Knock listener
I have only combined different knock listeners to this awesome design. The first part is a 100mm long 10-12mm copper pipe that is hammered flat in one end. A hole is drilled in the flat section, bolt the copper pipe to the engine block close to the deck plane.
Connect a soft hose to the pipe and run it into the car, now drill holes in a pair of ear protectors and install hose fittings in the holes. Run hose from the hose fittings to a hose tee, also attach the soft hose from the copper pipe to the T. Done!!!
This seems to conduct the sound (pressure waves) to your ear.
An adaptable analog design to enhance the above scheme
What does it have to do with the above sound-wave-guide mechanical solution (where the signal aint go to electronic domain at all, right?) ?
| [Analog Amplifier] |
Are you talking about a board with a microphone input, digital signal conditioning in the ARM and a headphone amp? I don't think that anything short of that would compare well with an analog listener. My only doubts is if the analog design is worth the effort when comparing with the hose and earprotector design.-Jörgen
No.
I think the main problem is finding the knock frequency. The FFT helps you find that. It's easiest to do with a notebook soundcard, but it is theoretically possible to do with the ARM ECM.
When you know the freq, you can configure TPIC8101 so it will tell the knock level (Will a narrow bandpass earphone amplifier won't be of much use than ? ).
If you don't know the freq, you cannot do much better than apply a wide bindpass filter (appr. 6..10 passive components, RC) use a simple earphone-amplifier (easiest from an old walkman), but that won't help you find the knock freq unless you have absolute-freq hearing. There are guitar-tuning circuitry out there that might help to differentiate between close frequencies, but the notebook soundcard (or other digital sound recorder) and FFT seems a better idea.
I also think that the hose design is simple and nice.
The knock listener is not meant for finding knock frequency, it's meant to listen for knock with. The bandpass filter should be wide. Something like 3-10k. Listening to the engine can't be replaced with electronics. An experienced engine tuner can often pick up knock before it is clearly visible on the in cylinder pressure graphs. -Jörgen
But letting the ARM FFT the signal from the knock sensor is a good idea too. Beeing able to log that information would be of huge interest. -Jörgen
Back to DetonationDetection