Running wire for knock sensor GGSX

[teet]

For the Archives:

When you have to run a wire for the knock sensor in your turbo project Galant GSX...you have 2 wires. One ground and one that plugs into #7(I think...but could be wrong) in the ECU harness.

The factory wiring from the sensor to the ECU is SHIELDED.

The factory ground goes from the sensor...connects to the shielding around the signal wire (side note...this is a cheesy connection...might have something to do with phantom knock at either end of the ground...)...all the way to the factory ECU...then connects to the ground where the ECU grounds to the chassis...NOT through the ECU in any way.

The signal wire goes straight to the ECU.

IF you run it just a normal wire to the ECU...and a ground straight to the Chassis at the engine...thinking...eh...it should work /ubbthreads/images/graemlins/wink.gif It won't. Your car will knock so loud you can hear it /ubbthreads/images/graemlins/shocked.gif and you'll nearly wet yourself when you realize it isn't something rattling in your engine bay...its knock. /ubbthreads/images/graemlins/crazy.gif But cause your "knock sensor" isn't reading anything bad...you think..nah...that can't be knock. Wrong.

I fixed the wiring now I have a knock sensor that works. /ubbthreads/images/graemlins/blush.gif Now comes the boost /ubbthreads/images/graemlins/wink.gif

john

 

Actually, running a knock sensor un-shielded with standard wire and a ground to the engine bay won't result in a knock sensor that "seems" to read ok. It won't read correctly at ALL and you'll likely get loads of phantom knock.

See, the knock sensor should read in hertz, not voltage and when you ground through a chassis ground and use unshielded wire, whatever signals are floating around will likely interfere and cause the knock sensor to get ultra sensitive.

The knock sensor must use shielded wire and should not be ground to the chassis...At some point, someone had replaced a knock sensor on my car and for some reason they hacked off the plug on the sensor and replaced it with a generic GM type plug. They wired into the harness using standard wire and ground to the chassis...

I recently had it all replaced with a brand new mitsu knock sensor and a spare harness. All is well now /ubbthreads/images/graemlins/wink.gif

 

[teet]

Sorry. You are wrong. The end of the stock wiring is grounded directly to the chassis...under the dash. Not through the ECU. Look. Trace it back. I did.

The knock sensor reads out in Volts...not hz. 0 to a max of 5 volts.

I ran this way for a while (unshielded, grounded to chassis in engine bay)...and logged with the AEM the direct voltage output...which reads direct sensor voltage. It read some voltage, but nothing much over .5 volts...when I should have been reading like 2-3 volt spikes.

I'm not talking from "theory" on how I "think" it should work...i'm talking from direct experience.

john

 

talk to keydiver about how the ecu reads knock. just cos there is volatage on the wire, doesnt mean thats what the ecu reads...there might be something other than voltage. i always thought the ecu read knock in hz as well, and judging by the code on the ecu, i think it recognises hz too...
I may be wrong, will someone shed some light on this matter...

 

Here you fellas go. The knock sensor generates voltage in that it is a transducer that converts mechanical impulses to a voltage by means of moving magnetic coil principle. It is a microphone for all intents. I don't know it's sensitivity (voltage produced for certain level of excitement). It is resonant at a frequency centered somewhere around 6kHz. I haven't had one characterized in a lab yet but that is the fundamental frequency of knock on this engine configuration (based on study of knock sensor output) and that is what the similar toyota 3s-gte from the gt-4 all-trac sensor's resonant frequency is. It resonates (exaggerate) at 6kHz its input so as to make other common engine vibrations mute in comparison. Think of it this way, you have a microphone that is for all intents, only sensitive to what you are interested in studying. That way, lifter and valve clatter is harder to confuse as well as other engine vibrations. So, in effect, you were both on the right track, the sensor is putting out this waveform that can be characterized by it's magnitude (voltage) and frequency. The ECU has signal conditioning circuits to further eliminate noise from the signal (maybe bandpass filtered) to help keep the crap out of the ECU.

Here is a recording from a DSM knock sensor:

http://home.netcom.com/~bsundahl/knock/sound/rtb11_1_10.wav

As you can tell, most of what it is picking up is valve train noise.

Here is the (excellent) website describing the DSM knock sensor output as studied with forier transformations.

http://home.netcom.com/~bsundahl/knock/sound/KnockSounds.htm

Here is an instrument capture showing what I believe to be representative of our sensor's frequency response:

and that fellow's excellent site pertaining to knock sensors:

http://deviantmethods.com/bigmoose/pages/knock.htm