jcsbanks

This is what the owner just PM'd me:

"I have been out in the car with the det cans. I carnt hear any loud clicks etc but i can hear what i think is a lot of tappet noise its constant and can be heard at tick over and gets more intense the higher the revs get, and i must admit they are quite noisey do you think this would be picked up by the sensor?

I have also done a compression check and that is ok, Also the car ticks over nice and smooth "

Some say it could be the tappets and have seen this, some say they think it unlikely.

TTP Engineering

iTrader: (465)

With 16 years of lifter tap, I am sure that the Mitsu Engineers would not have an issue like this.

HOWEVER, I think this may call for a controversial KNOCK FILTER tweaking thread if anyone is so brave as to venture into that territory. Would be interesting to see nonetheless if this phantom knock could be "tuned out" using the knock filter in that specific rpm range.

The downside to that is that it appears in a location which is also known for peak torque, an area you WOULD want the proper protection from REAL knock.

mrfred

iTrader: (50)

Has a different OEM knock sensor been tried yet to determine if the knock sensor in that car is broken?

dan l

iTrader: (3)

Logging my '90 and logging a IX Evo it is pianfully apparent how the code seems to be nearly identical. The Evo's WILL pick up Lash Adjuster tick just like a DSM. Heck, the register location in the ecu is the exact same from a 1990 DSM to a IX ecu for the knock sum.

l2r99gst

iTrader: (2)

I'll second that.

Even in my 99 ECU, I would get phantom knock from lifter tick. Luckily DSMLink had a feature that let you ignore the knock sensor based on certain variables, such as TPS, RPM, etc.

But yes, I did get a lot of phantom knock just from noisy lifters. My GST had 117K I believe and the phantom knock started around 100K or so from the lifters. Used DSMLink to ignore knock sensor for low throttle/load in early RPM range and all was well.


Eric

TTP Engineering

iTrader: (465)

Jack and John, you reading this? Its a wonder why this wouldn't be a feature from the factory. What boosted car knocks under 100kpa?

If your rods are knocking you have a bigger problem.

We definately need a way to differentiate off load noise. This is a ridiculous problem to have with such advanced technology.

dan l

iTrader: (3)

Actually in the DSM code the ecu would ignore knock, however it was at a super low level. It was so low that it only really ignored knock when the car was idling. This was probably only done to keep the car from yanking 20 degrees of timing and stalling the car.

MalibuJack

iTrader: (5)

I wouldn't be surprised if there was more... After seeing the most current partial disassembly, there are nearly a dozen new data items that need to be defined for ECUFlash.. Not to mention quite a few more that arent documented yet. I do recall seeing a few specifically related to knock control and thresholds.. but I need to scrutinize the new data a little closer..

John posted his EUDM IX rom's disassembly in progress on aktivematrix if anyone wants to look at it and see if any of the data looks familiar. John did notice that quite a bit of the items are similarly located (if not the same) for many of the roms 8's and 9s included.

jcsbanks

Sorry TTP, time zone differences make it difficult to reply promptly.

The knock control seems to be disabled until you are almost out of vacuum from this log 0,0,0,0,11,36,36,36, for example.

I've not found knock filter stuff like the US models in the UK/JDM ROMs. I had a brief look. Trouble is with 12 or so tables in some of the models, which to adjust. The knock control code is convoluted in the extreme.

If we get custom comms going I might be able to setup a way of telling which maps are being accessed.

MalibuJack

iTrader: (5)

I'm fairly certain the reason its convoluted is I think their all used simultaneously.. I think each map represents a range of frequencies to discriminate since its the combination of frequencies combined, and not the total voltage of the signal, that determines the difference of a plink and a plonk if you know what I mean.. This is why there are situations where 1.5v might trigger knock in some circumstances, and not in others.. It allows the ECU to ignore background noise and focus on frequencies where there are signature spikes that indicate knock.. Altering the displacement of the engine alters these resonance frequencies and therefore the new noises are unfamiliar to the ECU and can fall into map where a range of frequencies are more sensitive..

jcsbanks

I had imagined that the filtering was analogue (excuse my quaint English spelling, my favourite colour is grey and it will put me in gaol as I run through my tyre collection).

I'm trying to think how an interrupt driven/loop driven system would have the ability to process an analogue signal and break down the frequency components without using DSP and large buffers to store the sample after each spark - I see no DSP opcodes in there.

MalibuJack

iTrader: (5)

It would probably pass through a filter that blocked all frequencies but the range in the specific map.. and do it sequentially until it made it through all of them.. considering the speed of the ECU I suppose its "fast enough" to get acceptable results..

And their probably not DSP routines, but signal traps on different inputs.. this would allow the ecu to quickly cycle through each range and test it.. Its simple enough to have several filters to attenuate the signal for each specific range and not have to do it programmatically.

burgers22

Hi

have a look at post 14, it links to a document about knock detection, where the document details knock filtering, analouge and digital. Might be of some help determining how the ECU filters knock.

Knock thread

Also here Tom Collins makes a brief metion of how the ECU+ does its knock detection, the link above comes from this thread

ECU+ knock

MB

Ultimate CC

iTrader: (122)

this looks like the exact same issue I am having with my evo right now...

4TUN8

iTrader: (18)

yes it does but does your car have forged guts like this one does?