Diy Knock Monitor?

[StealthX]

Hey sau.

Stumbled on an idea today, dunno if it's feasible...

The stock knock sensors send a signal to the ECU right... what sort of signal is it? 0-5V sorta thing or is it a frequency?

What I'm getting at - is it possible to hook up an indicator light straight to one of the wires of the knock sensor? Sorta like the PFCs in skylines how they flash the engine light...

The other thing is, is there a way to simulate knock to test if such a device would work? Try tapping different size spanners on the head or something?

What other DIY knock monitoring can I look at? I know that my knock frequency is 6.4kHz, you can't tune some sort of modified radio into that freq. can you? Even short wave AM freqs are still hundreds of kHz right?

[GTRNUR]

Knock sensors are basically microphones.

Detonation when it occurs in a cylinder induces a specific sound at a specific frequency within the cylinder block. This frequency is determined by the bore size and block design/material.

The knock sensors pickup this sound, along with all other engine noise, and pass it along to a band-pass filter which filters out all but the frequency in which detonation has occurred. The signal is then sampled with an analogue to digital converter, where it produces value that the ECU can make use of. In the case of an FC, it is a byte value (0-255). The ECU can then act on that data based on thresholds you define relating to knock. Different ecu's have different capabilities here of course.

The nature of how the Power FC does this internallly is anyone's guess. But on an RB as there are two sensors, I would say that the sensors sample an averaged signal amplitude every 3 ignition pulses from the front and rear cylinder banks.

6khz is an audable frequency, so knock sensor signals can also be fed into an amplifier and listened too. This is basically how the knock box systems works. You mount the knock sensor on the intake manifold and it will pickup noise reflected back past the intake valve when detonation occurs.

[Adriano]

Added to that, factory knock control like late evo, only detect knock when the pistons are at a certain range of position in the cycle to further increase the accuracy

[GTRNUR]

That's smart of the EVO's for sure. Faster ECU processors, faster ADC's and another 10 yrs of ecu development make a difference. Allows false positive noise detection to be reduced too, like when the engine moves around under accel/decel.

[DVS JEZ]

Afaik Most standard nissan knock sensors have a wider frequency so they can read false knock especially when when the motor has been modified.

[rob82]

Afaik Most standard nissan knock sensors have a wider frequency so they can read false knock especially when when the motor has been modified.

Hmm I was led to believe that they were tuned frequency sensor's not the broadband style. Pretty sure the are tuned piezoelectric sensor with an output around 0-1v depending on knock intensity/frequency.

[DVS JEZ]

U may be right. I have never tested only what i have heard.

The setup that Status is designing sounds exactly like what the OP is thinking

[naga]

Fairly straight forward, Silicon chip have published an article and designs to do this: http://www.siliconchip.com.au/cms/A_108910/article.html

Basically takes the audio from the microphone/pzt sensors feeds it to preamplifier -> bandpass filter -> rectifier + low pass filter. If you didn't understand that it's essentially a sound level meter which is only sensitive to knock frequencies.

I copied their schematics and tuned the filter to the rb25. So far i've only bench tested it with an occiloscope, will be installing it in my car soon and hooking it up to the inputs of my techedge wideband.