driving, recheck A/F Learning A/B/C/D.
- If one range is heavily positive (e.g. +15–25%), that’s a vacuum/boost leak in that load range.
- If trims are reasonable but MAP/boost readings are still off, lean toward sensor calibration (MAP) or tune issue.
-
Confirm the P2431 isn’t dragging anything else with it.
- Scan all modules, not just engine, for pending/history codes.
- Make sure you don’t have any other hidden codes (MAP, MAF, EVAP purge, etc.) that the AP screen might be glossing over.
Possible parts (if needed)
- Turbo inlet pipe – if smoke test shows leaks or you see splits/loose ports.
- BPV / BPV gasket or hose – if it’s leaking under smoke/pressure.
- TGV / intake manifold gaskets – if smoke is coming from TGV/manifold junctions.
- EVAP purge solenoid or hoses – if purge is flowing when commanded closed or leaking smoke into the intake at idle.
- MAP sensor – only if mechanical gauge disagrees with ECU/Accessport and wiring checks out.
A few questions
- What exact map are you running on the Accessport (OTS stage X, custom tune, etc.), and did you change maps or just clear codes?
- Do you have any mechanical boost gauge or only the AP reading, and have you ever logged A/F Learning before this to compare?
- Any recent work under the hood (intake off, TGV work, turbo/IC/BPV changes) right before this started, or has the P2431 been the only ongoing issue for a long time?
Get me a short log (idle + one clean 3rd gear pull) with the PIDs I listed and I can tell you pretty quickly if this is a leak vs sensor vs tune/learning.