Knock tables/routines unraveled (evo)

[scheides]

http://forums.evolutionm.net/showthread.php?t=382625

Two major things AEM EMS holds over the stock ECU at this point: speed density, and adjustable base knock thresholds. SD is being tested already, but the knock stuff is probably more important for most people. Why?. Some engines are louder than others. OEM components have different harmonic resonances than aftermarket ones. Heck, some OEM components have different harmonic resonances than other OEM components!

I will paraphrase, because it is a long-ass thread. The ECU listens to the knock sensor and digitally filters out background noise as time progresses. Using some newly discovered tables, the values used to determine how loud (and at what rpm) this background noise is can be adjusted!

This is a major, major, major breakthrough. Many people have shied away from built motors on the stock ECU simply because they can be 'louder' than OEM motors. No more will this be the case, and those people will be able to spend the $1500 and hours upon hours upon hours they spent on playing with AEM and put it towards making their car faster in other areas!

Discuss.

[scheides]

Interesting post, just for you Kracka:

Originally Posted by jcbanks

The adaptive bit is pretty quick, one filter will be done after 16 ignition events, then the next one after 8, so it doesn't take long to completely clear out any memory effect. I can't see any long term memory involved at all.

Like many of us have noticed, you often get knock on the first pull more than subsequent ones, I think most people put this down to carbon embers in the combustion chamber that build up during cruise and then act as knock promoters. Doesn't sound like this entirely explains how it only happens after a flash in your observations, but knock is an unpredictable phenomenon and sometimes I know I look for patterns/predictability that simply isn't there.

http://forums.evolutionm.net/showpos...&postcount=122

[scheides]

Here's what the tables look like:



The top one is rpm vs load. You can tell the ECU to simply ignore knock under a certain load/rpm range, up to a load of 160 (peak torque loads are 240-300+, cruise loads are 30-80).

[scheides]

And oddly enough, check out these patents:

http://www.google.com/patents?id=K5w...BAJ&dq=5419180

http://www.google.com/patents?id=K5w...BAJ&dq=5419180

They describes the code that jcbanks deciphered almost exactly, and guess who it was issued to? Mitsubishi, in 1995.

[asshanson]

Can you do more than just filter out below certain rpm/throttle? That would just seem to get rid of phantom knock at low load levels, and wouldn't solve the problem of a loud built motor.

[scheides]

A comment comparing AEM and the DSM ecu. Can anyone else comment on these?

Originally Posted by dan l

Wrong, it is MORE powerful than the most popular standalone for our cars in the US, the AEM. Also the processor speed in the OEM ecu is amazing. That it is working at every ignition event is REALLY fast.

This system is much more sophisticated and even proprietary to our engines from Mitsubishi. Even if a company wanted to mimic the functionality they can't, its patented. When properly tuned this system will provide the most protection bar none.

I've seen a AEM logs where knock voltage never went over 2 volts and a guy lost a motor, parts through the block on the top end of the track. The AEM system relies on a really good tune and the assumption that the user keeps up on the tune to keep the engine from grenading. Even at that, if a fuel/ignition parts failure occurs mid pull the AEM system can not always decipher knock correctly and save the motor. Another one is a local car that pushed the front balance shaft out when the rod broke off after piston damage an an AEM. For the average street guy who hands the keys to his sober friend to drive his drunk butt home from the bar the stock ecu is the best option.

I know its a bad example because the DSM system works fundamentally different (analog vs digital) to an evo. However the voltage spikes I believe are still picked up on the DSM which is similar to the Evo. At any rate I went to a T&T day last fall in my DSM. Made about 20 passes. All day long the car was either slow or fast but knocking. I couldn't figure it out, so I turned the boost up . Well come to find out when I got home (and the car stalled and pumped fuel through the motor, big fireball). I discovered that the cam angle sensor trigger had worked its way loose. The ignition was advancing +15 deg and retarding -15 degrees at will until the trigger plate jammed and stuck at TDC and stalled my motor. With a total swing of about 30 degrees the stock knock sensor system in my archaic DSM kept my motor alive, pulling timing when needed on a 130mph trapping stock motored street car. The evo system is even more sophisticated, so it should do a better job!

Andrew: Yes, using a combination of that and adjusting the background noise numbers will accomplish this. People are just starting to test it to see how well it works, but its only a matter of time now!

[Shane@DBPerformance]

The top table is similar to what people have done with eprom chips for the DSMs to avoid phantom knock during cruise.

[asshanson]

Scheides you have to be logged in to see the pic. Can you copy it to this server?

And this quote is awesome:

Originally Posted by dan l

I couldn't figure it out, so I turned the boost up

[scheides]

Fixed, sorry about that!

[FattyBoomBatty]

I guess I'll upgrade to the evo harness and computer once they figure out speed density mods for it. Then I wouldn't have to worry about this (I come to find out in this thread) crappy AEM blowing up my motor.