Thanks Mark and Ted for thoughtful replies.

Here is one way to put it... call AEM and ask them, would you recommend tuning a car relying solely on their knock control system? And I'd suspect they would say no.... but it would be curious to see how they would reply.

Me personally, with it being even a little bit controversial how well a knock control system can be, I'd err on the side of caution and use det cans of some sort.

Seems like a nice safety feature though if you have it.

 

This kind of sucks because many people on 4G64's or stroked 2.3 4G63's are having problems with phantom knock, which I remember was a big problem with the DSM's. There has to be a proper way around this problem (without resorting to friggin headphones lol). There are plenty of people out there with extremely high power 4G64's and 2.3's with totally different engine internals. I wonder how their tuners dealt with false knock?

 

Probably the the only means of accurate knock measure is with a gas bench, which is expensive to rent and requires experience using it on the same car for some time. Other methods work, just not qualitativily, most professional racing uses them but it's not quite an option for us yet. Hopefully they'll be available similar to lambda O2 in the future but at this rate we might be forced to drive electric cars by then

 

Update:

I have found three very interesting articles for those who want to know more about detecting preignition using one's ear:

Part I

Part II


And how to construct such a setup:

DIY

I've ordered the parts for three audio setups and will construct them when I receive everything. We'll see how well it works and report.

 

ted be i remeber somewhere that jcsbanks did some testing with detcans and evoscan and he says that at 15knock sums only some slight pinging was heard and the ecu would retard timing very fast that only slight dets would go through...
the stockk ecu is just great for saving your engine...

 

I have tried a couple of these types but did not like them as it wasnt clear enough for me and all you could do was turn the noise up or down so it still picked everything up lifters and all,
I have mapped on the engine dyno with the operator using that style noise meter and I was using my own copper tube and pipe and we both heard dett at the same time, He preferd his own and didnt like mine as he was used to his just like me and mine.
Ted you should also try the copper pipe style also if you havent already.

Mark

 

FYI the headphone setup Mark is referring to above looks like the attached pic below. For those that don't know, Mark is a long time tuner in England running a shop call MADevelopments, one of the largest Autronic resellers in the world.

He has produced many 700 hp GT35R cars... just recently mapped a 1500 hp Skyline that does low 8 seconds using Autronic engine management. Also sells his own 2.2 stroker which is the most expensive kit out there for the Evo!

 

Just to elaborate further on the above headphones..

A copper pipe about 1/2" in internal diameter and 8" long I believe with one end flattened and drilled is bolted to an intake manifold stud. Rubber hose is then pushed into the other end of the copper pipe, which then leads to the headphones pictured above.

 

Using a hard plastic pipe offers better sound but this is hard to put through the bonnet and front panel, However I do use this on the engine dyno due to the longer lenght away from the engine, If you use 8mm fuel pipe for the connection keep it short and make sure the bonnect doesnt close the pipe up when you shut the bonnet but you can hear the change in noise when you close the bonnet.
The copper pipe is around 15mm diameter.

Mark

 

Interesting. What was the originally intended use for those headphones? I ask because they appear to akin to a closed-ear stethoscope, and I don't think I've ever seen them.

 

They are just normal ear defenders with holes drilled in the bottom for the pipes,
Using these that cover your ears almost completely take the car/road/background noise so you can concerntrait on what the engine is doing.

Mark

 

Thanks, I'm going to try them. I'm also going to try the electronic model with a pair of noise-cancelling headphones. I'm curious to see how many knock counts the factory ECU records vs. the point where I can detect it audibly.

 

I would only use copper pipe around 15mm min or max the larger size will pick up lifter noise and injectors ticking, I havent tried smaller pipe thats why I recomend the 15mm as this would be the correct known starting point,
I would bolt the pipe to the middle of the inlet manifold as this is easy and where I put it,
On a V8 I would put a pipe either side of the V under the exhaust manifold if poss and 1 pipe to each side of the headphone if you are not deff in one ear most people have a dominat ear for hearing so it would be best to find out which one or hear with both.

Mark

 

Thanks, the good info is appreciated. I'm very curious at this point, especially if the results from comparing audible vs. factory ECU knock detection are useful enough to use the audible method to program the KS noise floor settings for an aftermarket ECU.

 

Right.....
The knock sensors are tuned to a particular frequency to ignore the other "noise" going on in the block.

This frequency the sensor is tuned to "listen for" depends primarily on the geometry of the combustion chamber. So....changing cams, etc, etc, isn't going to change that frequency when detonation occurs.

As someone else posted...It seems in the worst case, the knock sensor will respond to something it shouldn't (like "falsing"), pulling timing when it doesn't need to (i.e. fail-safe). Seems very unlikely that it won't respond when it's supposed to.

MAYBE changing pistons (i.e. changing the volume or shape of the CC) could cause some shift in the knock frequency so the sensor would require recal, but that's only a speculation.

EDIT : of course, failing safe and pulling timing isn't the solution the go-fast guys are looking for.....sounds like the ion-sensing is the way to go. I was looking at this some more and was surprised to see how widely it's already used. Google ion sensing Delphi, for example.