Determining the natural frequency of an engine block? (for knock detection)

[RL7]

From the Stealth Technical Information Manual:

"Detonation Sensor
The detonation sensor detects cylinder block
vibration due to knocking and outputs voltage
proportional to its strength.
The number of cylinder block vibrations due to
knocking vary for each engine and in the
detonation sensor, there are a vibrational plate
with the same number of vibrations (natural
frequency) as the cylinder block and a
piezoelectric element to generate voltage when
the vibration plate vibrates. When knocking
occurs, the vibration plate resonates with the
cylinder block vibration and the detonation
sensor outputs a high voltage. When knocking
ceases, the detonation sensor output voltage
falls because the vibration plate does not
resonate. The engine control unit uses this signal
to retard the ignition timing according to the
strength of the knocking."

!?!?

When they say "each engine," do they mean one engine model to another or one partcular 6g72 (for example) to another? I doubt the factory determines the natural frequency of each block and fits the "correct" knock sensor, but that is almost how this makes it sound.

I bring this up because I was considering using a knock sensor from my stealth block to monitor knock on a 300zx. Apparently the knock sensor on a z32 is not very good at all and doesn't even function above 3000 rpms. Obviously, that is unacceptable and it partially explains why z32 people are so negative about putting turbos on the NA 10.5:1 block. I like the way the 3/s knock sensor works in general and I was going to use an e-manage or something similar to log it, but after reading the above, I'm less sure.

I believe changing certain things on an engine changes it's natural frequency, so wouldn't certain upgrades interfere with the operation of the detonation sensor? And wouldn't a 6g72 sensor not work correctly on a 74 or 75 block if this was the case? Is there some way to measure the natural frequency of YOUR motor and build a knock sensor for it? I'm thinking mabye hit the block with a hammer and record it with some sort of software to determine the frequency? I don't currently have any idea how to build the corresponding knock sensor, but it sounds like a fairly simple piece of electronics. Thoughts?

[green-lantern]

I don't think that will work very well. That sensor it designed for the 6g72 and that plate. I would think just a different plate would cause undesirable results. It could show knock all the time or not show knock when you are actually getting it.

[x2percentmilk]

Most kock sensors are all the same, just a piezoelectric sensor, its the preprocessing circuit (usually found in the ecu) that might "tune it" and filter out unwanted noise, and how the ECU interprets the signal, that change. Of course, quality can differ from sensor to sensor, but alot are made by bosch or some other major manufacturer.

[J. Fast]

I think you could probably use the sensor since it's really only designed to pickup noise off the quench pad. The filter parameters are a different story. If it's an expensive engine I'd just buy an aftermarket knock sensor kit like a Plex or a Knock Link. They come with the sensors and they dumb down everything and they're only a few hundred bucks.

[RL7]

-cut bad info-

I looked at the aftermarket options and they seem a little bulky. I haven't looked too hard though. I really just want to be able to reliably log a measure of actual knock and possibly have some sort of gauge or light associated with it.

Turns out the hitting with a hammer test is actually used. It's called a bump or rap test and basically entails hitting the object with a hammer to make it resonate and measuring the frequencies with equipment with that capability.

[i3igpete]

not quite true. combustion knock frequency depends purely on the cylinder volume when the knock occurs, and the mix in the cylinder. it's frequency is basically the mixture's speed of sound bouncing around the cylinder walls. This type of analysis requires CFD/wave dynamics and can't be easily recreated except with a running engine that is tuned on purpose to knock. the frequency is high enough to be barely audible to the human ear.
http://www.youtube.com/watch?v=GGJZ6kKzLus

complete engines vibration is of a much lower frequency and analyzed with completely different software (FEA/sysnoise). This is a mechanical noise that can be experimentally recreated (the hammer test you mention). This sounds like the familiar thud noise.

[KeithMac]

I would have thought the Z`s knock sensor will pick up genuine knock over 3000 rpm but the ecu just ignores it (because they were too lazy to program the filters).

[J. Fast]

If you hooked up an amp to a spare knock sensor attached to your block, you wouldn't need to bother with internet BS. There's several bands of frequency, you can only hear at best the lower ones (bore diameter). The most important ones are out of audible range, the frequencies are determined by the distance between the piston and the roof of the combustion chamber.

Steve

Are you talking about that microphone knock device frankensteined together on that 1st gen DSM awhile back? I remember reading about that on RMDSM and CODSM and there was a little more to it than just hooking up a mic. They guy actually used a spectrometer and a soundamplifier with a filter and made recordings and then analyzed them on a computer and compared them in real time with his logger. It worked but was a big ass time waster.

You do realize sound waves bounce all over the cylinder chamber like a raquetball or crazy pong game but they will not get picked up until they hit the quench pad? Do you know what a quench pad is and where that region lies in an engine and why it was designed? It was designed strictly for one purpose, when you compress a mixture in a cylinder it's the only region where harmonic waves from detonation report and generate a spectrum wave that can be deciphered with a knock sensor. It's also the only region in an engine where you will pickup soundwaves from fuel splashpoints, and more recently a region that has been reshaped to help with fuel scavenging and swirl by cross-hatching.

[J. Fast]

Steve, you’re kinda right, but in this case everyone is right. Pete in one hand is describing a Resonance Sensor, and you are describing a Mass Sensor, and x2percentmilk mentioned most sensors are piezo electric sensors and that’s correct as well. To get everyone on the same page let’s make sure to note there’s several types of knock sensors.

Most of us are familiar with a basic Piezo Electric Knock Sensor which is the one you generalized in the thread. A basic Piezo Knock Sensor produces voltage based on noise or when it senses vibration. To test for proper operation you grab a voltmeter and set it to AC Volts and tap into the sensor lead and then on the negative side ground to the block and bang the block with a hammer. If it produces voltage that means it’s working. Hammer type freq’s is all it’s responsible for reporting… such as loud banging, piston slap, rod knock, and generally any audible to ear engine noise.

The second type of knock sensor is a Mass Piezo Electic Knock Sensor. It’s basically posesses all the characteristics of the basic Piezo sensor except it was modified to respond to much broader frequency range and it accepts all sound frequencies. The benefit is that it works and can be tested with a hammer test. The bad is that I don’t want to have my stock ecu pull timing when it hears noise from a ticking lifter or a timing belt or useless engine noise which is why we move into the sensor that Pete pointed to.

The last type of knock sensor is a Resonance Piezo Sensor. A resonance sensor is tuned specifically for a frequency range that’s been identified as specific knock. The advantage to this type of sensor is its sampling rate. A standard Piezo sensor sampling rate is whatever it hears (which is slow btw). The sampling rate of a Resonance sensor is something crazy like 5000 times per minute and it’s only listening for a specific frequency programmed into the specific knock filter. Consequently a resonance sensor will not respond to a hammer test unless by some strange twist of fate you can strum the exact frequency range with a hammer and extension by beating the piss out of the engine before your arm tires. The only way to get it to respond is to actually make the engine ping or knock by manipulating the air, fuel, or spark delivery and monitoring the sensor voltage with a logging device.

In a nut shell that's the 101 on types of Piezo Knock Sensors.

[MaxClass]

I also thought of the hammer test as I began to read your post. Stradivarius used the same type of test on trees to find the one with the best sound for making his violins out of.

Maybe the ultimate sensors would be replacements for spark plug gaskets. That way each cylinder could have its own detector directly at the source and know exactly what it is looking for and not pick up extraneous unrelated noise like lifter tick’s, etc.

Just my two cents.