john banks

Skimming the patents you linked Andy, I didn't notice anything about how knock is actually detected, but more the ignition timing determination and learning. My problem was always that the actual knock detection appeared fairly deaf on JECS which appears to have similar logic with different hardware. Mitsubishi also changed their CPU but kept similar knock control logic for about 20 years, but it seemed much better to my ears about detecting knock and could be trusted.

Welloilbeefhooked

iTrader: (3)

Ok, eventually got around to logging some data and found that the knock event count remains at a value of 1. Absolutely no change at all at any load or speed.

Am I supposed to be seeing a rising value with engine speed and load, even without det?

The Rig

iTrader: (13)

At least yours got to 1, mine stayed at 0 in the same scenario as you, altho, i did do a battery reset and it was at 255 for ages

Welloilbeefhooked

iTrader: (3)

What does it show now? I'm assuming you changed the sensor?

Just replaced mine with another one that I believe to be good. Will test it tomorrow.

Dmacnz

I think the sensors are rather sensitive to how they are torqued up too.

The Rig

iTrader: (13)

Mine still shows 0, drives fine etc so havent chnaged sensor, i tested it and the ohms etc were within spec of a knock sensor

Dmacnz

I would say ya sensor is stuffed, I has similar. Replaced the sensor and started counting again. Mine had a crack at the bottom on the plastic surround. Think is pretty common!

Welloilbeefhooked

iTrader: (3)

Ok, logged the knock count and it stays at zero all the time. No other values after 4 power runs and ecu reset beforehand. No det through cans at all.

I'm assuming I should see some noise being picked up like you would on the Pfc commander I had on the previous scoob.

Welloilbeefhooked

iTrader: (3)

When you say counting, what sort of values did you see?

The Rig

iTrader: (13)

same readings i got then, hmm, so when should the value change ? when det is detected/heard ? as im learning det being heard and det being detected are 2 very different occurances it seems

My det cans led to det being heard, knocklink lit up like xmas tree, yet ecu knock control was perfect and cyclic knock count didnt budge from 0, no knock sensor error on cel codes check either and knock sensor passed test for ohms within spec.

would be nice to see a video of the cyclic knock count moving so we know what should be happeneing

Welloilbeefhooked

iTrader: (3)

I'm going to check the following.

Resistance of sensor (don't know value yet)
Check both open and connected signal voltage from ecu to sensor. I think this should be %5v from ecu when not connected and 2.5v when connected?
Continuity of all wiring and connectors.

The Rig

iTrader: (13)

it needs to be around 400k ohms or above when tested


im sure our sensors are good, otherwise they would be throwing up errors

Welloilbeefhooked

iTrader: (3)

Checked the voltage from ecu with ign on but not running and sensor disconnected, which shows 3.3v. With the sensor connected under the same circumstances it reads 2.1 volts.

The old sensor reads 550k ohms and the one fitted also reads the same and I mean identical reading of 550k.

The sensor is earthed correctly as I can read the same resistance from the block to the sensors plug.

Both sensors are part number 22060aa061 which according to import car parts are correct for the v3 type r.

Any ideas?

The Rig

iTrader: (13)

Nope, im out fo ideas as well, its weird how we both have a cyclic knock count of 0 and both hear Det thru the cans yet the esl software doesnt detect any knock.

I basically dont rely on the ecu detecting knock at high rpm/load and mapped the car with the det cans and manually adjusted the timing until no det was heard and rely on the knocklink i have which has been very accurate, when ive heard det, the knocklink has responded the same way, i.e, amber/red lights illuminating.



Maybe Andy can shed some light ?

Welloilbeefhooked

iTrader: (3)

I have one more test to do tonight which involves hooking up the laptop and logging the data while I tap the block to induce a knock signal. I will report back.

Welloilbeefhooked

iTrader: (3)

Right, all sorted!

First of all the first knock sensor works as it should and all tests confirm this. So to test the knock sensor on the car through the esl software the car needs to be running. I wanted to tap the top of the knock sensor so moved it on top of the block nearest the gearbox so as to get a socket extension to it. When connected and with the engine running I can tap the top and get my knock count to 7! Not sure if moving it has helped but I have also been out taking some logs to monitor things.

To provoke a bit of det, audible through my earphones, I loaded on my previous map that I had det present. Det could be heard at about 5k and the esl showed a count of 6. Removing some timing and starting again I could hear no det but had a count of 4 this time. Timing added again with det present and this time a count of 8. Back down with the timing and a count of 3 rising to 4 but with no audible det. So I'm thinking now that a count of 6 to 7 gives signs of det but the knock control map still shows no sign of correction, even though the fine ign control in the logs shows a decrease from 128 to 122. This is of course on my specific application.

Thinking back to what Andy has told me the knock count may well be a simply tally of how many times a knock event has occurred and not its severity, although I seem to get the same number each time knock is heard through my earphones.

Anyway, I'm not going to map purely via the knock count on the esl but it does look like it is pretty consistent at detecting knock. I'm still left wondering why the change in location of the knock sensor has caused the system to work properly?

Welloilbeefhooked

iTrader: (3)

Forgot to say that iam still rises to 16 even when the knock count reaches 8.

The Rig

iTrader: (13)

Did you tap the knock sensor in its original location and had no cyclic knock count ?

i have my knocklink installed where you have put the knock sensor now.

will tap my original knock sensor and see if i get a count in the ESL software.

The Rig

iTrader: (13)

This is what i dont get, if knock count went to 8 yet the IAM still stays at 16, does this mean its knock but not enough to retard timing ?

Welloilbeefhooked

iTrader: (3)

I couldn't tap the sensor in its original and moved it simply to make it easier to get at really. I suppose taping the block should have the same effect. Try it first before moving the sensor.

My sensor is now on the same bolt at the ic bracket on the turbo side of the engine.

Bob Rawle

If you are hearing det then the engine is detting ... so adjust your maps accordingly. Do not rely on the ecu to take care of it. Logged parameters don't always show change I have had parameters "stuck" at a value when its been obvious they are not fixed.

Before any ignition activity Jecs does things with fuelling as a det suppressant, the fuel compensation maps are not "exposed" though like they are for MY99-00 in EcuTek.

thats something I was hoping to discuss with ESL at some point

bob

Welloilbeefhooked

iTrader: (3)

Cheers bob!

I'm now mapping for no audible det and no adjustment when logging the fine ign learning as the fine learning is the only thing changing when I can hear det.

Andy Stevens

The cyclic knock count is a tally of events picked up post hardware, it isn't a level. That's why the differential of the signal is important, ie. big jumps, quickly. If the system is working it will tick up slowly.

The ignition is pulled before any fuelling alteration.

Cheers.

Welloilbeefhooked

iTrader: (3)

Andy, I saw ign learning changing with det from its normal 128 value to 122 ish when I witnessed my knock events. The fine fuel learning only changed when I was pushing way too much ign, which backs up your statement of ign then the fuel being adjusted.

Is there no way it could be made to record severity and possibly flash a cel? I only saw ign changed by the ecu when it quickly tallied up to 7 or 8 on the cyclic knock count. I'm assuming a single event now and again doesn't really indicate a problem but in quick succession it does.

Bob Rawle

Nope I don't think so Andy ... we need to meet up as discussed, I see relatively significant afr movement with no ignition movement anywhere, you have to have very fast equipment to catch this though. Even more critical to MY99-00 ecu.
This can catch an inexperienced mapper out. There are related circumstances which exist after a "reset" as well.

all good stuff

cheers

bob

Andy Stevens

Reading the control code, fuelling is modified at IAM<7.

There are lots of other reasons why fuelling may be changing however.

Happy to have a chat though Bob, we can reveal further parameters as required.

Cheers

Andy

Welloilbeefhooked

iTrader: (3)

Just thought I would add a bit of info.

I recently fitted a 3 port bcs to replace the 2 port. After logging a run and checking with det cans everything seemed well with only a very small adjustement on the min and max duty tables. What I did find is that the boost comes in a little earlier and uses the ign and fueling of earlier rpm cells on the map. While this is great it did throw up some changes in the knock control map. Its all sorted now but I thought that I would show the correction the ecu made, and thats without me hearing any det. Ecu was reset so it shouldn't be residual data.

Im assuming the 2 port still allows the wastegate to open slightly even with 100%(actually 91.6%) duty.