Ecumaster forum

Okay guys, so I've been messing around with knock sensor setup for the past few days, and THINK I'm going in the right direction. So I figured others would chime in as well.

First and foremost, you have to figure out what kind of kHZ your knock sensor works at. This is just a rule of thumb, and will require tweaking.

The formula is 1800/(3.14*X). X is your cylinder bore in MM. So since I have a RB20, my bore is 78mm. So 1800/(3.14*78) = 7.349. So essentially 7.4. However, the EMU software gives you a few predetermined ones, and I selected what is closest to my value (not rounding up or anything) and came to 7.27. Close enough.

So that was the easy part, here we get to much more difficult areas of setting this all up.

1st, go to the Log tab and open up one of your "Graph"". Before you begin driving, you will want to go back to your Knock Sensor Parameters and set up a few things. First turn your Gain to the highest setting (2.0) and turn your Integrator to 100ms (more on this later).

Now you're going to want to log your RPM, Knock Sensor Voltage, and you'll want to drive normally and make sure you are getting knock voltages. This ensures that the knock sensor(s) are working.

So I HAVE NOT made it passed this part, but hypothetically this is what I will do, and should be fine.

Slowly adjust your "Gain" setting to where you get close to 3v. You do NOT want to EXCEED this value, 3v is what the EMU sees as "normal" meaning, NOT knocking. Pretty much what I am going to do us chose the maximum amount of Gain (2) and slowly work my way down. While logging to make sure I 3v or less. Hopefully the log shows it's ABOVE 3 volts, and I can adjust the gain lower and lower. But if the knock voltage is too low, I'm not really sure what I can do.

Finally the Integrator is more or less a sensitivity dial for the knock sensor. If it's higher than it should filter out engine noise. If it's too low, it'll pick up more noise. The suggested value is 100-200ms. I personally will do the same as gain, I will start at 200 ms and just work my way down.

Hopefully someone else has setup their knock sensors to provide some input on if this method is correct.

gain usually needs to be quite low maybe only 0.5 and set the engine noise by logging sensor value and engine noise making sure there is a good margin between them at all times when you slowly free rev the engine up close to redline. If everything is working correctly knock will show up as very large spikes

So an update. Just as RHD has said, I ended up with .50 gain. I dropped the integrator down to 150. But I am still getting ghost knock in the middle of the RPM range, but goes away in the upper end.

However, it only happens in load, so it has to be too much timing (or so I thought). I am only running 20 degrees in that area. I pulled a little bit of timing, and it's still there. I ended up adjusting the Engine Noise Table in that area, but haven't had a chance to try it.

you can also try to change the frequency in little steps up or down , or try to use second harmocal frequency .... basicly your desired frequency calculated from bore multiplied by 2 or 4

I've tried second and third resonances and still the same deal. Altering timing or fuel does nothing. I'm starting to believe I have a faulty knock sensor.

Hi,
its not the knock sensor fault, its probably fine. I guess you need a bit more tweaking. For example on my RB25, when i use my headphones the engine is so noisy that you u think its knocking almost constantly in the upper range. This is the reason why , when i hooked up a Z32 ECU to it, it was pulling timing due to false readings. I read somewhere that in order to use the Z32 ECU with RB25 you need to change the sensor to an SR20 one , becasse the Z32 ECU is looking for different noise, which is probably correct. Anyway i ended up removing the knock sensor totally, because i decided its not needed for a track car. I have tuned it with headphones and on RB engine the strong knock, is quite interesting to hear. It sounds like someone hitting an anvil with a heavy hammer Usually you can easily hear it, when pushing the throttle from a very low speed on 2nd gear, when on a slight uphill. This was the situation where i was able to induce knock almost every time, probably the mixture getting lean due to TPS not enriching correctly. I didn't bother with this one though, its not dangerous. So you can try this test, to induce knock on your engine, and see how does the graph looks in the log. If there is a difference you can adjust your settings to filter. Hopefully it will filter this noise that you get at midrpm.

Resonant sensors will have much higher voltage output than flat response sensors so the gain and integrator settings will be different between the two types. You will have much better knock detection with flat response sensors. I would set it up so the knock-free Knock Sensor Value gets up to around 1.0-1.2v at redline. The output range of the conditioning chip is 0-5v and you'll want to have knock events register in that range. Setting the gain and integrator so the knock-free the value is 3v will clip the output when there is a ping so the EMU can't see intensity of knock.

what engine do you have?
it could possibly still be knocking, i have an older 6cyl bmw thats really prone to showing knock at around 2-2500 particualry with hot air temps. It could very well be light knock even with very low ignition timing but it could also be something like an engine bearing getting a bit sloppy which would only show up under full load or some other harmonic rattle on one of the accessories or timing chain etc, but one would think the latter would show up under all rpm not just under high load.

Have anyone a one log with Real knock?

RB20DET

probably look something like this

I get no voltage in my knock log ... is this to be setup, or knock is faulty? Wires are checked, knock is sr20det and shows 560Ohm resistance, so according to manual its ok. Any sr20det knock setups to share?