NineMeister

Whilst I admire Jussi for his work and would like to offer help, I've also tried desperately to stay out of this discussion because I have been in a similar position myself and got tired of people telling me what I could not do, not helping me with things that I knew I could do. Subtle difference, but I'm sure you can understand my viewpoint. Now I'm here, maybe you've just caught me at a moment of weakness Harald?

To answer your points, you're right Harald, 9m do not lose engines due to detonation - provided the customer runs the correct fuel that the engine was mapped for and that nothing goes wrong with his engine or installation. That said, knock control is a great safeguard and given a choice anyone would be a fool not to have it, however for the purposes of producing powerful engines it is definitely not essential.

The way that knock control works (as far as I understand it) is that when an individual cylinder knocks or detonates you will get hot spots in the cylinder which propagate further knock events through consecutive combustion cycles, hence most knock control systems (individual or otherwise) when faced with multiple knocks will retard timing by 5 to 8 degrees in one hit, then the algorithm re-advances the spark to a safer level.

Engines will not make more power with global or individual cylinder knock control. Not even one tenth of a hp. In my view the object of the (tuning) exercise is to set the ignition timing and fuel mixture to optimise the transient response of the engine and hence give the car maximum acceleration, a status that will not be achieved by measuring a notional mean best torque on a dyno to deduce the ignition timing, or indeed by ensuring the mixture is exactly La=0.88 (or whatever else you aim for) at some notional measurement point in the exhaust pipe.

That will have to do for now, my cold is getting the better of me.....

evoderby

Hi Colin,

Please have a look at the power graphs provided by an Ilmor engineer (for those not in the know designers of Mercedes F1/Indycar engines) using the same aftermarket individual knock control that I'm using (albeit not on my Porsche):

http://www.jandssafeguard.com/VampirePage/Vampire.html. (bottom page)

Written testimonial on the same engine:

http://www.jandssafeguard.com/SafeGu...timonials.html. (bottom page Ian Whiteside)

Rgds

Harald

NineMeister

Thanks Harald.

Interesting concept that you can completely over-advance an engine (with distributor or ecu) and just rely on a purple box to stop it blowing itself to pieces.

Personally, I would rather rely on tuning it correctly.

Juha G

Hi Eric,

That would indeed be a very good alternative. You can just plug in a wide band lambda sensor in the exhaust and use a standalone lambda unit to get the AFR's correct. i.e. no problem that the stock unit has only narrow band lambda. Lambda correction is made only in the closed loop operation and does not affect WOT mapping.

But, I think the method of using a Ostrich EPROM emulator is much work and I doubt it will be cost effective unless you do all the work your self!? Correct me if I'm wrong but it requires removal of the original eprom chip etc. And the user interface is not really tuner friendly.
Nevertheless, it's a valid option but not exactly a plug and play like what we are working on.

My friend is on the way to my house and we will work on getting the MAP sensor installed and dialled in. Stay tuned!

evoderby

Hi Colin,

A very similar functionality is built into the Pectel EMS that Neel has been using on the 964 'Cosworth' engines. I don't necessarily see the Safeguard knock control example as the 'absolute truth' in engine tuning. It just shows the potential when used on a certain engine previously tuned for MBTminus a few degrees for safety. The example stems from a controlled test environment and a credible source.

For those not having read the detailed info in the links, the knock control 'simply' allows winning back those power compromising 'safety degrees' in complete safety. Neel has also commented along similar lines, especially when turbocharging is involved.

One thing that got me thinking real hard:

"Colin:In my view the object of the (tuning) exercise is to set the ignition timing and fuel mixture to optimise the transient response of the engine and hence give the car maximum acceleration, a status that will not be achieved by measuring a notional mean best torque on a dyno to deduce the ignition timing, or indeed by ensuring the mixture is exactly La=0.88"

I find this a really interesting view, which also has me genuinely puzzled in trying to digest using basic laws of physics stating Force=mass x acceleration. If I understand you correctly developing for maximum acceleration is the key (as a driver, something hard to argue with from a seat of the pants point of view). With mass being a given constant we must therefore optimize Force (Torque) at every point in the effective rev range to optimize acceleration. Isn't this exactly what tuning ignition for Best Torque tries to achieve?

At the same time I know how the (proper) acceleration jets on a trusty old Weber increase throttle response (= engine acceleration) over SU's not so equipped, even if both carb types flow the same amount of air. Surely there must be an optimum to be found in acceleration enrichment that can be optimally programmed in an EMS(?)

I understand that an engine performing through its rev range is a highly dynamic if not complex environment not easily captured to the fullest 100% by theoretic concepts. Accordingly I'm genuinely interested in learning more advanced views since unfortunately no EMS I know of has a "Newton's 2nd law self programming box" to be clicked.....

"Colin:Now I'm here, maybe you've just caught me at a moment of weakness Harald?"

And look at how I've now taken advantage of the situation, I hope it hasn't caused your cold to get any worse now;-)

Best regards!

Juha G

Good discussion! Keep it coming!

I certainly hope Colins feeling is nothing like my hangover (from yesterday's trip to the Helsinki Boat show) right now....

evoderby

993inNC

Read all nine pages, way above my pay grade so I'm just going to subscribe and learn. Nice to see Loren getting over his illness (Verburg's disease*)


*The long time regulars will understand that last statement.

NineMeister

Well, no, not exactly. From what I understand the Pectel ecu is tuned accurately by Neel on the dyno and the knock control is used as a safeguard, which in my opinion is the correct way to go. The purple box you mention advocated seriously over-advancing the engine and then relies on the box's algorithms to retard the ignition to the point where knock does not occur. Maximum ignition advance without knock is not necessarily optimal.


OK, so how do you know what is happening in the actual combustion chamber during individual burn events? How accurate is the in-cylinder mixture going to be when it is measured by a single lambda sensor located a metre down the exhaust pipe? How can tuning an engine for MBT on a fixed inertia dyno optimise engine acceleration in all gears?

"Newton's 2nd law self programming box" . Interesting. To do that you would have to measure what was happening in the engine, maybe with in-cylinder pressure sensing, high resolution rpm sensors and a very powerful processor running at a fast clock speed? Yes, that would work.

Much better from a good night's sleep, thanks Harald, and yes, advantage well and truly taken.

Geoffrey

I've done a fair amount of knock programming with the MoTeC knock box and it is a complex environment. One knock sensor is sufficient to know which cylinders knock individually, and manage the knock on each cylinder irrespective of what happens in the other cylinders. Since the ECU knows where each cylinder is at any time, it also knows which cylinder is knocking when the knock sensor activates.

There is a frequency in each engine wher knock is best identified and can be thought of as s signal to noise ratio. You use an analyzer to determine where in the frequency range knock is best heard and separated from other engine noise. An example is where a car has standard engine mounts and you replace them with solid engine mounts, it may cause noise at the same frequency where you are listening for knock, making it harder for the ECU to differentiate between knock and engine noise.

Second, once the engine starts knocking, it will never return back to the thermal environment as before knock without significantly reducing the volumetric efficency (less throttle), even if you retard the ignition timing.

Third, with the MoTeC knock module and I suspect others, there are many parameters to control knock. How much ignition timing is taken out, how much is put back and how quickly. Additoinal parameters are available for what happens when knock occurs when putting timing back.

You cannot simply let the ECU run the engine on the knock module, you'll never get full potential out of the engine. Further, steady state ignition timing values will be different from transient and that is a hard concept to understand and program.

Juha G

Good information Geoffrey! Thanks for posting!

My friend is over and we were drinking coffee and chatting about the knock control. It turns out the MS3 DOES have reservation for knock control.
Since the 993 engine is fitted with knock sensors from the factory, only additional thing required to enable knock control is to have a aftermarket unit added. So this ECU is in that way no different from other aftermarket ECUs.

But, my understanding is that the KC is most probably not worth the hassle. (As posted by the engine gurus here too).

Then a question to the gurus; does the original ECU use the cylinder head temperature for anything?
Should the timing be retarted at some temperature?
With no load on the engine we only see 100-120 degrees celcius but it's a different game when you put load on the engine.

NineMeister

Can't speak for the Motronic, but in all 9m Motec conversions CHT is one of the the primary correction factor for the Motec fuel and ignition tables, in particular for the warm-up cycle fueling and idle speed control.

In the aircooled 911 engine CHT is the best indicator of what is happening in the cylinder than oil temp. Under full load on track (or on the dyno) I would expect to see a race engine up at 180-200C.

Lorenfb

"In the aircooled 911 engine CHT is the best indicator of what is happening in the cylinder than oil temp."

CHT is only a 'proxy' for the actual cylinder temp, as it's an average over time and
gets averaged with the case temp. It's not really that good for use in preventing
detonation of certain cylinders and under unique conditions. Just a few degrees
of retard for temp was used in the pre-knock control 911 engine (911 3.2 DME),
and far short of what's required for detonation safety.

So, it appears, as a result of some of the recent posts, that more seem to
'see' a value in knock control as being an integral part of any EMS where
the goal is to achieve maximum torque without detonation and not just
to deal with octane issues. That's good!

The decision as whether to use or not use knock control is quite simple;

Option 1

A. One can maximize a torque curve by tweaking the ignition timing under
various loads and conditions, e.g. temps, and form a timing map.

B. Then based on a combination of potential unknowns & untested on the dyno,
e.g. loads, temps, & octanes, 'dial back' the ignition maps to attempt to provide
a margin of safety to avoid detonation, but still run the risk of detonation. Or,

Option 2

Do Option 1A, but minimize the extent of the 'dial back' of Option 1B by
having a knock control system and thereby achieve an enhanced torque
curve over that of Option 1B and provide for better detonation prevention
that Option 1B.

"Since the 993 engine is fitted with knock sensors from the factory, only additional thing required to enable knock control is to have a aftermarket unit added. So this ECU is in that way no different from other aftermarket ECUs."

It's not that simple! The key, as previously mentioned in another post (944 guy), is the
algorithm used to discriminate noise from the actual knock signal. Do a search on papers
written by Bosch (early '80s) and Nissan about knock control and the techniques used.
Remember, the goal with knock control is to determine the real knock and not the false
knock which can result in a loss of torque when unnecessary.

Juha G

I think we can all agree with Loren's post above. And it seems that both methods are widely used.


Yes, of course it is used for cranking and warmup enrichment + idle control but my question was if there is also a upper limit of HT for safety reasons?

On this ECU we have used the IAT based temperature correction for the ignition and fuel when engine is in normal operation temperature.
There is also possibility to factor in the CHT (% of CHT used in IAT vs. rpm).

Juha G

Yes, it is not simple but my point was that it is possible and the situation is same as with any other aftermarket ECU.