Knock alarm

[Tony Bryant]

> Is there somewhere I can go to learn more details about how knock sensors
> work? 

There an Appnote on TI's DSP web site (Try finding via alta-vista)

> It sounds like they use a resonant sensor, chosen & tuned for the
> application, and maybe look for a build-up of oscillation?
> 
> I had this idea to use something like a (non-resonant) quartz transducer,
> maybe one per cylinder, to get a signal when each cylinder fires. Then I
> could monitor the signal under normal conditions, to get a baseline. A
> knock should give a stronger, shorter signal, a bit earlier, right? Even if
> it needed a DSP, and tuning for each engine, it'd still be useful to me.
> 
> I had heard that some aftermarket ignition system had this kind of setup,
> but I can't find any info on any such thing. 
> 
>    TIA,
>    Chris Conlon
> 
> p.s. Hi, I'm new :)
> 
> 

I did once see an project in a hobby electronics magazine (Silicon 
Chip, I think) - Can't rememeber the exact details, but I remember 
some facts.

It used a piezo transducer from a microphone.

It only checked between the ignition event and a few degrees (ms?) 
after that to reduce false occurences.

It stated that the Knock frequency is roughly inversely proportional 
to the bore size. (i.e. the period of oscillation is proportional to 
the bore size)

It was a pretty simple cicuit, three or four op amps in a reasonably 
tight bandpass filter arrangement.

I suspect that this circuit could pick up serious knock that you 
could hear yourself, but the subtle detection needed for running 
closed loop may be beyond its capabilities.

r.e. the DSP idea. Good idea, but it could be a problem trying to 
sample a single baseline, I suspect the average engine differs wildly 
acoustically across its rev range, as various resonances come and go. 
You may need a baseline sample for each band of 1000 or even 500 rpm.



$$$ MAKE CAR FAST $$$
bryantt at psc.fp.co.nz