Knock sensors revisited

[Force Fed Mopar]

So, long story short, I did an immense amount of playing around with the knock sensor vs rpm table in the Lebaron, trying to find a happy medium between engine/combustion noise and accurate knock sensing to tune from last year, without much success. Pretty much ended up just about disabling it and going by ear/seat of the pants/educated guess work.

I am using a new knock sensor, but do not have Teflon tape on the threads. Is this causing it to be overly sensitive? I know many people have put it on theirs to keep it from being so, but was it that way from the factory? Is this a good way to make it usable?

I know also that the type our cars use is very primitive and designed with specific parameters in mind. Possible that at higher than stock boost and power levels that it's no longer useful at all? I have been told as much by some accomplished tuners on other sites. I do plan on making a wideband knock sensor setup (as has been described in other threads) for it to compare against, so I may be able to answer my own questions at that point.

[5DIGITS]

So, long story short, I did an immense amount of playing around with the knock sensor vs rpm table in the Lebaron, trying to find a happy medium between engine/combustion noise and accurate knock sensing to tune from last year, without much success. Pretty much ended up just about disabling it and going by ear/seat of the pants/educated guess work.
I am using a new knock sensor, but do not have Teflon tape on the threads. Is this causing it to be overly sensitive? I know many people have put it on theirs to keep it from being so, but was it that way from the factory? Is this a good way to make it usable? I know also that the type our cars use is very primitive and designed with specific parameters in mind. Possible that at higher than stock boost and power levels that it's no longer useful at all? I have been told as much by some accomplished tuners on other sites. I do plan on making a wideband knock sensor setup (as has been described in other threads) for it to compare against, so I may be able to answer my own questions at that point.

"...trying to find a happy medium between engine/combustion noise and accurate knock sensing to tune from last year, without much success."

What approach are you using to identify the different levels of background noise?

Thanks

[Shadow]

I've never understood the difficulty others have had with the factory knock sensors in these mtrs? I've used them from day one, as is and they've always worked Great for a factory piece! (except when they were bunk and needed replace that is)

I don't know where everyone is getting all of the sensitivity from? Typically I'll see .02-.04 fluctuation on the DRBII which I've just found to be normal operation. Then, IF the mtr starts to detonate the knock is going off way before I hear anything and the ECU is pulling timing to compensate. After that it will become audible.

I've always used and learned to trust them "as is" and run Zero knock tunes. Although I have had other built Turbo Mopar/ Dodges come into the shop that just wouldn't stop knocking and the owner had given up. Very rarely was it the sensor, but just a mix and mess of everything being put together poorly. (a jumble of parts bought off the net and bad advice from ppl who unfortunately don't know what they're talking about)

I'm starting to think it has a lot to do with ppl tuning their own cars and messing everything up in the ECU?

[Ondonti]

headphone setup is what I would do before ignoring noises or relying on your normal ears. Read plugs at least after a pull etc.

[Reaper1]

I'm curious where this goes. The tune I got from Rob (ShelGame) for my Z was great! It ran much better than the other tune I had, especially in the mid range and part throttle. I had him include the CEL knock detection in that tune. No matter what I did the CEL would always flash at me under boost. I tried lower boost, adding fuel, different fuel, adding water/meth injection, adjusting the timing, and putting Teflon tape around the threads. Nothing ever made it better (or worse, just to be clear). I never did get around to changing out the sensor. IIRC Rob did tell me that the MOPAR knock sensor was better to get and could absolutely be relied on compared to a "parts store" version. Now, knowing what I do about "parts store" parts, depending on the manufacturer (usually Wells), they are the same part you would get at the dealer. However, I do also know that sometimes the OEM versions of the parts hold a tighter tolerance or are built to specific specs where as the aftermarket versions are more generic.

The question whether the engine in my car was actually experiencing knock all those times is debatable. The plugs never indicated it. When I popped the headgasket, though, the gasket itself showed signs of higher than normal temps/pressures, #3 piston had a small amount of pitting, and the rod bearing had a touch more wear on the top shell compared to the bottom shell. The thing is I can't say for certain whether that happened when I popped the gasket, or from some other time in the past.

The long block of that engine was completely stock. It made no noise, so piston slap or something like that is out of the question (or it should be).

An overly sensitive knock sensor can be just as bad as a broken one, or not having one, especially when tuning. I am very curious how to filter out background noise, though. From what I understand this would be like a bandpass filter because knock normally produces a certain small range of frequencies. That range is mostly defined by the design of the engine (I think bore size has something to do with it).

Ken, I am VERY interested in hearing your expert input on this. I remember you going over it at a SDAC tech session, but that's been a long time ago.

[shackwrrr]

I had to disable mine below 3k rpm. but I have loose fitting forged pistons. My timing is dead flat 0 degrees over 5 Psi of boost but I still get cel flash sometimes. If I push it I get audible knock and when its cold out it gets much worse.

[Shadow]

I had to disable mine below 3k rpm. but I have loose fitting forged pistons. My timing is dead flat 0 degrees over 5 Psi of boost but I still get cel flash sometimes. If I push it I get audible knock and when its cold out it gets much worse.

Something seriously wrong there. I'm surprised how many drive around with motors that have audible knock and can't figure out the problem.

[Force Fed Mopar]

What approach are you using to identify the different levels of background noise?

Thanks

At first I copied the MP Stage 2 LM table, but it was for a 2.2, so then I went to the 2.5 MP Mtx table. At 10-12 psi on stock 2.5 MP timing tables it seemed okay, but over that it would set the light off. I was fairly sure that at 15 psi it wasn't detonating, so I just bumping up the knock threshold until I didn't see the CEL flash anymore at that boost level when it spooled up. That seemed good except that it would flash on above 4500, so then I bumped it up some more above that. That worked fine, until I went up to 18-20 psi, then I had to do it all over again.

Now at this point the threshold is way higher than stock. Maybe not a big deal, maybe it is a big deal? Every so often it will flash on a normal acceleration shift too, no boost, just run it up to 2500-3k and shift and it'll blink on a sec, so I move it up a little in the lower rpm area also.

I have Wiseco pistons w/ about .004" clearance, and stock original valve springs and 2.4 lifters (the ones with the bleed hole in the nose) properly shimmed. New knock sensor, no tape on the threads. Stock head, stock valves, stock cam, stock 2-piece, 52mm TB, water-to-air intercooler, stock T2 turbo, ported exhaust manifold, 2.5" sv, 3" exhaust. Ran 80 mph in the 8th on street tires. Car weighs 3300 with me in it. 568 trans, 3.77 gear.

Here's my current knock table compared to the stock MP 2.5 Mtx table (green line).

[zin]

I'd also install a "det can"/headphones before disregarding a knock sensor... now that I have a workable soldering iron, maybe I'll put together my electronic version, got high hopes for it.

Mike

- - - Updated - - -

I'd also install a "det can"/headphones before disregarding a knock sensor... now that I have a workable soldering iron, maybe I'll put together my electronic version, got high hopes for it.

Mike

[Ondonti]

I'd also install a "det can"/headphones before disregarding a knock sensor... now that I have a workable soldering iron, maybe I'll put together my electronic version, got high hopes for it.

Mike

Yeah, a lot of stress over something that you can 100% prove with a headphone setup. I have tried this even in a bone stock car and recorded the audio and then spectrogram'd it. Knock stands out, especially when you can hear it over the exhaust. Trying to identify knock without listening to it is like a blind man trying to learn sign language. Once you get the audio to match up with your knock sensor behavior, you know you are on track.

[Force Fed Mopar]

Also, here is my current WOT Timing from Map table. The red line is stock MP 2.5, the blue and green lines are what I've tried and it would set the light off either way. The stock one flatlines after 15 psi so I can see why it would knock above that. The blue I just basically continued the slope, the green is where I pulled a bunch of timing out trying to get it to not show knock.

Attachment 49344

[shackwrrr]

Here is a copy of my Cal, if you open it up you can see that I have less timing than factory at 15psi and my knock voltage scale is raised significantly. I still get random flashes and audible knock. Rob M. do you ever get audible knock?

[Force Fed Mopar]

Here is a copy of my Cal, if you open it up you can see that I have less timing than factory at 15psi and my knock voltage scale is raised significantly. I still get random flashes and audible knock. Rob M. do you ever get audible knock?

Only on the stock MP timing and at more than 15psi.

I assume you have a new sensor also?

[Force Fed Mopar]

My 2 main problem areas with knock were right when it spooled fully (about 3k), and then above 4500. I did try pulling timing from those points in the Timing from RPM table also. It just always seemed to get slower the more I fooled with it, always just ended up going back to what I have now, or close to it.

[5DIGITS]

As indicated within the thread, this area of calibration does not typically require excessive effort and recalibrating the knock sensor should not be considered an approach to mask mechanical issues or support increased spark advance.
But for those who are interested in gaining additional information, the following steps can help identify what the levels are and where the peak values are realized.

The production knock sensor calibration values were set to avoid any and all audible knock, for customer satisfaction purposes.
This also includes an effort to preserve the head gasket, avoid dealership repair downtime, and the resulting warranty costs.
The result has the spark reduction process somewhat hyperactive and over-sensitive even in the stock configuration.
Considering that many of the vehicles on the forum are modified, the challenge becomes understanding how the modifications have affected the mechanical and combustion noise levels, relative to the knock sensor.
This is where a CAS (Combustion Analysis System) would be very useful but without it there are a couple ways to understand the back ground/operational noise levels, specific to your vehicle.

Considerations:
The engine should be in good mechanical condition, for obvious reasons.
Please make sure all engine and sensor grounds are intact and the controller is free of any faults that would alter the results.
High output ignition systems can impact the sensor noise levels - consider testing with and without.
Confirm that the injectors metal body do not contact/touch any portion of the intake manifold.
When this occurs, the injector will transfer its mechanical noise into the intake.

Getting started...
Considering that combustion noise will vary as a function of load/MAP and the mechanical noise is best identified with minimal cylinder pressure, we can attempt to isolate them and determine a general baseline for each.
The following steps are best completed while monitoring/recording engine RPM, MAP, and knock sensor voltage, via any safe means possible.
Again, this is far from combustion analysis but it can provide added insight, for those who are interested.

Mechanical Evaluation :

• Retard spark advance via the distributor or in cal (for DIS) by 4-6 degrees throughout all operating conditions
  (Depending how much spark advance is being used, this reduces the opportunity for detonation while testing)
• With a warm engine, start at the lowest possible throttle angle/load, record the knock sensor voltage while slowly increasing the engine RPM from 1000 to the safest peak RPM
• Decelerate the engine slowly, to avoid rapid low-load engine unwind
• Let the engine idle for several minutes to de-aerate the oil and minimize any resulting valve-train/piston noise
• Repeat the data collection process and confirm consistency with each previous run
• These steps should not exceed ten (10) inches of vacuum, to remain at minimum cylinder pressures
• At this point, a 'general' mechanical baseline has been identified, as seen by the knock sensor
• If there's interest, a test can also be done between 2500-3500RPM while dithering the throttle
• This induces varying combustion chamber pressure levels which excites wrist-pin and piston slap noise

The mechanical traces should be fairly consistent producing essentially a single knock sensor voltage curve, for the narrow MAP range used.


The next steps duplicate the previous process but introduces load.
The load intervals should be selected based on what you're willing to repeat up to elevated boost levels.
(For example: 2.5, 5, 10, 15psi, increments etc... )
This typically requires dragging the brakes under elevated torque conditions - USE CAUTION !

Combustion Evaluation :

• Retard spark advance via the distributor or in cal (for DIS) by 4-6 degrees throughout all operating conditions
  (Depending how much spark advance is being used, this reduces the opportunity for detonation while testing)
• With a warm engine, start at atmospheric load, record the knock sensor voltage while slowly increasing the engine RPM from 1000 to the safest peak RPM
• Let the engine idle for several minutes to de-aerate the oil and minimize any resulting valve-train/piston noise
• Repeat the data collection process at the selected load to confirm run to run consistency
• With repeatable results, increase the load to your next chosen level and repeat the previous steps
• Considering the testing process, repeat the above steps only to a peak boost level that your comfortable with
• At this point, the impact of increased cylinder pressure is identified as the growth of knock sensor voltage that exceeds the previously recorded mechanical results

The load based knock sensor curves will elevate as MAP increases but they don't always increase at the same rate between the tested boost levels.
This is the reason for collecting multiple load results and this should be considered when determining the knock table voltages for boost levels beyond the test levels.

I hope this helps and enjoy.

[Force Fed Mopar]

Helps very much, thank you! I had done some of this testing you describe, such as revving the engine in neutral under as light of throttle input as possible to get to max rpm and watch for the knock light. Did a little logging with MP Scan but not enough apparently I think I will throw out what I have and start back stock again. I do have an EBC now also, which doesn't allow it to spike boost as bad as the G-valve I had on it. So it should be easier to verify the knock levels as well as tune each boost level properly.

[5DIGITS]

The method of using the CEL to indicate knock was brought up earlier in the thread and its connection to the knock sensor calibration table is worth discussing.

Essentially there are two methods to support CEL illumination when knock is suspected.
The first is the old and currently used method of activating the light when the knock sensor voltage exceeds the voltage values calibrated in the knock sensor table.
Although this does work, its increased sensitivity while using a single input condition allows the light to illuminate during erroneous single low-level 'burst-knock' events, noise spikes, and/or with mechanical influence.

The second uncommon method is an approach that relies on the controllers determination of actual knock by illuminating the CEL based on multiple input conditions.
The multiple input determination helps minimize the erroneous events by only illuminating the light when realistically necessary.
Knock sensor voltage levels that exceed the knock calibration table value, knock events determined as something more than noise, and knock retard of any amount impact the determination.

What's common in both methods:

• The opportunity for the driver to make a corrective decision to back out of the throttle or take their chances
• Dependency on the knock sensor calibration table accuracy
• Nearly immeasurable differences in controller determination time

What's different in the second method:

• Ability to predetermine the legitimacy of actual knock occurrences
• Minimized erroneous CEL illumination
• Reduced tendency to modify the table to dangerous levels that nearly numb knock detection

The initial use of the CEL for knock detection came up in during the race days of the 2.2 race programs.
Coincidentally, the same approach was used with a fuel rail mounted sensor to turn on the CEL when fuel pressure dropped while cornering, due to pump cavitation.


When things are done properly in the controller, backing out of the throttle is no longer needed and the use of the CEL for knock indication is reduced to 'an event occurred but it's been handled'.

"Stay Tuned" and Enjoy

[John B]

I have the knock sensor trigger my check engine light, and it comes on when I hit a pot hole.

[ShelGame]

Ken, the current implementation of the CEL knock indicator is that the light comes only when the ECU is retarding timing due to knock. IE, any one of the 4 knock retard values is greater than 0. Knock retard is ignored below a set coolant temp (to reduce cold false knock).

So, its not just simply looking at the threshold.

[wallace]

Have you guys tried using higher octane fuel to see if this reduces or eliminates the knock? I have had 2 problems that were not knock cause the cel to flash. I traced one to bad alternator bushings and the cel went away. The second was knock on light load when engine was not up to operating temp. This second I believe was a rod bearing on its way out...I haven't taken the engine apart to see but I believe it spun while being flogged. With mpscan can we see which cylinder is detonating in the log?