Re: Knock Sensing with MACXXX ECU...

["Bernard Littau" <bernardl@acumenassociates.com>]

[snip]
>Anyone electronically and technically
>inclined have insights into how the MACXXX ECUs do the knock
>sensing? Is it simply the conditioning circuitry listening to the
>knock sensor/microphone and filtering out all frequencies below
>the "knock" and then outputing a "high signal" when the frequency
>of knock is heard? What would be the "knock frequency" of the
>MC motor? Like a high-pass filter/crossover. Or am I as usual on
>the wrong track here and should just rip open a Hitachi ECU and
>"super-reverse-engineer" from the knock input to the lines to the
>CPU top half? :)


As usual, I lack Audi specific knowledge, so YMMV:

The knock sensor is a piezoelectric microphone, embedded somehow so it is
insulated from much of the noise of an engine.  When it hears a loud sharp
noise, like a knock, it sends a large signal to the ECU.  I believe the
signal is threshold sensed at the ECU -- the microphone sends some low level
signal all the time while the engine is running. (I believe this is how the
ECU can sense bad knock sensor -- if it gets no signal, but I digress)
There is no frequency analysis going on, it is strictly level driven. (The
microphone is very likely to be embedded such that it hears some frequencies
better than others, so some spectral filtering is going on, but there is no
active signal analysis on the ECU).

When the ECU senses a large signal from the knock sensor, it retards the
ignition. (Again, guessing, on the Audi, the ECU may do more, like mess with
the mixture to make it richer, as well as retarding the  ignition timing.)

One can test the knock sensor on many cars by using a timing light.  The
test is to sharply strike the engine with the plastic handle of a
screwdriver -- you can see the timing retard for a moment each time you
strike.  I need to try this on my Audi...

Best,

Bernard Littau
Woodinville, WA
'88 5kcstq