Trying to simulate and detect det.

MichaS14a · Post by **MichaS14a** » Fri Apr 17, 2009 9:21 am

I have found some websites again with some informations about engine knock:

Engine Knock Theory
Listen to your Knock Sensor
Knock Sensor Sounds

A post from a subaru forum: engine noise vs knock detection

Perhaps it will be possible to build an own knock monitor with these chips (at the moment I do not fully understand what is missing for a complete solution):
New Texas Instruments TPIC8101 Knock Sensor Interface Chip

I hope some of these informations are useful.

So I am not completly wrong with the idea to apply a bandpass filter at the recorded signal.

chris2712au · Post by **chris2712au** » Fri Apr 17, 2009 6:28 pm

I am all for it.. I would love to see what you come up with.. just have learnt from experience that road tuning is just so difficult.. it gets you in the ball park of where you need to be.. and it is very time consuming.. its just so much easier when you can dial in a load and hold it there especially for timing a specific area on the map.. holding a load point using a dedicated driver and a very long hill is difficult..

subaru's had problems with their knock sensors and it is very common to change them... then again the nissan ones are not so crash hot either especially if you skimp on the fuel quality regularly..

I guess to prove the bandpass theory just get a mp3 file from that site i linked before and play it back through winamp and adjust the EQ..

MichaS14a · Post by **MichaS14a** » Sat Apr 18, 2009 9:52 pm

chris2712au wrote:I guess to prove the bandpass theory just get a mp3 file from that site i linked before and play it back through winamp and adjust the EQ..

There is an easier solution! Get the freeware Audacity here: http://audacity.sourceforge.net/

Install the software, load the mp3-file and go to "Effect - Filter - High Pass Filter". Then play around with "Rolloff" (I try 12db), "Filter quality" (I use the default 0.7 for now as I am not really sure what this entry adjust) and "Cut off frequenzy" (I set this to 5500 Hz for testing). The result is that only frequenzies above 5500 Hz are left over. So you are able to listen the mp3 much louder and there are no more lower frequenzies in the sound anymore.

I will try it with the knock test file and see if it works to only hear the knock itself.

EDIT:
Thats strange... If I apply at the test file http://home.netcom.com/~bsundahl/knock/ ... 1_1_10.wav a high pass filter I did not hear any of the knockings anymore as it will be filtered out. I must apply a low pass filter (Rollof: 3db, Filter Quality: 0.7, Cutoff frequenzy: 1000 Hz) to get (mostly) only the knock itself. But this works good! I think this sample does not match our conditions with Nissan engines.