[mikelr]

A little snip it from the repair manual about the 918's knock sensor...

EMM.2 - T KNOCK SENSOR
The ignition timing required for optimum performance can lead, under certain operating conditions;’ to
detonation of the fuel mixture in the combustion chamber, causing excessive heat and pressures and a characteristic
‘knocking’ noise. If allowed to continue unchecked, major engine damage can occur. A single knock
sensor is mounted in the crankcase ventilation cover at the front of the crankcase ‘V’, and is able to identify
detonation ‘knock’ and produce an AC output voltage which increases with the severity of the knock. The
knock control strategy programmed into the ECM allows the engine to adhere as closely as possible to the
optimum ignition timing and turbo boost settings without a damaging level of detonation.
Knock sensor
When the knock sensor detects the onset of detonation, the ECM first rapidly retards ignition timing to a
safe level, and then progressively advances ignition until detonation is again detected (if at all), and the cycle
repeats. If however, retarding the ignition does not stop the detonation, the ECM also rapidly reduces boost
pressure before slowly allowing it to build up again (see Sub-Section EMM.2 - R).

 

[ksgrimsr]

A little snip it from the repair manual about the 918's knock sensor...

EMM.2 - T KNOCK SENSOR
The ignition timing required for optimum performance can lead, under certain operating conditions;’ to
detonation of the fuel mixture in the combustion chamber, causing excessive heat and pressures and a characteristic
‘knocking’ noise. ...

When the knock sensor detects the onset of detonation, the ECM first rapidly retards ignition timing to a
safe level..

Thanks for the helpful snippet. I'm familiar with this section of the manual and it's the primary reason I have been asking so many questions about the knock sensor and its failure modes (whether it fails in a way that makes it trigger all the time) and how to confirm it is indeed triggering by measuring it (my OBD-II reader doesn't cut it).

In the absence of a Tech-1 what are my options for measuring the knock sensor? I have already read the section of the manual mentioning the 2.5V bias and A/C voltage the sensor induces, etc., but it's a little vague about where I might be able to tap into the signal to make a measurement. Is there a convenient harness connector that is accessible that I can gain access to this signal conveniently?

The way I see it there are three options I need to narrow down with some diagnosis:
1) It is indeed knocking (and then I can chase down the reason for the knock of which a few possibilities have been suggested)
2) It is not knocking and the sensor has failed in a way that is resulting in it sending spurious knock signals to the computer (in which case I have a bad sensor to replace)
3) It is not knocking but I have something else rattling around in the engine such as a bad rod bearing that is causing a slap that is triggering the knock sensor (any mechanical ping will trigger the knock sensor and it doesn't have to be induced by detonation).

Before I attempt to fix something, I'm first striving to properly diagnose what is going on so I know what to fix and why. My first angle is to see if I can determine whether the knock sensor is indeed triggering and then proceed to determine why if that's indeed so. So in the absence of a Tech-1, does anyone have info on how I'd go about measuring the knock sensor signal?

Thanks all for your help!

Knut

 

[sailorbob]

Have you tried the OBD-II s/w by Peter Maeers? It works with ELM323 or ELM327 OBD-II interfaces.

 

[ksgrimsr]

Got my hands on a Tech-1 with the Lotus V8 module in it (thanks to Guy). I'll do a couple measurements this coming weekend and post what I find.

I have a new knock sensor on its way just in case it is the culprit. My fear is that there is something banging around that is triggering the knock sensor which is a little concerning since worst case it's a rod bearing slapping around. My heat shields are split and rattling but I don't think that has the kind of impact that would trigger the knock sensor.

Knut

 

[ksgrimsr]

Knock sensor access

In the handbook it mentions "... remove or partially remove the intake plenum to provide sufficient access to the sensor..."

So does this imply that sufficient access to the sensor can be achieved by not completely removing the plenum? Can it be released and simply lifted at the front to gain sufficient access?

 

[mikelr]

I don't see how you could really get to it without pulling the upper intake. Besides if you disturb the seals you needs to pull it completely to reseal properly.

 

[ksgrimsr]

Tech-1 readings from performance pulls

So I had a chance to plug in the Tech-1 I borrowed with the Esprit V8 module in it (thanks Guy) and take a couple measurements with it on my lap while doing some performance pulls.

Since it only displays 2 lines of live data at a time I first looked at the knock sensor voltage and the KNK IGN Retard value. The knock sensor voltage came up to about two tenths of a volt during the pulls and the KNK IGN Retard value came up to 0.75 degrees, which is nothing.

I next monitored IGN Advance and KNK IGN Retard values and indeed during the pulls the advance drops down (dropped to 7.5 degrees in my pulls which were a little less aggressive) while the knock retard rises to the modest 0.75 value.

I thought maybe the KNK IGN Retard value displayed might have the decimal point in the wrong place, but when I just rev the engine not under load I get 30 degrees of advance with the same 0.75 degree knock retard value so I don't think there is some snafu in interpreting the knock retard value.

So something isn't adding up. The knock retard does not account for the retarded ignition timing I'm getting and at high RPMs having advance in the single digits doesn't seem correct and I get little pull at high RPM as a result.

So what other inputs does the ECU use in determining whether to retard ignition timing other than the standard stuff of throttle position, RPM, MAP, and inlet air temp? Those parameters are typically static tables in the ECU. All these readings look perfectly fine on my OBD scanners.

I thought for sure I was getting timing retard as a result of the ECU picking up noise on the knock sensor, but the latest measurement seems to indicate this is not so. What else causes the ECU to retard timing?

Knut

 

[mikelr]

What kind of map values are you seeing? I think it would be helpful if you made a list of all the pertinent sensor values during a pull and posted the values here so we could evaluate them.

 

[jtrealty]

What kind of map values are you seeing? I think it would be helpful if you made a list of all the pertinent sensor values during a pull and posted the values here so we could evaluate them.

Refer to my previous post (16)
If any of the inputs mentioned are not within range it would cause the timing to be reduced to reduce power and protect the motor. You are just going to have to examine each sensor to find one causing the problem. Instead of driving the car it may be more useful to use a dyno if you can get on one. My first guess would be IAT and MAP but reduced fuel pressure could be causing this and you would not necessarily see an error code.
David Teitelbaum

 

[jtrealty]

What kind of map values are you seeing? I think it would be helpful if you made a list of all the pertinent sensor values during a pull and posted the values here so we could evaluate them.

Refer to my previous post (16)
If any of the inputs mentioned are not within range it would cause the timing to be reduced to reduce power and protect the motor. You are just going to have to examine each sensor to find one causing the problem. Instead of driving the car it may be more useful to use a dyno if you can get on one. My first guess would be IAT and MAP but that should throw codes. Reduced fuel pressure could be causing this and you would not necessarily see an error code.
David Teitelbaum

 

[mikelr]

Refer to my previous post (16)
If any of the inputs mentioned are not within range it would cause the timing to be reduced to reduce power and protect the motor. You are just going to have to examine each sensor to find one causing the problem. Instead of driving the car it may be more useful to use a dyno if you can get on one. My first guess would be IAT and MAP but that should throw codes. Reduced fuel pressure could be causing this and you would not necessarily see an error code.
David Teitelbaum

I get a gold star if that ends up being the problem!

 

[ksgrimsr]

More data

Below is some more data on the basic logged engine parameters during some test pulls. This chart has a lot of data on it and some of the colors are pretty similar but you should be able to make it out. This data does not include the knock retard or knock sensor voltage that I reported previously since that is only available from the Tech-1 which doesn't really do logging. Also, because I'm collecting too many channels the update rate is pretty slow and as a result you see the same value being repeated a lot and the time resolution is poor. Some of the samples might fall in the shift points and look funny due to the time resolution coarseness.

At about 8 seconds before I start the pull you see the advance is around 30 degrees and then I put the hammer down a little before 10 seconds. At about 12 seconds the boost is 1.75 bar, TPS is at 100%, RPM is 5000 and IAT is 35 C. Everything looks sweet except the timing drops down to about 8.

Inadequate fuel pressure has been mentioned a couple times. If low fuel pressure doesn't result in knock (no knock or knock retard recorded by Tech-1) what input does the ECU use to know the fuel pressure is low in order to know to retard the timing? I have O2 sensor readings as well for the pulls in the chart below (not included on chart since it would clutter it even worse), and O2 sensor readings do not indicate a lean condition (both banks record voltage during wide-open throttle in the 0.85V to 0.9V range which is slightly rich).

Thanks for the help guys.

Knut

 

[jtrealty]

The way the chart is posted, with the banner along the bottom covering the legend, I can't figure out what line is what. You are correct, there is no direct input for fuel pressure and I also am not sure what the ECU would see that would tell it to affect timing because of it. Of course that could explain why there are no error codes but still doesn't explain what is messing up the timing. The good news is fuel pressure is easy to check. Hard to believe it is a sensor since any values out of range are supposed to log an error. For weird problems like this, one of the best methods is to substitute known, good components and hope you get lucky quickly. Is there *anything* that was done to the car prior to this behavior occurring? That can give you a clue as to where to begin. The Tech 1 can be hooked up to a PC for data logging but it requires the proper cables and software. I don't know if it is even obtainable anymore.
David Teitelbaum

 

[ksgrimsr]

I'm not sure what banner you are referring to that's covering up the legend and the chart in the posting looks fine to me, but the image below is just the legend to help clarify. Unfortunately, there are 3 blue colors Excel decided to choose for the chart and I didn't bother changing the colors to better distinguish them, but from context it's easy to see which is which in the chart.

The power reduction happened between laps 2 and 3 of the second session of a track day a while back, so there is not a clear correlation of the loss and something that was serviced on the car.

I'm presuming that the timing values that I'm measuring are quite retarded since I'm not used to such low advance values on other cars and it feels like retarded timing since the pull at higher RPM's is poor, but I have to admit that I don't have a baseline on the 918 engine to compare against to even know whether the timing values are indeed retarded compared with what they should be. Does someone have timing values for their V8 under high-output conditions and can confirm that my values are indeed different than they should be?

Knut

 

[jtrealty]

I was just grasping at straws hoping to uncover some clue. At this point we have to assume a hard failure since the problem seems repeatable and it occurred on it's own. I got the legend but the colors are confusing and the "stuttering" makes it hard to interpret. As you mention, we are just guessing the timing is the problem since we don't have a baseline for "normal". Since there are no error codes (and there wouldn't be for low fuel pressure) I still say you should start there.
David Teitelbaum

 

[Wayne Ellison]

Sorry to be late to this thread, I've been out of the country.

The timing pattern you are seeing is quite correct for the stock Lotus 918 ECU! The timing will retard all the way to 6* under high IAT temps (225* F) and boost. You also can't accurately interpret the voltage reading or duty cycle of a narrow band O2 sensor, you will have to install a wide band O2 sensor to have any faith in the readings. The narrow band sensor is very non-linear off of stoichiometric.

I agree with your concern that the ECU cannot respond to PID requests intermittently, as being an indicator of something going on. I believe in fixing the obvious first, what ever has convinced you that you have a flakey cam sensor, and they do go flakey and cause all kinds of weird failures, fix that now! It may be swamping the ECU with trash. Timing is based on the crankshaft sensor, but the ECU tests the timing relationship between the crank and cam sensors. The car will not even start if the timing relationship is very far off!

I use a PC based scanner from OBDII.com which has the ability to read the enhanced PIDs from Lotus (like knock count among others), and also allows me to set the sample rate . I have had a knock sensor, generate a false error code, which was caused by the spark being coupled into the knock sensor from a close plug wire. I made a shield, with an attached ground, which covers the knock sensor. However this never did create high knock sensor counts.

Your comment about the power loss being sudden and absolute at a track day indicates some type of failure. I don't think it is spark timing related, spark is behaving correctly. With everything you have checked I would lean toward secondary injectors. Get a wideband O2 sensor.

Maybe you got into the elevated IAT region for the first time, see what you logged for IAT, if it is over 200* F. your intercoolers aren't working and that's a problem. There is another boost/timing restrictor, which is undocumented in the manual, oil temperature. An oil temperature sensor was added to the drain plug around 1999 MY, it will cut boost and timing, but I don't know the thresholds, Lotus is quiet on that subject.

 

[Wayne Ellison]

I just looked at your logging data, 78 KPA of boost is correct, but really only 40 * C of IAT, where's your IAT sensor? That's very low IAT ???

 

[ksgrimsr]

Thanks for chiming in.

Unfortunately the chart I posted has 3 blue lines on it which confuses things a bit. My IAT is the flat blue line that starts out at 35C and slowly rises a couple degrees to a little under 40C in the chart during the pull, so my IAT looks pretty nice and the intercoolers are working well.

So my timing looks reasonable to you? Unfortunately I don't have a baseline to compare against and with poor power at high RPM it felt like retarded timing so it's one of the things I checked (after chekcing boost, etc). Does anyone have a baseline they can post for what is considered normal on a well-running car? What PC based OBD-II reader are you using for your V8 that supports the Lotus specific values?

Thanks for the help.

Knut

 

[ksgrimsr]

Looks like our messages crossed each other.

I have air/water intercoolers on my V8 and although the engine was up to temp I hadn't done a lot of pulls yet when I logged the data in the chart so the coolant for the intercoolers was nice and cool giving me a pretty nice IAT. Even at the track when things are all warmed up and I've been flogging it for 20 minutes the IAT usually holds around 60C. There is a writeup on my intercooler setup back in the archives.

Knut

 

[Wayne Ellison]

Here are two PDF's of my stock ECU with intercoolers.