I've been tearing my hair out dealing with a problem with cylinder 2 knock and really need some help. 2007 civic si Fully built k20z3 rods, pistons, built head, cams. Basically fully built but stock sleeves. This is a boosted application, running a 6262 turbo. I'm getting tuned by imw and after doing the initial logs I was told to investigate cylinder 2 as it had a high knock count. So I pulled the coils, plugs switched things around... still cylinder 2 well I thought maybe injector? I mean they're brand new id1300 but I'll give it a shot. Still cylinder 2. Then I was thinking maybe it's my turbo manifold hitting the top of the fire wall so bent it back (note this maybe still a problem so I'm not ruling it out) but still no change. This is where it gets weird... I had an aftermarket knock sensor and I thought maybe it's a faulty sensor, so I bought an oem sensor. Yeah nothing changed... if anything it started picking up more. I just need some thoughts, I've checked for air leaks in the intake system, I will provide all information datalogs, spark plug pictures, photos of the engine bay. I just need help I'm kinda at a loss.
8th gen civic si ecu, running flashpro
(Note I am aware that the location of the sensor "hears" more from cylinder 2)
Build list
https://docs.google.com/document/d/1-GD ... p=drivesdk
Datalogs
Original datalogs sent to Derek
https://drive.google.com/file/d/19H7361 ... p=drivesdk
https://drive.google.com/file/d/1atoFEd ... p=drivesdk
https://drive.google.com/file/d/1ohXmWM ... p=drivesdk
Test driving with oem knock sensor
https://drive.google.com/file/d/1rX6q6U ... p=drivesdk
https://drive.google.com/file/d/1_y5Vo2 ... p=drivesdk
Spark plug pictures (cylinder 2 and 3 look a lot richer than 1 and 4, no indication of oil burning)
https://drive.google.com/file/d/1xMLD0C ... p=drivesdk
https://drive.google.com/file/d/1hyW15d ... p=drivesdk
			
			
									
						
										
						High knock count need help
- 
				saltyrabbit90
 - Posts: 14
 - Joined: Fri Mar 22, 2024 10:26 am
 
Re: High knock count need help
What I would do is to pull 2 degrees of ignition in the areas where you are getting knock (you must datalog, then make the ignition table follow the datalog cursor or position). If knock decreases: Then its real knock and the tune is just too aggressive. If no change? try pulling 2 more, then if knock count does not go down, then go back to the original value and look somewhere else. It is highly likely it's not a real knock, and just noise.