Just another "is this real knock" thread

[dynoiasi]

Hello everybody,

Since I have not finished yet my DIY "detonation listening device" I made a few runs using only the engine knock sensors and very conservative tuning, trying to spot knock signal waves on the logs. To my surprise I found a few spots that look a little suspicious which I would love to hear your opinion about. It's hard to believe that this might be real knock considering the crazy rich AFR and the low timing but still I would bring these to your attention. My engine is a M50TUB25 with about 9.5:1 compression and a smallish turbo.

Problem 1: 

There's a small spike on cyl2 knock level. It settles back and then rises again along with the boost pressure. Might this be knock, or knock should be an even bigger spike compared to the previous average noise?

Problem 2: 

The cylinder 2 noise seems to raise pretty suddenly but the cylinder 6 even more so. I think that cylinder 6 might be caused however by the  RPM cut which happens at the exact same time but I am not quite sure which happened first. 

 

I also attached the base map and the log file. The first problem is in the SOL section, the second one in the rpm cut section, timestamps should match the attached screenshots. Thank you very much for you help!

 

Have a nice day,
Florin

dinmasinapolizuiasi.pclr Log 2019-03-31 6;32;48 pm.zip

[Adamw]

To me it looks like just your knock target is a bit conservative at higher RPM.  The small lone spike on cyl 2 at -12:24 maybe knock - but sometimes even an odd bump in the road can cause random spikes like that when the background noise is low.  Really the only way to know is to confirm with audio.  Your knk level on cyl2 & 6 in your screenshots above look worse than they are due to you havent different scales on the Y axis.

I would drop the RPM high lockout a few hundred RPM so it disables before the RPM limiter kicks in.

[dynoiasi]

Many thanks for your opinion Adam. I did not know about the bumps in the road causing false positives and I was wondering why in my log there are very rare but big noise spikes while engine has basically no load. Maybe this might be it, since we certainly have a lot of bumps in this part of the world...

One more thing: for me the best way to discern whether some noise is real knock or not, would be to see that after the knock control timing intervention the knock went away on the next stroke or at least it reduced the amplitude. Can you see this on a log? As much as I found you can only see the knock ignition trim tables per cylinder but they are "static" in nature. There is no real time parameter that shows the ignition trim due to knocking but only other ignition trims. Is this true?

 

Best regards,

Florin Tufescu

[Adamw]

You need to connect audio knock equipment to confirm what knock level is real knock then set your threshold accordingly.

The individual knock ignition trims are real time, just will remain in place until there is no knock detected in that region and the advance delay has expired.  So if you see a knock ign trim of "-3", then that cylinder has 3 deg less timing at that time than what was commanded.

[dynoiasi]

So I finished my diy electronic detonation listening device (just a preamplifier connected to the factory knock sensor and a software real time filter running on an android tablet) and I am heading for the dyno tomorrow. I really hope that on the dyno, adjusting timing in steady state and listening I will be able to tell when knock occurs. Because so far I am not quite sure. To me it sounds like the engine is already knocking with what I consider to be the safest settings making very little power for this engine. 

Please take a listen to my recordings: 

The first link is to a bigger sound file containing multiple pulls with no filter. Just what the knock sensor gets going through the preamp:

https://drive.google.com/open?id=1i1l5j_ZUo3_CQZJrHFkkKv5ujRviF0FW

The second link is to a shorter filtered pull:

https://drive.google.com/open?id=1VJV729P3c6bU82_WBDJSQ6KRad_4ZYDo

Does this sound like knock? Could this be caused by the strong Vanos rattle which I am currently experiencing?

 

Thank you very much!

Best regards,

Florin

[Adamw]

The unfiltered file I cant really tell.

The filtered file sounds pretty odd however it definately has all the spatter and pop characteristics that I would normally associate with knock noises.  I would suggest deliberately making it knock at some relatively low load area to see if you get a similar sound.  You can do that by putting some really large timing numbers into a small group of cells in the ignition table and use your throttle to move the engine operating point into one of those cells.

[dynoiasi]

Hello everybody,

Update on my knock research and a new question: Firstly, it seems like the sound recordings I attached on april 30 were not ok. I think the amplifier input was too high and so clipping occurred which somehow created the pops. I reduced the amplifier gain a little and now the recordings are much cleaner. However I can still hear a couple of pops here and there.

The first recording is made with 16 degrees advance and the second one with 14 degrees. They are both filtered with 6 to 6.5khz band pass. What do you guys think?

https://1drv.ms/u/s!Amk_BITNgVuBs3sXuVLSgmNuPfml?e=UdVfoQ

https://1drv.ms/u/s!Amk_BITNgVuBtAIk-lqVsaiYJzKA?e=ABjnuq

In addition to these recordings I have one big theoretical question I hope someone more experienced will be kind to answer: is it normal for the engine noise to increase when you are increasing boost and timing, or should the average noise at any given RPM remain largely the same? On one hand, the engine mechanical noises should be the same , but on the other hand if the cylinder pressure increases I would expect more "bang" and as a consequence more noise on average. To make the question more clear: if you increase the boost/timing/power of an engine, would you also have to increase the knock detection threshold or would it be the same regardless of the power as long as there is no "real" knock?

[Adamw]

I can hear barely anything in either recording.

7 hours ago, dynoiasi said:

is it normal for the engine noise to increase when you are increasing boost and timing,

Yes, this is normal.  Timing doesnt make a huge difference to the noise in general (assuming we are not talking a massively retarded starting point), but boost certainly does.  Typically with a NA engine or a turbo engine that runs a fixed boost level you only need a 2D knock target table, on a turbo engine that runs various boost levels then you need the 3rd axis on the target table to adjust the target based on load.

[dynoiasi]

Thank you very much for your answer, Adam, I really needed to find out the answer to my question. Regarding  the recordings, I forgot to describe what seems important about them. Especially the first one is too long and only contains strong acceleration at around second 40, with some "popping" sounds happening from 49 to 53. The second one has some spikes/pops at around second 4. However in both of them the pops do not sound like I usually hear in knock recordings, like metallic taps that are different from the rest of the sounds, but they are more like random spikes that sound the same like the rest of the "normal" sounds, that is why I can't really tell whether they are knocks or not.

I tried to induce knock to hear how it sounds and I went as far as 50 degrees at around 2500RPM with very small boost but I got nothing. Could it be that it was really not knocking because the load was too small for my compression, or could it be that my whole recording system is somehow flawed, although it definitely seems to pick up the sound that the engine makes?