[acannell]

So I cracked a piston and wanted to see if I could hear any detonation (after it was fixed...lol)

I've read about homebrew ways to listen to your knock sensor, some on here, some on google.

Here is my setup, and a sound clip too!

All the parts for this were bought at radio shack:

Pocket Speaker/Amplifier (277-1008)
9V battery (230-0875)
Mini alligator clips (270-0374)
RCA to 1/8" mono adapter (274-0330)
12ft bare wires to RCA mono plug (420-2372)
1/8" mono to 1/8" stereo jack (274-0882)
Headphone volume control (420-2559)

And you will need some headphones.

A brief summary of how all this works:

The knock sensor gets unplugged. You connect it via alligator clips and the 12ft cable to the headphone volume control (to attenuate it, NOT to connect it directly to headphones), and then to either the pocket speaker/amplifier, or a laptop. If you will be using a laptop, you dont need the pocket speaker/amplifier. Anyways, from the laptop or the amplifier, you can connect it to headphones. With a laptop you can record the sensors output, which is nice because you can listen carefully to it later when you aren't driving, or share it with other people on here and see what they think about certain sounds.

A more detailed description:

The 12ft RCA/wire cable will connect the knock sensor to the inside of the car to whatever you will plug it into for recording or listening. Its got two wires sticking out of one end of it, and a male RCA plug on the other. Solder an alligator clip to each of the bare wires. Unplug the knock sensor from its harness and connect one alligator clip to the sensors little stud. Connect the other alligator clip to a nearby ground (I used the O2 sensor mounting bracket on the passenger side of the intake manifold). You should extend both the bare wires and heat shrink their solder joints so they dont short on anything.

Then run the cable inside the car (i just do it through the window). Plug the RCA to 1/8" mono adapter into the end of the cable. Then plug THAT into the headphone volume control.

The headphone volume control is required because I noticed that the output of the knock sensor was overloading my laptop when I was recording its output. (clipping) Even with the laptop's microphone setting at absolute minimum it would still clip at anything but idle. Basically the sensor is just TOO LOUD for a microphone or line-in input, so the volume control lets you attenuate its output.

Now if you have a laptop, you can connect the output of the headphone volume control to the laptops microphone in jack, or maybe the line in jack. Experiment and see what works for you. Just remember to adjust things so that when stuff gets LOUD (anything greater than idle really) the output of the knock sensor doesn't start overloading whatever is attached to it. (this is really only detectable by looking at the sound waveform with a laptop or an oscilloscope).

If you wont be using a laptop, plug the headphone amplifier output into the pocket speaker/amp and plug your headphones into the amp too using the stereo to mono adapter so you get sound in both ears.

Off you go!

Attached is a picture of the setup (minus the heapdhone volume control, which I got after I took the picture), and a link to a sound clip.

SOUND CLIP!

The sound clip is me idling at a light, then turning left and flooring it through 1st and 2nd. Peak boost was about 5 psi, as I have the wastegate connected directly to the intake manifold at the moment.

knock sensor 5 psi.wav

Anyone hear detonation in it? What does detonation sound like?

I recommend you download this file and play it on headphones.

 

[puppet]

Very slick :thumb:

Now the dumb questions .... are you listening to what the sensor is reading(hearing/sensing) or actual knock?
Can your device be made to display in a visual format ... like a NB A/F meter?

 

[acannell]

im not sure how one would listen to "actual knock". im listening to the direct output of the sensor, which is basically like having a mechanics stethescope on the engine, except that i can listen to it while im driving.

displaying the output of the sensor on a gauge would be like displaying the output of a microphone on a gauge. doesnt really make sense. unless you do a lot of post-processing of the signal to discriminate between knock and no-knock...but then the whole point of this is to use my brain as the post processor because i dont trust the ECU

im definitely curious what real knock looks like in the sensor output waveform.

 

[puppet]

OK .... does the sensor make noise? I don't know if it does but if so, how do you differentiate that from actual knock?

The web homebrews use a direct connection to the head ... much like the can and string approach to listen for knock.(this is more like a mechanics stethoscope) Just wondering if the electrical output of the sensor can be processed by the brain ... better than the ECU. Acoustical feedback should be easy once you know what your listening for. Don't know what if anything the sensor adds to the noise? What's the scoop on that part?

 

[usetaboost]

can't you just hook a knock alarm up to these cars?

 

[puppet]

Forget the company but there are audible and visual knock readers available form them.
Most of the fellas modify the calibrations to include a "check engine light flash" on knock. Not quite sure how knock specific this mod is however. I know for a fact that it'll pick-up internal block noise and flash the CEL. I busted a crankshaft and would get the flash for months before I realized ... no take that back ... before the crank let me know that it was about to let go completely.

So readings can be false.

 

[risen]

Basically the sensors in our cars generate a 0-5 v signal from a pizeoelectirc element that is excited by vibartion. The ecu reads this signal and based upon a table determines wheter or not there's knock ( the table is sensor volts vs rpm). If you could watch the 2 inputs to the table you could pretty easily monitor it yourself. However you're not going to be able to pilot a TD in a straight line @ WOT and read rpm vs volts quickly enough to do anything. Not to mention that the ecu already does this with precision enough to pinpoint which cylinder knocked.

Anyways, if you feel that the ECU isn't sensitive enough for your liking you could always edit the table and make it more sensitive. Interesting project though.

 

[speeduphoria]

Personally I would just use a bolt or copper tube and glue a microphone(s) on it and bolt it to another hole on the block. Then keep the knock sensor hooked up. A guy on neons.org did this and said he would hear a lot of exhaust noise.

I dont think listening directly to the knock sensor is going to help much because it will get louder with rpms. Like said above the computer listens to this but only at certain points and it has a tolerance that increases with rpm to decipher knock.

 

[usetaboost]

Definitely not trying to steal a thread

Caspers electronics sells some nice ones for the TB community but I'm sure they'd work on the dodges also. I also read that the computer pulls fuel to slow the motor down if it gets about 14.4 or so lbs of boost. How can that be true? If the motor went lean that definitely wouldn't be good

 

[acannell]

Re: Definitely not trying to steal a thread

wow its amazing how much mis-information there is on this topic

1) ever hear an engine knock? just with your own bare ears outside the car? thats the exact same thing you are doing by using the knock sensor as a microphone with an amplifier and headphones. the only difference is its far more sensitive because its in contact with the engine, so you can hear knock that would otherwise be inaudible, and you can also listen to it while you are driving. you are not trying to process electrons with your brain. you use the knock sensor as a microphone and listen to its output, EXACTLY as you would with a normal microphone.

2) THE KNOCK SENSOR IS A MICROPHONE. it does not output a 0 to 5V output signal. it does not have a "voltage input". it has an OUTPUT (the stud on top) and GROUND (the intake manifold). THATS IT.

3) the check engine light flashing feature on modified ECU's depends on the ability of the ECU to distinguish knock, which is questionable. thats the WHOLE POINT of listening to the knock yourself with headphones.

4) listening to the knock sensor is a FANTASTIC way to identify knock. it doesnt matter if the engine gets louder with rpm. it is a COMMON PRACTICE to listen for knock while tuning on a dyno by listening to the knock sensor or something like it (stethoscope, etc..)

5) there are reasons to believe the ECU is too sensitive to knock, especially at higher boost levels than stock, when the engine is even louder. this means the ECU may detect false knock and pull timing when THERE IS NO KNOCK. many variables could affect this. its unclear just how good the ECU really is at detecting knock given the modified and/or worn state of our engines. THATS WHY WE WANT TO LISTEN TO THE KNOCK SENSOR OURSELVES.

 

[speeduphoria]

Re: Definitely not trying to steal a thread

1) ever hear an engine knock? just with your own bare ears outside the car? thats the exact same thing you are doing by using the knock sensor as a microphone with an amplifier and headphones. the only difference is its far more sensitive because its in contact with the engine, so you can hear knock that would otherwise be inaudible, and you can also listen to it while you are driving. you are not trying to process electrons with your brain. you use the knock sensor as a microphone and listen to its output, EXACTLY as you would with a normal microphone.

2) THE KNOCK SENSOR IS A MICROPHONE. it does not output a 0 to 5V output signal. it does not have a "voltage input". it has an OUTPUT (the stud on top) and GROUND (the intake manifold). THATS IT.

1) The knock sensor is not a normal microphone, so the sound will be different then audible detonation "pinging". Here Is the 1st site that came up and has a sound recording of what knock sounds like with the sensor the way you have it hooked up. Now it may be a different sensor so the pitch may be a little different then yours(sounds like a bird chirping).
Knock Sensor Sounds
another
THE KNOCK BOX - Knock sound recordings

2) Again the knock sensor is not a normal microphone so the sound of detonation is going to be different then what you would normally hear. All knock sensors output voltage(at least 0-1V), the ECU doesnt listen to sound it reads voltage, there are voltage limits in the cal that it uses to decipher knock. Many people who do custom cals also change the threshold in the cal so it doesnt pick up false knock. Look on D cal or mopar chem for more info.

I find it amusing that you always come on here like you know what your talking about and have an attitude. Your the one having all of the problems, I would hope you would tone it down a little and be open to discussion.

 

[usetaboost]

wow its amazing how much mis-information there is on this topic

So what you are saying is I can crank the boost on my we4 and stap I microphone under the hood to listen to "audible knock"?

 

[acannell]

Re: Definitely not trying to steal a thread

I wasn't saying that detonation heard via the output of the knock sensor would sound like detonation heard outside the engine with your ears. I'm saying that the output of the sensor has a very fast response rate and lies within the audible spectrum.

Yes I realize that the ECU doesnt have an ear and doesnt listen to "sounds". Kinda like I dont hear electrons yet somehow I can hear my stereo.

Any transducer can have its output converted to sound with another transducer and if its spectrum is in the range 20hz to 20khz then it can be heard by your ears and therefore processed by your brain. Its still a SIGNAL. Its totally irrelevant where the signal is coming from. You can take a photodiode and convert the light from your florescent overhead lamps into an audio signal and hear the 60hz hum. The information is in the signal you are just converting it into a form your "inputs" can read.

Just like they still use people to listen to sonar signals on submarines and (strangely ignored) just like they still use people to listen TO ENGINE KNOCK ON DYNOS, people are just better at processing sounds sometimes.

 

[acannell]

Well my plan is to start at low boost, listen for knock with the "microphone", check the plugs, etc... Then turn up the boost a little bit more and repeat. If you want to just turn your boost up all the way THEN listen you might need a new engine!

 

[usetaboost]

just kiddin around man:thumb: I mean c'mon-I do own a turbo car

 

[speeduphoria]

Well my plan is to start at low boost, listen for knock with the "microphone", check the plugs, etc... Then turn up the boost a little bit more and repeat. If you want to just turn your boost up all the way THEN listen you might need a new engine!

I agree that you have to step up a little at a time.

I would do as the guy in the link posted did and splice into the wiring instead of disabling the knock sensor.

I think you should get some equipment and learn to make your own cal, I think this is the best way. Besides making more power safely you also learn a lot more(or are forced to learn more).

 

[usetaboost]

WHOOOA, disabled knock sensor?? Livin on the edge. Why try to reinvent the wheel and just hook a knock alarm up to it? If I"m wrong here someone help me out. Maybe the TB and TD worlds are different but I'm just here to learn cause I found a 88 shelby z I might be buying when the old man that owns it decides to move away.

 

[puppet]

It's not too uncommon for guys to disable the sensor on track days. If your tuned just so it's not a bad idea to insure best 1/4 mile times.

acannell ... it's why I asked if you were listening to the sensors output or using the sensors mounted location as a acoustical post. There is a difference. If you figured out how to convert the sensor output to audible sound .. cool. Now, heh ... how do we understand the music? Pretty sure the sound the sensor makes as a result of knock doesn't sound like knock to us. Like a language that needs translation.
Is an oscilloscope needed to decipher your output? Like you asked before then ... what does knock look like as a waveform?

... me head hurts.

 

[acannell]

i dont know exactly what detonation sounds like as a sound. speedeuphoria posted a link to a website where there is a sound clip of detonation, and on that guys engine with that guys setup it sounds like a twang.

there is no difference between using the sensor as an acoustical post with another microphone, or using the sensors direct electrical output. both involve using a transducer which is attached to the block. both involve converting the electrical output of a transducer into sound. the only difference is the knock sensor has a narrower bandwidth than your typical microphone, which should make it easier to pick out the right sounds.

the good thing about using a laptop to record the knock sensors output is that you can visually observe the waveform, just like an oscilloscope.

the bad part is you cant trigger on ignition pulses.

this whole ordeal is because I want to PROVE there is knock in my engine. i dont like the fact that the only evidence i have is the ECU thinking there is and a cracked piston. i know how this sounds, that should be evidence enough right?

the problem is the wideband AFR gauge said 11.4 or below, my EGT's were probably below 1600F, the plugs never showed signs of detonation, i have a huge intercooler, and new injectors. i think the piston cracked because the hall sensor was bad and cut fuel several times at high boost, and/or the fuel injector harness is intermittent.

i really need to nail down what is going on here and im sick of guessing. i want to actually OBSERVE knock.

 

[usetaboost]

It's not too uncommon for guys to disable the sensor on track days. If your tuned just so it's not a bad idea to insure best 1/4 mile times.

That's making my ENGINE'S HEAD hurt :bash: You guys are nuts. Unhooking knock sensor and running down the track? Hell why have it hooked up during street driving then. I've seen bad stuff from knock, not just cracked heads.
When it comes time to get my shelby z this will be one thing I won't do. If I'm wrong about this whole thing maybe you guys should teach some of the buick crowd ( turbobuick.com )about this. I am obviously a little uneducated on this subject