Rage

The questions I have is with the knock analysis/control issue. Both the Merc and GM ECU's have been set up by Merc? and by GM? engineers with a calibration and a decision algorithm to process the raw knock sensor signals and to "decide" what is a "knock" and what is not and then what is bad enough "knocking" to require protective action.

The knock sensor wil be picking up all kinds of vibrations from a running engine with the knock event vibration mixed in. This signal would seem to be worthless with out means to interpret.

Both the GM and Merc ECU's can be hooked up to an Innovate LM-1 o2 recorder/reader or other data logger to record real time engine sensor ECU input signals and ECU out put signals for later analysis. What I have been told is that the GM ECU generates a "signal" that says when a "Knock Event" (per its internal programming analysis) has occured and the Merc ECU does not. It would seem comparatively easy to use trial and error analysis to tune out an engine knock with a GM ECU versus with a Merc ECU.

Is there a means to tap into the Merc PCM555 ECU and data log real time the "signal" that says when a "Knock Event" (per its internal programming analysis) has occured????????????????

The Merc ECU unit I understand will store a record of "Knock Event" "trouble codes" that can be accessed at a later date with various electronic trouble code readers. These "Knock Event" trouble codes I do not believe can be associated with the real time engine sensor, A/F ratio or other conditions at the time of the " Knock Event". This is better than nothing but stone age compared to the GM ECU real time "Knock Event" output.

I have a Merc PCM555 unit and for various reasons would prefer to stay with it. Any quidance that can be provided on how to better/easier identify "knock Events" in the process of tuning for engine changes, I would sure like to hear it !

Pat McPherson

I know that most of us are Merc guys and the subject of this thread is about the Merc ECU, but how about the system that Volvo Penta uses.

I have heard that for 06 VolvoPenta will be using Sequential Port Fuel Injection with a new 'E Controls' ECM vs. the old MFEI setup.

How will this compare to Mercs fuel injection?
Better or just different?

Rage

Anybody?

800XCR

How are you able to make changes to the PCM555 ECU?

Dave_N

I don't know if any of this will help you, but after giving it a lot of thought (and reviewing a lot of data), I think the only way you could monitor knock sensor activity (real time) in a PCM/ECM555 system would be to use a lab scope with an inductive low amps probe on the circuit. Even then, it would require some time to figure out what to look for (I have the equipment, but have never attempted to do this).

I tried to determine if knock sensor activity could be detected by looking for the resulting activity (retard) of the ignition timing. My results were inconclusive, but I did notice what may be evidence of knock retard in some WOT data recordings. Basically, I noticed that timing advance was usually very steady at cruise speeds, but started to flucuate a lot at WOT. Was this flucuation evidence of knock sensor activity/retard? I don't know for sure.

Have a look at the recordings below (from a 2003 496HO). The first two show steady timing advance at cruise speeds, and the third one shows the fluctuation at WOT.

Dave

stevesxm

good stuff. looks like the data aquisition that we used in the cars.

not certain i agree w/ your conclusion however... first trace shows timing at 26 ish just right.... second trace shows timing at 23 ish but nice steady state ... just what you would expect to see .... a small linear retard at higher rpm.

third trace shows esentially same timing at essentially same rpm as trace two except trace has some hysterisis or noise.

normally if the knock sensor 'trips" and the ecu " believes" it
i.e it satisfies the parameters for a detonation even as defined by the programmer... then you will see an abrupt and virtually instaneous timing retard by some significant amount...4 or 5 degrees... sometimes more.

so.... IF trace three were a knock event, i would have expected to see the timing go to 17 or less instantly and remain there for some period of time defined again by the prgrammer until the operating parameters returned to nominal and it " thought" it was safe to return to preset timing.

everybody uses a different algorythm for this.... but one thing is absolute... detonation can damage the motor VERY quickly so i don't think you would see that sort of subtle correction that your trace 3 implies....

motors running that close to the edge really need exhaust gas temp sensors.... those tell you the real answers.

Dave_N

I thought that I was pretty clear that I was not drawing any conclusion...just making an observation.

You won't necessarily see dramatic evidence of knock retard by watching the timing advance, as you suggested. Consider the recording below from a MEFI1 system (where you can monitor knock retard). Even when you KNOW the knock retard is activated, you can see that the resulting effect on timing advance is not as clear.

Dave

stevesxm

conclusion poor choice of word.... so the total retard is 3.2 deg for a knock event ? sounds about right.... and you can get this data off the second system but not the first ? interesting.... why the dif in resolution bet the two systems... do you know ?

Rage

Dave,

Thanks for the useful information.

It is my understanding that when a "knock event" is detected per the programer's ECU software algorithem, both 1.) timing is retarded and 2.) the fuel delivery is enriched by longer pulse widths for the fuel injectors.

You have shown timing curves. Do you have any similar information regarding the injector pulse width that may be useful in identifying a knock event per the PCM555 ECU?

Dave_N

The knock retard displayed in the numeric display (3.2 deg) is the retard for the specific point where the vertical yellow line is. The retard went as high as 5.0 deg during the portion of the recording shown.

Yes, the second (newer) system does not have knock retard output in the data steam.


This one is my fault, the graph displays shown were captured using different resolution settings.

You are correct, but the ECM will fatten up the pulse width upon knock event before retarding the timing, I think thats why you don't see a direct, proportional and predictable change in timing advance.

No, the PCM/ECM555 does not output knock retard or injector pulse width, so watching the effect on timing advance is the only hope to "see" knock sensor activity in the real time data output (that I know of). Wish I could be of more help

Dave