SV: Ignition Controllers

[Stefan Olsson]

----------
> Fr=E5n: Tim Drury <tim.drury at gtri.gatech.edu>
> Till: 'diy_efi at coulomb.eng.ohio-state.edu'
> =C4mne: RE: Ignition Controllers
> Datum:  den 25 juli 1997 16:28
>=20
> >Does anyone out there have any advice as to the best method to impleme=
nt
> >this ? Ideally what I would like to do is advance the timing to the
onset
> >of knock and then retard it slightly back into the safe zone. The
> >controller will be used on a fuel injected 12A turbo rotary engine.
>=20
> The first thing I would do (and I already did this), is find out what a
knock
> sensor really is.  Here's a starting point: a knock sensor is simply an
> accelerometer tuned to high frequencies.  Knocking causes the engine
> block to vibrate at high frequencies.  Which frequencies?  Where in the
> block is the best place to measure?  Good questions. =20
>=20
> As far as where, I've heard the rule of thumb is to locate it fairly
centrally.
> In the case of large motors with lots of pistons, sometimes 2 knock
sensors
> are used.
>=20
> At what frequencies?  It depends on the engine, which is why I didn't
include
> this feature in my system.  Engines knock and resonate at different
frequencies.
> You can look at all the technical papers you want (and this is a good
idea
> to learn how knock detection works) but unless you see one for your
application,
> the frequencies may be a little off.
>=20
> Someone mentioned the Harris knock detection chip.  I've seen this (it
was years
> ago) and believe it could be tuned to detect knock in a narrow band.  I=
t
probably
> is nothing more than a switched-capacitor bandpass filter, but this wou=
ld
be a
> good start.
>=20
> Without previous knowledge of what frequency to tune to, you'll have to
experiment.
> We've tried hooking a knock sensor up to a graphic equalizer with 10
bands.  We
> didn't have any luck; I believe the bands were too broad.  If cost were
no object,
> I would install a knock sensor, hook its output to a spectrum analyzer
tuned for
> narrow bands between 5kHz and 20kHz.  Then advance the timing on the
motor
> _very_ slowly and look for spikes to occur.  This is _probably_ knock.=20
Be very=20
> careful though, too much knock for too long and you'll blow a hole
through Mr. Piston.
>=20
> The cheaper method would be to install the Harris chip or a narrow
bandpass filter
> to the output of the accelerometer and run the same test as above witho=
ut
the
> spectrum analyzer.  Manually tune across the band using the filter and
look for
> higher voltages coming out of the filter (use an envelope detector).=20
This method
> requires more time and more knocking from the motor.
>=20
> And to be honest, I really don't have a clue what frequencies knocking
occurs at.
> 5-20kHz is just a guess.  Read some SAE papers to get an idea.
>=20
> -tim
>=20

Have any one seen the instrument that Link is offering ? Its called
Knocklink.
Its made so You can tune it to hear pinging from different engines and it=
s
cheap to ($100,)


Stefan