How to determine limit on advance timing...............

[toysrus]

Yep, that's right, how do I know how far to advance it ? Does it affect AFRs' or is there another way of determining the exact amount before knock/detonation/pinging ? and also to determine exact amount of timing for highest power output ? without using a dyno ofcourse and I've got my own wideband 02 sensor with rpm input and logging :P

Plus a Microtech LTx8s so I can just plug in my own fiqures :D but I prefer to use some common sense and try and do it in a educated fashion

Thanks in advance

[flappa]

If it blows up you've gone to far :)

Seriously, if you were going to do it in an educated fashion you would be using a dyno.

[David]

Just stop to think a little bit. Every tuning shop in the world uses a dyno to achieve peak performance. If they could do it without a dyno, they would.

The best you can do without a dyno is to have someone drive (preferably up a steep, long, regular hill) and tune it via a laptop or hand controller.

That method *does* work, but it's not accurate, and you risk damaging your engine. You also won't come close to peak performance.

By the way, it has nothing to do with common sense. Common sense means not putting water in the oil filler hole. Tuning an engine is engineering. You'll need to understand what you are doing and why you are doing it.

Read some books and learn about how to tune engines.

Good luck :)

If you try hard enough, you'll succeed, but it's not easy... Just take your time and be prepared to make a lot of mistakes along the way :)

[Powerslave]

Watch this space , I'm in the process of designing a circuit that connects to the knock sensor to provide a visual warning of detonation .... :beer

[mrb1]

toysrus,
you can connect an oscilliscope (a Fluke scopemeter works well!) to the factory knock sensor. But I guess not everybody has one of those :?

Go for it Powerslave I was going to do that a long time ago. Also I think there was a circuit in Silicon Chip a while back as I recall.

[toysrus]

I would then have to connect the Knock sensor up back.......but yes I will probably do that instead of using the gtech hp reading down the same street over and over again :)

Powerslave, if you need any help, or can pass on all the info you so far have done, I'll be more than happy to help out.....I am doing EEE degree at Adel Uni so I know what I'm doing and 'cause I need it sooner I'll probably advance your project fairly quickly......hopefully by the end of next week I'll have it ready :P

[mrb1]

I am doing EEE degree at Adel Uni so I know what I'm doing

You would never hear that statement from an experienced engineer :D Toysrus the more you know the more you will realize you don't know! Once you have got your head around that concept you will be a good engineer.
Don't get me wrong I wish you well in your studies, good luck :)

And yes do a circuit!

[toysrus]

There is so much you don't learn at Uni, even when your in a job, it only teaches you a certain spectrum of everything electronic so yeah there is alot I don't know and hopefully don't have to know :D Otherwise I wont be able to fit any Starion ideas into my head

[Fuel]

Watch this space , I'm in the process of designing a circuit that connects to the knock sensor to provide a visual warning of detonation .... :beer

Can i place my order now? :)

[flappa]

Earlier this year I had a really interesting conversation with one of the engineers from motec about knock detection... I'm not going to say too much because half of it was over my head, but the short of it was in his view the problem with knock detection is not so much the issue of actually detecting knock, it's determining HOW MUCH knock and when that knock is occuring.

In my understanding of it almost every engine will have different accustic properties, and depending on the number and placement of sensors makes it a pretty problematic thing to try and quantify. Some Euro F3 teams running mega dollar Bosch systems are mounting their sensors to the back of individual cylinder walls in an effort to try and get a more accurate detection. All interesting stuff, over my head tho :roll:

[mrb1]

In my understanding of it almost every engine will have different accustic properties, and depending on the number and placement of sensors makes it a pretty problematic thing to try and quantify.

It is affected by the block design/stiffness, bore size, and sensor placement. Unfortunately the only way to fully characterise the signal it is to make the engine knock on each cylinder and see what the "knock signature" looks like coming out of the sensor.

All this relies on making the engine knock to do measurements which might be fine for Mitsubishi but not for me if I have to rebuild the engine!

Anyhow I am sure a reasonably simple circuit can be designed to detect bad knock. It would be interesting to reverse engineer the ignitor box because that is where the knock detection is done. The "the engine is knocking" signal must be in there somewhere. The JB/JD one is sealed but the JA ignitor can be opened I think.
Maybe have a look in there guys. At least to see how the raw signal is conditioned on the input.

[toysrus]

Okay, now the issue is that I will be doing this to the E-101 Ignitor AKA Cordia so it might be a bit different to the Starion one....

What is knock actually ? isn't the basic theory that knock = bad even if minimal for long term reliability ? Also the detenation sensor itslef, does this just output a continuous signal and let's the Ignitor determine if knock is present or does it ouput a signal as soon as it detects knock ? It wouldn't be too difficult to 'reverse engineer' it and see how it accounts for the detonation sensor output, just a bit time consuming you know :( Which leads me to believe the 'detonation sensor' itself outputs a continuous signal.....

If someone has a spare Starion ignitor box which they can take high-res close up pics or let me borrow for a few days, I'll do both at the same time (ie Starion and Cordi) unless it's easier and more accurate to create a new board to connect directly to the detonation sensor and output ?relevant info.....

[David]

You know, there are a lot of circuits for detecting knock already commercially available as both kit and OEM.

I've heard of some really cheap ones.

However from reading about them (this is going back a few years) the common theme is that you usually hear the knock a few fractions of a second before the light comes on - ie, they are not nearly as good as the human ear at picking up the problem.

The general operation is a piezo-electic microphone attached to the engine block detects the characteristing ring caused by knock. I think the simpler circuits actually just use a bandpass filter at this frequency. I thought it was supposed to be around 18 KHz, but a site I found which discusses a bit
http://www.wcnet.org/~bjohnson/knocksensor.htm
that talks about a huge range.

I think the Starion circuit is just a bandpass filter, and it exists not so much to alert you to it, as to force the timing to retard so that you don't damage the engine. It's not that complex. I pulled one apart once, and while I didn't reverse engineer it, it did just look like bandpass filters implemented with op-amps. It has a bit of other stuff too.

The knock itself is caused by pre-ignition, and the resultant shockwave of the two flame-fronts colliding does a lot of damage. Think about the effect of two waves slamming into each other at the beach, as opposed to just butting up against the wall.

This tends to cause a characteristing ring which can be detected by the microphone (sensor).

I've always thought that with an accurate microphone attached to each cylinder you could pick up the ignition with a microprocessor and work off the actual ignition event itself to detect it more accurately... I've never bothered trying, but it would let you retard a single cylinder that was having problems... Might be just enough to win you a race.

Good luck with your circuit design. Don't forget to research on the net too. A lot of information is already there.

Regards
David

[toysrus]

Okay apparently now, from more research, detecting knock by attaching a microphone to a clip to then the block will allow you to detect it before any sensor could possibly do so.

Now the issue is to insulate the microphone from the outside noise to only concentrate on the knock noise passing through the clip or does the 'knock noise occur around the block and not through it's walls ?

[David]

That's what the knock sensor is - a microphone... A Piezo microphone, but nonetheless a standard high-audio frequency mechanical transducer.


You just amply the signal and process it.

You could look for it through other tests, but the piezo is probably the simplest and you already should have one bolted to the side of the engine. It shouldn't take too much to split the signal off.

It's usually located under the inlet manifold.

Regards
David.