View Full Version : Article About Engine Knocking and Knock Sensors

09-09-2004, 05:17 PM
Nice little explanation pulled from the Tec3 Manual: (just ignore the Tec3 specific stuff and you can learn some stuff that still applies to all engines/computers)

To compensate for poor fuel quality and other adverse
engine operating conditions, the TEC3 has the ability to retard
timing based on detonation. To perform this task, it uses a knock
sensor to sense when the engine is experiencing spark knock.
Tuning of the knock sensor is described in Section B.9. of this

The knock sensor circuitry in the TEC3 is designed to
sense detonation, not pre-ignition. Pre-ignition occurs when the
air/fuel mixture in the cylinder is ignited before the spark plug fires. This is generally the result of a hot spot in the cylinder. When pre-ignition occurs, peak cylinder pressure occurs after the piston has reached top dead center (TDC) of its compression stroke, but too early to produce optimum power. Optimum engine power generally occurs when the peak cylinder pressure is
between 10 and 15 degrees after TDC compression. Thus, pre-ignition causes the cylinder pressure to peak before the ideal 10-15 degrees after TDC compression, and in extreme cases, the peak cylinder pressure may occur before TDC.
Detonation is often referred to as “pinging” or “spark knock.” On many engines, the human ear can easily hear the sound made by detonation, since it occurs in the audible range (typically around 5000 cycles/sec). The knock sensor can hear the onset of pre-ignition as well, but better than our ears.

Detonation is defined as a pressure shock wave that develops in a cylinder as a result of a slow-burning flame front. Typically, detonation is caused by ignition timing that is too advanced, poor fuel quality (low octane), or poor combustion chamber design. Engines with large combustion chambers and poor air-swirl characteristics are especially prone to detonation. During detonation, the spark plug initiates the flame front, but the flame front moves so slowly
that there is time to compress and heat the unburned mixture to the point of spontaneous ignition. This results in a pressure shock wave that is akin to beating the upward-moving cylinder with a downward-moving hammer! It also results in a noise resonance through the engine block, just as a noise would resonate through the block if you hit the piston with a hammer. Piston, ring, and rod bearing damage is typical when an engine is detonating too much.

To obtain an accurate reading, the knock sensor must be screwed into the engine block. A plugged hole in a coolant passage is an ideal location, since the knock sensor is threaded for a 1/4" NPT hole. The knock sensor is essentially a microphone that is designed to be very sensitive to the frequency of detonation.

When the TEC3 senses detonation from the knock sensor, it will begin to retard the timing on all subsequent spark events. Once knock has dropped a sufficient amount, the timing will then begin to increase to the desired advance setting. When the TEC3 starts to increase the advance, it will not add any more timing than what is set in the Ignition Advance table. Since engines are more likely to experience detonation at lower RPM’s with medium or high loads, it is advisable to turn the knock sensor on for these conditions. On the other hand, since engines are less likely to experience detonation at higher RPM’s (since the flame front is moving too fast for detonation to occur), AND most engines will increase their mechanical noises that can false-trigger the knock sensor (like valves hitting their seats), it is advisable to turn the knock control off at high RPM’s. Mechanical noise can be identified by datalogging a light-load, high-RPM driving situation. If no detonation is heard during the drive, but the datalog indicates that there was knock in the upper rpm/light load area, mechanical noise may be false triggering the knock sensor.

Rotary engines are generally unable to use the knock parameter as effectively as a piston engine, since real detonation in a rotary will typically only occur one time…then the apex seals will be broken! It is still worth using the knock sensor on a rotary, but do not place all of your faith in it. Just keep in mind that ignition timing should always be VERY conservative with a rotary engine.

09-09-2004, 07:56 PM
Good article Hugh, useful information.

So according to this detonation is not likely to happen in the higher RPM range. WHat about the high load high RPM quadrant of the timing map? When we tuned my car and took it out for a rip we noticed high knock, it got up to 167. Since I have a PFC and no datalogit it was impossible to determine where in the RPM range the knock occured. We retarded the timing 4 degrees in the high load portion of the map across the RPM range.

Can I inferr from this article that I can advance the timing in the higher RPM load quadrant because detonation is not as likely to occur due to a faster traveling flame front? Just thinking 4 degrees across the RPM band was a little extreme in light of this information.

09-09-2004, 08:01 PM
I advance my timing after the torque peak. It is less likely to knock, and it holds power a little longer. I only go up one degree per 500 rpm though.

Knock is most likely in the midrange, right when the turbo spools or when its spooling up. The SR makes very audible knocking unless your car is extremely loud. (ie: open wastegate)

On the PowerFC without datalogit, first question I have is why don't you have datalogit yet? It rocks!

To monitor knock and figure out which RPM its happening at, you must bring a friend and have him pay close attention to the commander, not the road. I've done this many times myself riding with PowerFC owners.

09-09-2004, 08:08 PM
Yea I know it rocks, the reason I don't have it is cause I'm having money problems... it's first on my list as soon as I have a job again.

I'll try your method of increasing by one degree after the peak and bring a friend along to pay attention to the commander.

09-10-2004, 09:57 AM
As usual another great post Hugh. Looks like I need to download the Tec3 pdf file... http://forums.freshalloy.com/images/graemlins/grin.gif

09-10-2004, 10:13 AM
You could learn a lot from it

09-10-2004, 10:56 PM
Interesting article; made me think. When I first read it, I thought that it had the goofiest explanation of the underlying cause of detonation I'd ever seen: the flame-front moving too slowly?? But this is from Electromotive, who surely know a thing or two more than me, so I did a little back-tracking thru sources I trust... Hmm, while not explicitly mentioned, slow flame-fronts don't contradict them.

What I don't understand, but would like to, is the bit about the flame-front moving faster at higher rpm. Why would it do this?

I've always been under the impression that it moved at the same speed at any rpm, which is why ignition timing advances as rpm increases (so that peak cylinder pressure will occur at the optimum degrees ATDC). At the same time, using this belief as a basis, I've also always been vaquely uncomfortable with the way timing maps generally level off (somewhat) at a certain rpm. Sometimes I wonder if the advance curves are just relics left over from the days of mechanical/vacuum advance distributors, but I really doubt that. So why does the timing level off?

To recap:
1) Why does the flame-front move faster at higher rpms; and
2) Is this somehow relevant to the shape of the typical ignition advance curve?

Somebody educate me. Please.