Warren, as Grady said above these specialized microphones are tuned to a specific response range. Have you seen this very simple circuit to drive an LED indicator?:

http://www.gmcmhphotos.com/gallery/showpic.php?aid=59&uuid=mrerf&pid=514

That link above was posted in this thread:

http://forums.pelicanparts.com/showthread.php?s=&threadid=242234&perpage=40&highl ight=knock&pagenumber=2

I always assumed some actual signal processing would be needed but apparently not with the tuned knock sensors that are used. for a 911 we could use the 964 part number. The problem, as I mentioned above, is that we don't have a good way to bolt the sensor on (do we?) without the bosses that the 964 has for the knock sensor bridge to bolt onto (see Frere's 911 story section on 3.6 engines)

the pertinent part from John Cramers link:

How Damaging Is Detonation?

There are newly proposed "standards" that define "light," "medium," and "heavy" detonation. How those are arrived at is far too complex to go into here (which means "I don't know"), but suffice it to say that a little light detonation, even for hours at a time may not be harmful, and in fact, can be beneficial. It does a marvelous job of cleaning deposits off the top of pistons, for example!

The truth of the matter is, most of these engines can operate in the light detonation condition as shown in the graphics for several hundred hours with no detectable damage, . . .....

Now, am I recommending detonation? Definitely not! But at the same time, it is not quite the fearsome monster we've all been led to believe.

I'll keep this mind while reading.

Thanks.

dd74,

I didn't mean to imply that I was using an electronic detector back in '74 ... I was just experimenting with the advance setting on my '010' distributor to find the best setting that delivered/allowed use of full throttle without detonation ...

Just mixed a bit of 'thinking out loud' and reminiscences from the distant past ... suggesting what could be done, and apparently what has already been done, albeit on a motor home! The megasquirt guys aren't really breaking any new ground on this front, just collecting old data, and trying to cobble their projects together! Mixed in with beep's clamoring that 'acoustic is obsolete!'

Calling the piezo knock sensors 'tuned' is a bit misleading, though, as it is really just a matter of picking the dimensions of the piezo element, and living with the signals that are generated! Bigger piezo element = lower frequency sensitivity, smaller element = higher frequency sensitivity ... it is still a glorified microphone!

The Toyota manual page is really crude compared to Bosch publications' diagrams ... at least Bosch uses real waveforms, not an artist's drawing! Bosch superimposes cylinder pressure and piezo sensor waveform on the same scope trace and using the same time base setting!

Considering what is done with Bosch sensors on liquid-cooled engines ... mounted in pairs on the side of the block ... the sensor may not be required to be on the head at all! An easier retrofit may be to clamp a sensor to the top of a cylinder, or a 'bridge' connecting three cylinders on each side ...

The above of course is old anolog filtering technology but it does have it's place in the digital world. It saves bandwidth and focuses the computer on the job at hand, but here's something to keep in mind, voice recognition is way more complex and if it works then knock sensors should be a cake walk.

In otherwords, I would think the real solution in knock sensing would be teaching the computer what detonantion sounds like i.e. the "signature" Warren refers to. So, why not induce a motor to detonate, sample it, and tell the computer this detonation. When you "hear it" retard the ignition. This is pattern recognition and is more discriminating than just frequency triggering. Ah yes, the spin doctors. "It's tuned." :rolleyes:
I guess "only efficient in a very small bandwidth" doesn't sell as well. :)

This is an excellent discussion, all tech viewpoints need to be considered and even argued. But lets keep it personal, I don't like censorship but I don't like personal assaults even more.

If you have something to add, do so, otherwise follow my Dear departed Mom's advise "If you don't have something nice to say, keep quiet."

Here's what the Bosch engineers have to say on the subject

Ignition point (firing point) and ignition timing

Approximately two milliseconds elapse between the instant when the mixture ignites and complete combustion. Provided that the mixture strength does not vary, this period will remain constant. The ignition spark must thus arc early enough to provide optimal combustion pressure under all engine operating conditions.

The standard practice is to describe the ignition timing relative to 'Top Dead Center," or TDC at the crankshaft. The ignition timing is then specified in "degrees before TDC" or as the "ignition advance angle." Adjusting the ignition timing toward TDC is called retarding the ignition (or reducing the advance), while displacement in the opposite direction is referred to as advancing the ignition.

The ignition timing should be selected to satisfy the following criteria:

- maximum engine output,
- modest fuel consumption,
- avoidance of engine knock, and - "clean" exhaust gas.

Unfortunately, these demands are mutually antagonistic; compromises have to be made on a case by case basis. The most appropriate ignition timing for any individual condition will depend upon a number of factors, the most important being engine speed, engine load, engine design, the fuel, and the specific operating state (for instance, starting, idle, full throttle, trailing throttle). The basic adjustments necessary to adapt the ignition timing to the momentary engine conditions are performed by engine-speed and load-sensitive devices.

The high compression ratios favored in today's engines considerably increase the danger of knocking beyond that encountered in earlier designs.

Engine knock is produced by spontaneous combustion in portions of the mixture that the flame front emanating from the ignition spark fails to reach during the initial flame-propagation phase. This condition corresponds to excessive ignition advance. Combustion knock leads to higher combustion-chamber temperatures, which can cause preignition from hot spots. It also results in radical pressure increases. The sudden combustion generates pressure fluctuations; these are superimposed on the normal pressure pattern (Figure 3). A distinction is drawn between two kinds of "knocking":

- Acceleration knock under high load factors at low engine speeds (audible knock or "pinging"), and
- High-speed knock as encountered at high rpm during operation at high load factors.

High-speed knock is especially critical for the engine. It remains inaudible due to the general level of engine noise. Thus audible knock is not a completely reliable source of information on engine knock; however, electronic means are available to provide precise monitoring of the phenomenon. Continued knock causes substantial damage to the engine (destruction of the cylinder-head gasket, bearing damage, holes in the pistons) as well as to the spark plugs.

The preignition tendency depends upon such factors as engine design (for instance, combustion-chamber layout, homogeneous air-fuel mixture formation, free-flowing induction paths) and upon the fuel itself.
http://forums.pelicanparts.com/uploa...1131805767.jpg

Thanks Bill.

Let me explore what I think the failure modes are as a result of detonation. This is mostly my speculation from experience and not from professional journals.

The most obvious is the hole in the piston that I posted above. The hole looks as if you took a ball-peen hammer and “knocked” a piece out (pun intended). Note the large tapered area on the bottom side of the piston. Interestingly there is a small tapered area on the top also. The engine must have stopped immediately after this event because there are no visible combustion products on the broken surfaces. Remember, this a 2.0 with the tall combustion chamber.

This piston and almost all others that I can say exhibited detonation have signs of severe overheating. This results in the piston seizing in the cylinder and badly scuffing the sides of the piston.

On current larger diameter pistons the detonation tends to bend the peripheral area of the piston and close up the top ring land. This grabs the ring in a fixed position. The eventual result seems to be that the ring brakes into smaller and smaller pieces until pieces start to get past the top of the piston and into the combustion chamber.

A more insidious result is rod bearing failure. I think the shock wave from the detonation travels down the length of the rod and damages the big end bearing. The rod itself is strong enough to tolerate the shock. When the shock energy gets to the steel-copper-babbitt of the bearing shell the energy is dissipated at the interface of these softer materials. I think this causes the copper to separate from the steel shell and flake off in the classic rod bearing failure that we see. I think many rod bearing failures are not from lack of lubrication but are from prior detonation.

The very few broken wrist pins I have seen I have ascribed to a possible hydraulic lock. That is not to say it is impossible for detonation to have played a roll in the failure.

High compression, high temperatures, single plug to the side, poor swirl mixing, inconsistent octane and poor tuning all seem to contribute to detonation. The question comes back to what can be done when specifying and building an engine to reduce the propensity for detonation? The next question is what can be done in the tuning and operation? Finally, with the current electronic technology, what can be done to further reduce the detonation or mitigate its severity?


Best,
Grady

OK Bill, you deleted my last post, no problem. I respectfully request that you go back and delete Loren's personal attacks as well.

And, have you seen this thread?
http://forums.pelicanparts.com/porsche-911-technical-forum/250686-quick-question-dme-control-module.html

If that doesn't get Loren another time out, the moderators aren't doing their jobs.

Done, now lets stick to the topic , ;)

The signal from the knock sensor represents the pressure in the cylinder. The Bosch knock sensor is a wide-band accelerometer that has a natural frequency of more than 25 kHz. The sensitivity of accelerometers on the high-end of the spectrum does not preclude them from responding to lower rates, too!

Obviously, more than just a 'mic' ... but for purposes of discussion, the analogy is good enough!

The fact that the same Bosch sensor can be used on water-cooled commodity car engines and high-end water-cooled or air-cooled sports car engines speaks volumes about the basic data being gathered ... basic science does not change!

Also, there isn't any digital signal processing being done on the knock sensor signal in a Megasquirt system ... strictly analog data being converted to a yes/no binary signal to trigger the retarded ignition algorithm.

We might consider the extent to which knock is likely to be present in our engines - given the lifetime of the 911 engine it is low in stock form. What we really want - practically - is an idea of where the envelope lies so we can push it, given whatever gas is available...

Active engine mgmt. systems with knock sensors allow this 'gap' to be much closer than would otherwise be the case. If a human is in the loop - blue light or otherwise, then there will be a delay in response. How much damage is likely during that delay.

Detection via "knock sensors" technology is about as dead as the dog being beaten on this thread!!!

Can you spell Ion??

http://forums.pelicanparts.com/uploa...1131831416.gif

Oh really? count the number of OEMs using ion technology vs the number using "knock sensors". Then check out the availability of aftermarket ion technology.

Look, of course it's better, but are you driving a FWD saab?

Good gawd man the technology is still proprietary. If I explained it I'd have to kill you....

In the meantime you guys stay with buggy whips.

The earth is flat and the dog is dead.... try a search...

Muhahahahaha!

thx Bill

I'm with tsuter on this one. Can you spell $ ?

And afaik alcohol in the new smog gasoline increases the tendency to knock.

And a digital and bar graph EGT/CHT gage is excellent for tuning and detecting all kinds of problems. The higher end models allow down-loading. There is really no other method of real time info. Even a dyno with a million bucks worth of monitoring instruments doesn't duplicate real time. I guess a knock sensor is better than nothing, but it's not the best in my non pro opinion.

I've been experimenting with lowering octane from 93 to 91 to 89 currently. I waited for cold weather to do this for safety reasons. So far so good at 89. I can't seem to get even close to detonation & pre-ignition so far. After a couple more tanks I'll be trying 87.

$10-20K for an engine makes a few hundred for a gage seem inexpensive.

Warren,

I’m not sure I understand how the knock sensor signal represents cylinder pressure. I thought it was outside the combustion chamber. What is the correlation? I know there are pressure sensors that can be fitted in the head that directly read instantaneous cylinder pressures (within their spec bandwidth). That subject is worth an entire thread.

I thought accelerometers were axis dependent. Does the direction of mounting affect the sensor output? If so, can the direction help discriminate between normal engine noise and characteristic detonation noise?

What happens to the sensor output as the frequencies approach the natural (resonant) frequency?

What are the Porsche and Bosch P/Ns for the knock sensor? What others are available (US & Japanese)?

Is it reasonable to devise a circuit that gives a quantitative level of engine “noise” and an adjustable threshold for “knock” noise? Sort of like a sound meter with a weighted scale. Has anyone developed a “D-weighted” scale for Detonation weighting?

Are there software solutions using existing hardware?

I would be willing to foot the cost of components. I can build a simple prototype just fine. My EE circuit design skills are 30-40 years out of date and practice.

How would you calibrate a knock sensor? One point could be a 911 engine that is low compression, conservative ignition advance and very high octane fuel. Another point might be a diesel engine under load.



The point of all this is to help solve a significant problem with our nice old (pre electronics) 911s. For not much cost a vendor can supply systems based on knock sensing feedback. We can help drive that. This Forum has a wonderful mix of skills.

Best,
Grady

"The engine must have stopped immediately after this event."
------- the event happens before TDC. It happens as the piston is traveling up. If it happened after TDC the decreasing combustion psi would not allow it to happen afaik. Also the max psi is around 800-1,000psi without detonation. For entertainment there is an aircraft spark plug that is drilled to allow a power measurement.

"This piston and almost all others that I can say exhibited detonation have signs of severe overheating. "
------- Combustion temps are 3,000-4,000F. An Al piston melts around 1,200F. A boundary layer is set up under proper conditions. Detonation and pre-ignition scrub this heat boundary layer away.

"This grabs the ring in a fixed position."
------- the ring is disturbed from its normal position. It gets whacked around in the land. I read a description of the event. If you want a tech explain I'll try to locate it.

"A more insidious result is rod bearing failure."
--------- confirmed.. as per aircraft text.

"Finally, with the current electronic technology, what can be done to further reduce the detonation or mitigate its severity?"
-------- buy a EGT/CHT aircraft gauge with digital readout and a bar graph and tune to avoidance.

The future of EMS is now....

http://forums.pelicanparts.com/uploa...1131836028.jpg

Ron,

I pretty much agree.

The good news is there are several Pelicans who are searching for their MSME thesis. I’ll bet between the bunch of us and some industry help they could find additional funding and guidance. I’m willing, how about you and others? A Pelican graduate ME thesis advisory committee would be great.

For years (even pre WWII) there has been the ability to photograph (even high speed movies) combustion chamber events. Today’s technology carries this even farther. A 911 head is ideally set up for this kind of investigation. The easiest is to drill for the 2nd spark plug and use that port for pressure sensors, IR temperature sensors and video in various bandwidths.

It would be great if you could cite some references. I apologize I haven’t done that (yet).

Best,
Grady