LNF False Knock- Discovery might lead to cure???

[gmtech16450yz]

Ok I was getting a little tired of trying to determine false knock from real knock on my Redline so I tried something and got interesting results...

In the SI knock sensor replacement info it says this...

Installation Procedure
Important: Rotate the pigtail 90 degrees from vertical before securing the fastener.

Ok so I take a look at my sensors and one is at 90 degrees and the other one is just about vertical. I loosened both of them and turned them about opposite of where they were, not necessarily at 90 degrees. Drove home and guess what, massive amounts of KR, obviously false, and mostly from cyl 1.

So here's the homework assignment for anyone that's interested and is also fighting false knock, try rotating your knock sensors and lets see if we can figure this out. My next step is to figure out if the false knock showing on cyl 1 is actually because the front sensor. The computer knows what cylinder is firing when the knock occurs so it might not depend entirely on which sensor picks up the knock. I would assume the front sensor picks up cyls 1&2 and the rear 3&4, but that might not be the case. I figure I'll pull the sensor out and leave it plugged in but insulated so the computer stills sees it's there but it won't pick up anything.

Wouldn't that be cool if the false knock was caused by the sensors being "clocked" wrong. That might also explain why some cars have a ton of false knock and others have none. I've changed a lot of knock sensors and never thought about putting them back in a certain position, although most sensors will only go on one way. The mention in SI must mean they've traced some of the false knock problems to sensor positioning. If we can't get anywhere by trial and error rotating and logging, maybe I'll write a repair order so I can call TAC and see if they'll let me in on any secrets.

BTW, the illustration in SI shows the sensor pigtails facing forward on both sensors.

TIA for any feedback on this! Let's figure it out!

[sharptech]

hmmm you might be on to something here

[pantherqs]

Totally fucking with this tomorrow all day.

[gmtech16450yz]

Totally fucking with this tomorrow all day.

Lol, that was hella funny!

Tried something tonight, drove with both plugged in, then unplugged only the front sensor, then only back, then both unplugged. Definitely made huge differences in the logs. I now know it's my front sensor that is reading WAY too much knock, because I rotated it.

With the front unplugged I got 2-3 degrees KR, mostly on cyls 1&2.
With the rear unplugged I got 12 degrees KR (!), mostly on cyls 3&4.
With both unplugged it defaults to max (12) KR on any kind of accel or load.

Soooo, what I'm getting from that so far is with no sensors the computer is going to go full retard no matter what. (makes sense.)
With the front sensor unplugged, when the working back sensor sees any amount of knock or the motor is under a certain amount of load, the computer will put in, or show, more KR for the missing sensor's cylinders (front 2). (also makes sense.)
With the back sensor unplugged, and the front one WAY too sensitive because of it being "clocked" wrong, it puts in or shows a ton of KR on the cylinders with the missing sensor (3&4). (makes sense again, amazing!)

It does appear that the physical location of the sensors is biased towards the cylinders it's closest too, meaning KR in the logs showing it's on cyls 1&2 means the front sensor is picking up knock. That's what I always assumed, just wasn't sure if there was more to it than strictly which cylinders were closer to the sensor that picked up knock. Actually, there has to be more to it, how else would it know if it's cylinder 1 or 2 with the sensor between both? It has to go off of cylinder timing vs. knock detected, but leaning towards the cylinders closest to the sensor that detected the knock.

The next thing I'm going to do is rotate the front sensor 90 degrees from vertical like SI says, with the lead facing forward.

Getting closer! There's hope! (Unless we figure out how to "clock" the sensors for the lowest knock but it's still too sensitive.)

Thanks guys.

[Gimpster]

Totally unrelated to this, but on DSM's we have to be extra careful how tight the knock sensors are torqued... the clocking itself didn't matter though. If you had the knock sensor too loose or too tight it made a big sensitivity difference.

[shabby]

Totally unrelated to this, but on DSM's we have to be extra careful how tight the knock sensors are torqued... the clocking itself didn't matter though. If you had the knock sensor too loose or too tight it made a big sensitivity difference.

Twas not the tightness of the knock sensor but the will of the dsm gods that blessed you with phantom knock... they also cursed my gxp with it
I'll definitely give this a try in a few days, did you manage to get a torque wrench in there or did you feel the 18 ft/lbs?

[cmiller8006]

Are the knock sensors the two things next the the coil in the valve cover?

[Iam Broke]

I believe those are the cam phasers.

The knock sensors are on the front of the engine between cyls 1&2, & 3&4. Down under the intake mani by the injectors??? Not sure.

[gmtech16450yz]

The cam phasers are on the front of the cams, the ones on top of the motor are the cam position actuator solenoid valves (that's really what gm calls them) that control oil flow to the phasers.
The knock sensors are directly under the intake, right below the foam insulating material around the injectors. Not sure about the Cobalts but on the Kappas you need a mirror to see them without taking anything off. Pulling the dipstick tube off will make it pretty easy to get to both sensors.

Shabby, no I didn't use a torque wrench, after about 35 years of wrenching my hands are pretty well calibrated! I do use a torque wrench for critical stuff though. If anything they're tighter than 18 ft/lbs. I will try super tight and just barely snug when I get back to messing with it.

Voltage readings on Tech 2 of sensor 1 and sensor 2 were very different on the two LNF's I worked on today. On the one car sensor 2 was consistently higher voltage, on the other sensor 1 was higher voltage. I might try putting a dvom directly on the sensors and see if I can get consistent voltage readings with a constant vibration or noise put on the block. (hammer drill on low setting comes to mind) If it works, I can then rotate or tighten the sensors and see if the voltage goes up or down.

[cmiller8006]

Well I would of f'ed up the car lol.

[Japeatr]

I might try putting a dvom directly on the sensors and see if I can get consistent voltage readings with a constant vibration or noise put on the block. (hammer drill on low setting comes to mind) If it works, I can then rotate or tighten the sensors and see if the voltage goes up or down.

this caught my attention...let us know what you find.

[gmtech16450yz]

I'm not ready to break out the champagne yet but NO FALSE KNOCK FROM FRONT SENSOR ON LAST 3 DRIVES!!! (knocking on wood)

Last night I rotated the front knock sensor so it was exactly 90 degrees from vertical with the lead pointing forward. This is 180 degrees from where it was from the factory. Before I moved it I checked the torque, swear to god when I got my torque wrench adjusted up to where the bolt just started to turn I looked down and it was at 19 ft/lbs! I torqued it back to 19-20 ft/lbs after rotating it so the torque should be close to the same.

Drove it for half and hour last night and the log showed no KR from cyls 1&2. Drove it back and forth from work today and still nothing that I would call false knock from 1&2. I did get 3 degrees from cyl 1 on a hard run up to 6k, it definitely looked like real knock. Meanwhile cyls 3&4 are still showing a bunch of false knock, the lead on that sensor is pointing down, tonight I'll rotate that one to 90 from vert with the lead facing forward. If the false knock on cyls 3&4 go away also I'm going to be jumping up and down with joy!

BTW I put a knock sensor I had in my vise, put a dvom on it and tried to get some sort of consistent test I could use on the car, but judging from the erratic voltages I was getting it didn't really look like it was worth messing with. So I busted open the knock sensor to see what's inside! Basically just two thin metal plates with ceramic in the middle. Didn't really see what could make it directional, and even some info I found from Bosch says they can be mounted in any position. ???

Crossing fingers... I'll let everyone know what happens when I rotate the back one.

[PSYKOSTEVO]

.

Soooo, what I'm getting from that so far is with no sensors the computer is going to go full retard no matter what. (makes sense.)
With the front sensor unplugged, when the working back sensor sees any amount of knock or the motor is under a certain amount of load, the computer will put in, or show, more KR for the missing sensor's cylinders (front 2). (also makes sense.)
With the back sensor unplugged, and the front one WAY too sensitive because of it being "clocked" wrong, it puts in or shows a ton of KR on the cylinders with the missing sensor (3&4). (makes sense again, amazing!)

That made me think of the movie Tropic Thunder:

[mkriebs]

Intriguing!

[Iam Broke]

I'll crawl under mine this weekend to see what's what. I generally see more on #4, then #1 is a close second. Hoping to find mine 90* out or so.

[pantherqs]

Well, instead of fucking with this I A.D.D. kicked in and I hit mine with a 100 shot. Glorious results.

[mkriebs]

Ummmm, I hope you are kidding.




Might take care of the knock though, hahaha.

[Jaxon]

lmao, you run a dry shot or wet? so your running a 100 shot on top of spiking 30psi? lol

[pantherqs]

Wet, boost is fixed, replaced the busted wg solenoid. I'll up some vids tonight.

[mkriebs]

Got any dyno sheets of it yet?

Got a parts pile to build the motor? Lol.