Knock sensor sound

[Yves]

Is it possible to listen to the knock sensor sound via tunerstudio on the laptop ?

[yzn]

You can without tunerstudio just connect the knock sensor to microphone jack on the laptop

[Yves]

I remember something about that not working, but haven't tried it myself.

[Tjabo]

Does anyone know if it will work to split the signal from the knock sensor(s) so that one fork goes to the internal knock module, and the other fork goes to a small amplifier for a listening device? I thought this might be a good method for learning about the reliability of settings for the knock module. I'm just starting to try to learn about the knock module function.

Also, after searching around this forum and Googling for general results, I'm not really finding any success stories with the internal knock module yet. Are some people having success with them so far?

[Yves]

My sensor is picking up a bunch of noise. With the drive I made today, it was reacting during a decell. Looks like exhaust/valve train noise.

Not sure what to do. Options are limited around here what MS and a dyno is concerned. So I have to rely on my butt dyno

[Tjabo]

I was looking around for the explanations of the tuning parameters today, but I didn't find them. Are they in the Tunerstudio help files? I need to get a reminder of my license code for Tunerstudio I think, so I struck out on that effort also. Haha...

But anyway, I thought there was something like a band pass filter effect in the settings for the knock module to help you tune out the frequencies where your valvetrain noise is? I have used some electronic det cans a long time ago that I made, and they worked quite well. That is how I would want to verify the correct operation of the knock module if I tried to use it. It would require either splitting the signal of the knock sensor in use, or using an additional one.

[Yves]

I was thinking of an electronic det can, but have gotten some mixed comments about it's usefullness. Some say there is too much noise with difficulty distinguishing actual knock. If I look at my current signals, this would somehow seem to be the case.
The next question about it is : can you just plug the sensor output to a mic output of the laptop ? If so to what part of mic jack do you connect the sensor and to what part the ground ?

As for bandpassfrequency ... they only thing so far I've seen is knock frequency, time constant itegrator, gain and the window. I was contemplating going to use the second order frequency, but not sure how much that would do.

As far as my knock signal is concerned, I can clearly see that it follows the rpm, in the sense that low rpm = low knock %, high rpm is high knock %. With some notable exceptions, being that decell when backfiring in the exhaust also gave a pretty high signal. (happened during a test without AE on).

Tuning the ign tables is a mayor PIA.

So far the only useful thing I could do was looking at spark plugs, but even then...
I have considered using a high octane race fuel like the Sunoco leaded, but even then if the other noises at that frequency are too much, I don't think I can see much. And at € 5 per liter it is an expensive experiment.

[radial]

Do not use knockdetection for ign tuning. Get on a dyno for that.
You will usually loose power 3-5* before you enter knock, and for that reason alone knock ignition tuning is totally incorrect and a waste of time.
But as an extra protection against brakedowns at lower rpms? Yes.

Depending on knock sensor type, a real knock is supposed to be very easy to detect. Especially with the OEM "tuned" knock sensors(not the wideband style) that amplifies knock signal 5-10 times if knock occures.

But I think you are right on the trial approach.
But its probably cheaper to just remove all boost, and pull away 5-10* of timing cross the map (impossible to knock).
Then filter out the noise, and give it a try with boost and regular ignition when you are confident with the filtering.
I would probably ignore knockdetection above 5k with the window function though...it's not where it should knock anyway, and detection becomes difficult anyway.

[Yves]

Thanks. Dyno's are however not plentiful around here and those who do a dyno, usually sell their own EFI stuff (mostly MOTEC), so are not willing to touch any other system.

I have been playing with ignition timing and since I have a maf, I could see in some instances airflow went up with a certain timing retard. If actual airflow and therefor AFR in the exhaust changes with timing, it could be that you can adjust timing untill you reach the highest amount of VE (I use incorporate) at a given load point. If I'm not mistaken this should be MBT. So you could treat the point at which knock occurs as the maximum and retard timing from there untill you hit maximum VE.

I was contemplating on only using the knock sensor for protection but currently it seems to be rather useless due to all the extra noise.

[bradyzq]

Do not use knockdetection for ign tuning. Get on a dyno for that.
You will usually loose power 3-5* before you enter knock, and for that reason alone knock ignition tuning is totally incorrect and a waste of time.
But as an extra protection against brakedowns at lower rpms? Yes.

That comment is dangerous! It implies that you can keep advancing perhaps past MBT without the worry of knock.

When knock occurs is VERY dependent on fuel and engine design, not to mention manifold pressure. It may be that you encounter knock BEFORE reaching MBT.

Yves, your idea of listening to the noise the knock sensor hears is a good one. Your ears may be able to differentiate between mech noise and knock, allowing you to tune the knock control correctly. Of course you still have to tune the spark maps...

[Yves]

Do not use knockdetection for ign tuning. Get on a dyno for that.
You will usually loose power 3-5* before you enter knock, and for that reason alone knock ignition tuning is totally incorrect and a waste of time.
But as an extra protection against brakedowns at lower rpms? Yes.

That comment is dangerous! It implies that you can keep advancing perhaps past MBT without the worry of knock.

When knock occurs is VERY dependent on fuel and engine design, not to mention manifold pressure. It may be that you encounter knock BEFORE reaching MBT.

Yves, your idea of listening to the noise the knock sensor hears is a good one. Your ears may be able to differentiate between mech noise and knock, allowing you to tune the knock control correctly. Of course you still have to tune the spark maps...

Have you used a headphone set to listen to knock ?

[bradyzq]

Yes, but not with Megasquirt. You can probably separate the knock from the noise by listening.

Regardless, with headphones, you should be able to build your knock threshold/backgound noise curve.

[bradyzq]

Actually, yes with Megasquirt. 3SGTE PNP ECU with the built in headphone jack.

[Tjabo]

Brady, how does it work to listen to the knock signal through the 3SGTE PNP ECU? Does it provide some sort of a speaker out signal or something like that which allows you to wire in a headphone jack?

[yzn]

i remember that i have connected the upper part of an old microphone jack to ground and the other wire to the tip of that jack, it's a bit noisy but the knock sound is distinguish from other sound. if you don't want to make the wiring efiexpress sell and adapter cable :
http://www.efiexpress.com/catalog/artic ... icles_id=4
http://www.efiexpress.com/catalog/produ ... cts_id=112

[radial]

Do not use knockdetection for ign tuning. Get on a dyno for that.
You will usually loose power 3-5* before you enter knock, and for that reason alone knock ignition tuning is totally incorrect and a waste of time.
But as an extra protection against brakedowns at lower rpms? Yes.

That comment is dangerous! It implies that you can keep advancing perhaps past MBT without the worry of knock.

When knock occurs is VERY dependent on fuel and engine design, not to mention manifold pressure. It may be that you encounter knock BEFORE reaching MBT.

Yves, your idea of listening to the noise the knock sensor hears is a good one. Your ears may be able to differentiate between mech noise and knock, allowing you to tune the knock control correctly. Of course you still have to tune the spark maps...

The statement was to provide an example, because usually you do loose power before entering knock on normal fuels. Thats why ignition tuning based on knock is a waste IMO.
That's also what Greg Banish tells through his manual.

[Yves]

I have to argument as follows : the point if tuning in a knock limited engine is useful if you run incorporate afr. VE is then close to real ve. So if you encounter knock say at 30* you can back of on timing until you see ve go down

[bradyzq]

The statement was to provide an example, because usually you do loose power before entering knock on normal fuels. Thats why ignition tuning based on knock is a waste IMO.
That's also what Greg Banish tells through his manual.

No! I disagree! I have seen many instances where I had to stop advancing the spark due to knock onset, and I still had not reached MBT. And I have a dyno (I agree with that part!).

I am not suggesting that you dial in your ignition map by blindly advancing til knock then retarding by a couple of degrees. I just want it to be clear to whoever reads this that there is NOT always a safe buffer of however many degrees of extra spark advance past MBT during which power stays about the same PLUS another 3-5 degrees.

[bradyzq]

I have to argument as follows : the point if tuning in a knock limited engine is useful if you run incorporate afr. VE is then close to real ve. So if you encounter knock say at 30* you can back of on timing until you see ve go down

Yes, although the engine would not be knock limited in your example! On a knock limited engine, you back off timing until there is no more knock, losing power in the process.

Try to wire in a headphone jack, or make a little standalone knock amp/sensor. You can even make a mechanical knock listening device from copper pipe, flexible hose, and a set of ear defenders with holes drilled in them for the hose to enter. Bolt a flattened end of the copper pipe to a manifold stud or something near a combustion chamber. You'll be amazed by what you can hear, and, more importantly, that you can pick out asynchronous sounds that are more likely to be knock.

[Yves]

That is going to be my next move.

I made the point because last weekend I did a test at idle where I usually run 18.5° of ignition advance which gave me best vacuum. Have to add I also run a maf in my setup which is not used for fueling but it is logged. Closed loop iac was on as well. I could see the airflow go up through the maf until about 12°. By that time I was running one point leaner than usual. I readjusted the mixture to get the same afr and saw a little reduction in airflow. Map went up from a low 68 kPa with 18.5° to 71 kPA with the 14.5 AFR and when going back to 13.5 AFR I got 69 kPa.
Lots of things influencing things which I have not completly figured out. My personal take on this is that due to the lower timing and leaner mixture the exhaust pulse speed was higher, causing more vacuum during overlap that draws in more air. When enrichening idle mixture the speed went down again. My .02$.

All of this leads me to believe that you would want to know where you PPP is. But that is a whole other issue altogether. Maybe this is out of the realm of this thread but I was thinking of putting a strain gauge on the metal jacket of a spark plug. Since the combustion pressure will force the porcelain out of the jacket you should be able to measure the strain on the jacket and relate that to pressure. Only issue is that you need to be able to correlate the measured pressure with crank degrees and that is a whole other ballgame since we're talking about a shipload of data even if you only measure from 90° BTDC to 90° ATDC.