AEM Ultimate Resource Thread>>>>

[RobBeck]

Right. I never heard of problems with the NGKs either but I do not think the LM1 uses the NGK, it used the Bosch LSU4.2 or some crap like that. I guess the question should be what type of sensors does the AEM WB02s come with? I probably will go with the AEM sensors thanks, unless it seems to be an huge cost increase.

Thanks,
Rob

Originally Posted by mjannusch

I'm running my NGK sensor on the O2 housings. No problems so far. Lots of other people use them on stock O2 housings on various cars, including 3/S cars and I haven't heard of any problems with 'em.

[RobBeck]

Looks like the AEM Dual channel comes with Bosch sensors, not the NGKs like you have... I wonder if they will error out or become inacurrate like the LM1 does. Anyone on AEM Forum use actual Bosch WB02s with the precat locations?

[Type R]

I have used an AEM Dual Channel with bosch sensors (yes it came with it) in place on my o2 housings. From what I know, AEM recommends the sensors to be further away from the turbo due to extreme heat so if I had to do it again, I'd tap in bungs on my dp and run em there to save myself on those expensive sensors.

Other than that, the readings came out just cherry. They just need a GOOD ground on the unit and need to be calibrated correctly.

[RobBeck]

so i guess we can kick the narrowbands to the curve, and just put in the bungs further downstream if using the bosch sensors or just leave em in the OE locations if running the NGKs. The Bosch units are apparently REALLY tempermental with heat.

Rob

[FastOldGuy]

Does the AEM give a realistic knock count. I know the SAFCII counts knock but it is just raw knock so it is not run through an algorhythm so its is useless so I was wonder is the AEM knock the same as you would see with a stock computer and a dataloger or is it just Raw knock. I would assume its not just raw knock.

[pjayp]

Originally Posted by FastOldGuy

Does the AEM give a realistic knock count. I know the SAFCII counts knock but it is just raw knock so it is not run through an algorhythm so its is useless so I was wonder is the AEM knock the same as you would see with a stock computer and a dataloger or is it just Raw knock. I would assume its not just raw knock.

you get voltage reading from the aem. imho, this is way way better than any "knock count" given by a safc or factory computer. you can map what your engine does under non-knock conditions... and use that to determine when you are pushing things too hard.

[pacman3000gt]

Originally Posted by FastOldGuy

Does the AEM give a realistic knock count. I know the SAFCII counts knock but it is just raw knock so it is not run through an algorhythm so its is useless so I was wonder is the AEM knock the same as you would see with a stock computer and a dataloger or is it just Raw knock. I would assume its not just raw knock.

Realistic?

The EMS is infinitely configurable... you adjust what you would 'consider' a knock count. It's much better than relying on what the stock ECU's rendition of a 'knock count'.

You have a knock retard that kicks in once your knock parameters are exceeded. So it's more of a preventative measure when you compare it to what the stock ECU does.

NOW... this is entirely due to how accurately you have setup the knock calibration table... and more importantly... if the knock control is EVEN ON.

but back to what pjayp said... I almost never use the knock 'count'. I always check out the knock raw voltage.

[FastOldGuy]

Hmmm I assuming the AEM is reading the OEM knock sensor, correct? I guess I want to know how I decide when the knocks seen by AEM are important or not.

[mjannusch]

Originally Posted by FastOldGuy

Hmmm I assuming the AEM is reading the OEM knock sensor, correct? I guess I want to know how I decide when the knocks seen by AEM are important or not.

Yes.

You decide what the knock threshold is by running the car at lower boost through the rev range when you know the car isn't knocking (ie: provide plenty of fuel and use a conservative timing advance range), then review your log of raw knock sensor voltage, and then plot out a knock sensor threshold calibration that's a little higher than what you logged for raw voltage. After that, when knock raw voltage exceeds your threshold, the knock control routines will apply whatever timing and fuelling changes you set up whenever the "Knock Control" checkbox is checked.

Generally, anything over about 1.5V is almost guaranteed to be "real" knock. True knock will have spikes much taller than the surrounding samples.

One thing to be careful of is that the logger via the serial cable (realtime logging) is not fast enough to catch all the knock spikes, so you are only seeing some events and not all of them. To get higher resolution you can use the AEM's internal logger set on the fastest speed and then download the log to your laptop afterwards for review.

I generally watch my knock retard and knock fuel parameters to determine if I am getting knock beyond my threshold I set. Those decay slower, so you'll be able to easily spot them in logs and know that excessive knock occured right before the fuel/timing adjustment gets kicked in.

[pjayp]

Originally Posted by FastOldGuy

Hmmm I assuming the AEM is reading the OEM knock sensor, correct? I guess I want to know how I decide when the knocks seen by AEM are important or not.

you need to "map" what is not knock. turn the boost down, give it some good gas, keep the timing conservative... log some runs. you will generally get some voltage logging when the engine is under load. you want to use this to help create your knock table. just need to use a little common sense, there is not necessarily a particular voltage that "is knock". if at WOT at 10psi, and conservative timing at 6k rpm and you have a voltage reading of 0.32v. And then you go and bump the timing ten degrees and add 5psi you start getting readings of 3.4v, then I guess I would consider that knock.

I've seen people say 2.0v = knock. but personally I think if you pick a voltage to go by you are defeating the value of what the AEM brings to the table. fingerprint what YOUR engine does under conditions that shouldn't have "knock", then you use that to build a knock table.