Not a whole lot of s2000 specific information on knock out there for public consumption. I understand why people who make a living tuning cars may not want to share tips beyond basic theory, but what about all of us who want to do it for ourselves?

I have successfully installed a Bosch wideband knock sensor in the factory location, running to an AEM EMS Series 2 using shielded microphone cable. I played around with the knock settings in AEM Tuner and have gotten the knock signal to resemble the factory knock sensor pretty well.

Gain doubled from 0.4 to 0.8, integrator reduced from 300 to 180, and center frequency remaining at 6.64kHz.

I have also used this guide (https://www.nicoclub.com/archives/di...detection.html) to build some DIY knock ears or electronic det cans. An old stock Whisper 2000 was sourced off eBay for cheap, along with a set of battery clamps from Amazon, and more shielded microphone cable (Canare L-2E5 and a Rean 3.5mm jack) was used to assemble. A headphone amplifier is used to further amplify the signal from the Whisper 2000 for listening, after finding that once the electret mic was JB welded to the clamp additional amplification was helpful.

The mic is clamped to the engine block on a rib right next to the factory knock sensor location. I've yet to drive and test but will plan to update this thread after testing.











I also plan to plug the DIY knock ear signal into my laptop, induce light knock and record the signal. This will allow me to use software to analyze the frequency responses of a knock event and determine exactly what frequency my F20C knocks at (hopefully) and use that in the AEM Tuner software if different from the default (and calculated by theory) frequency. This idea came from a Supra forum linked here: https://www.supraforums.com/threads/...-setup.675960/

I'd appreciate anyone else with s2000 specific knock detection experience to chime in on this thread as well.


​​

​​​​​

 

First test completed successfully, which was just a test drive wearing an earbud to verify it being audible during operation.

The clamped electret mic sounds almost identical to the factory knock sensor (which I had tapped headphones into with a 0.1UF capacitor and 100k ohm resistor before replacing it with the Bosch sensor - https://www.k20a.org/threads/how-to-...e-knock.27258/ ).

Next test will be running the signal through the laptop and recording while attempting to induce low rpm mid load light detonation. A bit more of a nail biter than the first test!

 

Great post,

Here is an audio clip of my old f20c with some induced knock @2500rpm

 

Thank you for sharing, Sion.

It is my understanding that this was recorded using a Bosch wideband knock sensor bolted in the factory location and using the Phormula knock detection kit.

Let's do a little analysis on this recording.

Using a free online application located here - https://academo.org/articles/spectrogram/ :



Using Friture - https://friture.org/download.html :



Using Tone Boosters spectrogram plugin - https://www.toneboosters.com/tb_spectrogram_v1.html :



And using Audacity - https://www.audacityteam.org/download/ :



What we can conclude from consistent results between the various analysis software is that most of the spectral energy contained in the knock sound is centered at 5.9kHz and 8.0kHz.

When we look at the image captured from Audacity, we have an analysis of the engine sound before knock on the left, and during knock on the right. By doing this we can differentiate the "background" engine noise, which also tends to have a lot of energy at 8.0kHz, but NOT at 5.9kHz. This makes the sound energy at 5.9kHz a better place to have the EMS look for the sonic knock event signature.

If we examine the HIP9011 chip responsible for the knock control in the AEM EMS s2, we can find specs to indicate the bandpass filter has a Q (quality factor) around 2.4. (https://www.renesas.com/us/en/produc...gnal-processor)

Using the calculated theoretical frequency for the F20C of 6.7kHz (6.64kHz in the AEM Tuner software) we have a bandpass range of 5.4kHz to 8.17kHz. See the following link for more background on bandpass filter theory: https://sengpielaudio.com/calculator...requencies.htm . Using that setting we should be capturing the knock event, however, there may be more background noise present. This suggests we may be better off centering the filter lower, at 5.9kHz (giving us a bandpass range of 4.8kHz to 7.26kHz).

More experimentation is needed.

 

After spending a good bit of time street tuning I have the following updates:

• I was able to capture some light transient knock upon takeoff from 1st gear. This confirmed a lower than theoretical knock frequency, as also observed by Sion's recording, of 5.8-5.9kHz
  • 
  • 
  • 
  • 

• I've been able to customize the knock strategy in the AEM EMS Series 2 to perform quite well for my needs
  • 
  • With this strategy I just ever so slightly tickle the knock detection once in a while during WOT pulls (0.35-0.70 degrees of ignition timing retard on those occasions)
    • 

 

I worked out my AEM knock sensor calibration with E85 and while still NA. I adjusted it to the point it started pulling timing, then raised it slightly higher to keep it from thinking I was in the knock threshold. I've since moved to Haltech and literally adjusting that now as I was getting knock retard at lighter loads on E85. I am probably going to change to a different style of knock sensor, but for the most part, I don't worry about knock with my fuel and low boost levels.

 

Thank you, @yamahaSHO, for chiming in.
​​
Curious if you noticed the actual knock frequency was lower than the theoretical knock frequency also?

Any other details about your knock detection strategy you feel comfortable sharing are greatly appreciated!

 

I did not go deep enough to determine if the frequency was lower. I did the AEM setup over a decade ago, so I only remember getting to the edge of pulling timing from noise on ethanol and then just going just above that. For the Haltech, I did it mathematically and doing the same as I did on the AEM. I actually plan to put gas in it for the first time in over a decade to see how it responds with the FlexFuel setup. Maybe I'll go further with it at some point, unless I don't see a need to. I generally do not run gas, so knock is not something I worry about, especially with 10 PSI or less on the motor.

 

​​​​​In your experience, I can see that there is a good bit of knock resistance under mild-moderate cylinder pressure, which is good to note, and not unlike I've heard elsewhere.

It seems that on the low cam, the knock threshold tends to be met more easily than the high cam due to dynamic compression.

It also seems there are trends in ignition timing (which differ between the F20 and F22) suggesting the lower rpm efficiency increase to be significantly reflected in ignition timing requirements

​​