High RPM knock detection and knock theory

[Schulmann]

I read a few articles about knock problems in regular and supercharged cars. These articles are very interesting for ignition timing. When tuning the ignition, one of the problems is to detect knocking before it harms the engine. At low revs and at high load it is fairly easy to hear the ping in the engine (Example: Running 50mph in 6th speed and pushing the gas pedal).

However a lot of articles warn about high rpm knocking. All articles agree that it is nearly impossible to detect these knocks unless they are really severe. Most of devices which are available to detect knocking are not operating properly above certain RPM due to the engine’s vibration. For human ear it is also very difficult to detect these pings. So there are preventive measures to avoid these knocks. Running rich, retarding timing, using high octane fuel, water injection etc.

But when it comes to really fine tune a car how can we ensure that we will run with the optimal spark advance without those undetectable knocks?

How can I know that I gave too much spark advance?

Is this theory right below: ?
If knocking doesn’t occur at low revs and high load then running rich at high revs with “low” load will like prevent knocking from occurring.

 

[Russ M]

Best way to make sure your car isnt knocking is to dyno tune.

A good tuner will see knock on the dyno chart/spark plugs/sound.

Modern knock sensors are much more capable than past units, they can and will detect knock even at high rpm.

That theory does not hold water, it assumes to much in respect to timing/fuel quality/cylinder presure.

 

[cgmaster]

I would get a knock sensor set up on an ECU that could control everything such as Motec or Accel if I had any doubts.

 

[Irid]

If you're really trick you get ahold of pressure-sensing spark plugs (and accompanying hardware logger...). They are expensive (thousands of dollars) but that's an "OEM" way to moniter things correctly while tuning. You can also bore small holes in the head and use similar sensors.

This isn't something you'd run around with on the street, but rather something tuners could use while testing their systems to create various packages.

http://www.optrand.com/fliers/psiplug_%20060201.pdf
http://www.intertechnology.com/Kistler/Pressure_Model_6115A.htm

 

[Dyno Dave]

Hearing is not an effective way to measure knock. Especially if you've been operating a dyno for many years. Yes, I say "Huh?" alot.

The OEM's use cylinder pressure transducers, such as those made by Kistler. A passage is machined for insertion of the transducer from the outside of the head or block to the combustion chamber. They cost about $1500 each and require charge amps and an o-scope to read.

A less intrusive method is a pressure transducer equipped spark plug. They cost $2500 each and I don't find them to be as accurate because of their offset electrode, limited heat range availability and they have relatively poor durability.

Another method is ionization feedback across the spark plug gap. This requires an expensive ignition system, software, a sophisticated crank encoder and sometimes special spark plugs.
See (http://www.woodward.com/engine/smartfire.cfm).
I've used this system and it is very effective in measuring knock and misfire. Durability can be a problem because it requires the use of special spark plug mounted coils. The close proximity of the exhaust to the plug on a Viper is an especially tough environment for a coil.

Attention also needs to be paid on the sensitivity of the engine to knock with high coolant temps and elevated ambients. Humidity also plays into the equation. A humid summer day is like having water injection.

There was a comment on letting an ecu detect it. The first thing that has to be done is determine the engine background noise level without knock and set a threshold level above that which will be knock. Then put some low octane fuel in the engine and knock it on purpose and see if the sensor can distinguish the difference. As the engine runs up in speed it may go through some harmonic ranges where the background noise level will increase. This will need to be filtered but not inhibit the ability of the knock sensor to "hear" true knock. Placement of the sensor is critical. It must be able to hear knock from all cylinders equally (rarely the case).

So to complete my post on knock I will just say, it isn't as simple as I've read it to be on this board. I know. I've done it.

 

[Tom F&L GoR]

If anyone goes that far, get the ones with surface mounted pick ups and not tubes that run up the sides of the spark plugs. Anything above idle will create pressure waves in the tubes that make the signal useless. And yes, OEMs do a lot to reduce idle instability...

NASCAR types use individual cylinder pressure transducers to tune so that each cylinder is producing max power. They try to adjust for the different runner lengths, g-forces in turns, and used to grind irregular cams for point distributors, then irregular stars for electronic pickups to give each cylinder it's best timing.

Taking the intro from those NASCAR geniuses, good mixture preparation, swirl or tumble rates, and appropriate end-chamber designs are all (supposed to be) part of the package. Slapping parts together and adjusting the timing to stay out of knock isn't really tuning, it's finding the right size bandaid.

Other than restrictor plate engines, I thought NASCAR engine builders had more than enough octane because they had the burn rates so high. Part of the knock phenomena is having enough time for the compressed end gas to heat and self-ignite; if you were quick about it, it wouldn't have enough time to have the needed chain reaction.

Two stroke engines had spark plug metal gaskets that served as thermocouples. It's a little messy to run 10 wires, but you could look at plug seat temperatures and look for the "outlier."

Anyway, cool (pun?) stuff.

 

[Schulmann]

You are all right and this is what I feared. This type of tunig requires rocket science equipment. I could easier get an oscilloscope and a transducer to measure the high frequency vibrations ...

 

[FE 065]

The average person on a reasonable budget can't go redesigning OEM intake ports and valve angle for better swirl pattern in the cylinder, or combustion chambers and the new pistons to go with them. So knock sensors are a bit more than a bandaid in realistic terms...


I read an article recently mentioning Japanese tests that showed the higher the RPM the less probability for knock. Because it's all happening faster. I think it said that at 18,000rpm they could get away with something like 60 octane fuel.


Which doesn't help the thread subject does it...?

 

[Tom F&L GoR]

The average person on a reasonable budget can't go redesigning OEM intake ports and valve angle for better swirl pattern in the cylinder, or combustion chambers and the new pistons to go with them. So knock sensors are a bit more than a bandaid in realistic terms...

You don't have to redesign; there are just other parts of the package to consider. Smaller (or just OEM size) vs. larger valves affect charge velocity through the curtain area to mix fuel and air; cam profile and centerline affect scavenging, backpressure affects exhaust gas retention, compression ratio affects quenching and so on. Funny how we can say knock sensors are good, but how many of us would purposely retard the ignition timing to avoid higher power-producing cylinder pressures? (which is the same thing.) I didn't mean to be argumentative, just to point out that sometimes the best package feels counterintuitive, like putting on a smaller carburetor to make an engine run better.

 

[Dyno Dave]

Quote:"I read an article recently mentioning Japanese tests that showed the higher the RPM the less probability for knock. Because it's all happening faster. I think it said that at 18,000rpm they could get away with something like 60 octane fuel."

This is true. I have some experience with the 2.0L Touring Car engines of a few years ago and those engines had 14:1 compression ratio, made over 300HP with a rules regulated rev limit of 8500rpm and ran on 93 octane pump fuel.

A Viper would make 1200HP normally aspirated if you could get the same specific output/liter.

 

[FE 065]

The average person on a reasonable budget can't go redesigning OEM intake ports and valve angle for better swirl pattern in the cylinder, or combustion chambers and the new pistons to go with them. So knock sensors are a bit more than a bandaid in realistic terms...

You don't have to redesign; there are just other parts of the package to consider. Smaller (or just OEM size) vs. larger valves affect charge velocity through the curtain area to mix fuel and air; cam profile and centerline affect scavenging, backpressure affects exhaust gas retention, compression ratio affects quenching and so on. Funny how we can say knock sensors are good, but how many of us would purposely retard the ignition timing to avoid higher power-producing cylinder pressures? (which is the same thing.) I didn't mean to be argumentative, just to point out that sometimes the best package feels counterintuitive, like putting on a smaller carburetor to make an engine run better.

You must be registered for see images

OIC I thought you were speaking of the good mixture preparation, swirl or tumble rates, and appropriate end-chamber designs... part.



Well for a SBC Chevy there might be loads of ready made parts available. I think with the Viper engine you're kinda stuck with what DC gave us short of special ordering one-off high $ stuff



Unable to pre-calibrate timing for every circumstance in a manic +800hp Viper engine, accurate knock sensor timing control would seem to be as valuable and valid a choice as traction control for us mortals.


There ways to get better traction if you're willing to tub the car and redo the rear suspension and axle. Or just add traction control.


It's not the higher power producing cylinder pressure in and of itself that's causing the detonation anyway is it? The same cylinder pressure in a smaller cylinder bore, or being lit by dual spark plugs in the large cylinder bore wouldn't be as prone to detonation.



I'm with ya Tom! I just think an accurate knock sensor is the shortest and most cost effective route for anyone that doesn't have the budget of a factory team. And I'll bet teams with turbocharged engines are using knock sensors as well.