somberg

Peter

Maybe you should read the article I posted a link to in my previous post.
I think "essentially useless" means "can be improved upon because the proverbial 80/20 solution has been chosen". These cars are produced as a mass article; I think that if they achieve 80% of their objectives with 20% of the money they're satisfied. In the article you'll find acknowledgement for this. This also explains why a cheap solution like the one of the GT-R is working up to a certain point and why it is better than nothing. Tuners use this because it is all they have. It would be far better to have an individual cylinder detection + control strategy for the full RPM range because that will increase engine efficiency, but that costs a lot of money. Also note that the early R32 GT-R computer was a concept from the end of the '80-s ! I wouldn't say that the R34 computer is the same, but it's not a quantum leap. It is my opinion that car electronics as they are implemented are not always state-of-the-art to say the least. That has a lot to do with cost and the need (or better the lack thereof) for technological innovation.

Andre.

__________________
Bananaman..

Peter

Hi Andre,

It's too late to read another long technical article tonight, my head hurts! I'll summon up the courage later on.

I suppose thinking about the stock knock sensor logically, it was probably considered sufficiently sensitive to use with what is essentially a pretty simple det. prevention system on the stock car. It may not have been designed to operate to the level of sensitivity required by 'modern' tuning systems such as data logging with the FC Pro, etc. As with everything everywhere, the accountants had their say no doubt.

The over-fuelling/increased det. issue interests me as I was told this could be the case quite a long time ago by Mark and I have to say I found that hard to take on board but perhaps he was right after all....?

Cheers.

Peter.

somberg

Hi Peter

Last one for tonight, got to get some sleep

That's why I dislike accountants occasionaly. Don't think it's just the electronics that is affected... I've heard some weard stories about VAG engines in this perspective. Everything comes with a price tag attached !

As to overfueling/increased det: yes maybe. There's another thing you get from overfueling: bore wash..

One thing is for sure though: I will run my car only after I'm able to control and log all the parameters needed to keep everything in one piece

Andre.

PS there are some very nice knock sensors avalailable that will go under the spark plug

PS2 for sure we'll get to the absolute bottom of this

__________________
Bananaman..

22B

Mycroft,

2 things.

1. is that your name, or are you sherlocks older more intelligent brother called upon when he is stuck?

2. I have read your posts with extreme interest and have to say I believe what you have said. I must stress that as a physicist I am duty bound to hate all chemists but what you have said seems to ring true plus you have backed it up by some sound scientific theory that makes sense to me.

My car had previously been mapped for ignition based on the output from the knock sensor. I have a remappable ecu rather than relying on the standard jecs ecu in the 22B to retard the ignition.

What I would seriously like to know, and I suspect you may have no idea about this, is whether the masking of the detonation (or explosive deflagration as I have been told to call it) would still be detectable using det cans?

I think ultimately det cans are a far better way to map so future mapping will be done this way.

Also can you give details or links to the digital knock sensor you referred to and possibly the pricing also. I have just boyught myself a pectel T6 ecu which has very complex knock correction software, and am hoping it may have the capacity to receive inputs from digital knock sensors. Due to its complexity I am guessing it does.

more importantly can you let me know where you got the information about the 2k rev useful rev limit of the basic japanese knock sensors.

many thanks,

Adam

somberg

Adam

I've looked at Pectel years ago.
You wouldn't happen to have an electronic manual or a link to it, that you could provide me, would you ?

Andre.

__________________
Bananaman..

somberg

Faster than my own shadow http://www.pectel.co.uk/

__________________
Bananaman..

pat

Yikes, long thread

First a few basics... knock is in general an "abnormal combustion event". This can fall into many categories, such as pre igniton, "detonation" etc. Each has differing characteristics, and will "sound" different listening to the engine through "det cans". In most cases it upsets the cylinder pressure curve, and may or may not have an audible presence. I would assert that the only true way of detecting such events is by actually measuing the cylinder pressure curve using devices such as the optrionic pressure sensors made by Optrand and used by F1 teams. The actual detection system would presumably employ some DSP algortim... first gut reaction would be to use a nonlinear time variant tansfer function with a finite impulse response but I'de really have to sit down and think about it.

There are many knock sensors available with varying characteristics; for example NTK/NGK have low Q, high Q and flat response sensors. Simply using a high Q sensor with a gated detector would not be sufficent to tell the difference between the background noise and the actual knocking signal unless the signal processor had an attack filter, measuring the sudden increase in amplitude at the centrepoint of the response curve (which may or may not be 6.4kHz). Probably the best method of detection while minimising the number of false positives while not missing any events would be to use a linear response sensor with a DSP system that looks for abnormal variations in a spectral analysis of the signal composition. It would be reasonable to assume that there would be an occurance of a spike centred on 6.4kHz with further harmonics, a phenomenon which could reasonably safely be linked to the presence of kock of some sort. That doesn't eliiminate the possiblilty of weird effects related to (for example) valve bounce being coincidental with the knocking signal, but gating the knock detection system to a window centred about the combustion cycle would reduce the possibility of false positives. The trouble with engines with a higher number of cylinders is that you end up with more power strokes per revolution and thus the percentage of the revolution open to the detector by the gate will increase, thereby increasing the likelihood of false positive detection. It is for these, and other reasons, that I belive that cylinder pressure curve monitoring is the only true means of reliably detecting the presence of knock.

With regard to the effect of the octane boosters, much has already been said. There are two (probably more) angles to look at this from. Each phenomenon (pre ignition, "detonation") has differing criteria for the fuel's characteristics.

If we assume that (and this is not a safe assumption) there are no glowing deposits in the combustion chamber during the compression stroke then pre-igntion will be caused only by the adiabatic compression of the charge causing the temperature to exceed the autoignition temperature of the fuel and thus the mixture lights up before the spark lights the mixture. There are at least four ways to control this phenomenon... a) reduce the initial charge temperature, b) reduce the compression ratio, c) increase the autoignition temperature or d) increase the standard heat of evaporation of the liquid phase consitutents of the charge. Option a) can be achieved by means such as a more efficient intercooler, driving in colder weather, running less boost etc. Option b) can be achieved by either a piston swap or reducing the effective compression ratio by lowering the supercharging pressure. Option c) is the realm of fuel chemistry (over to you Mycroft!). Option d) can be achieved by mechanisms such as water injection, which effectively increases the relative humidity of the charge and thereby reduces the effective temperature rise due to adiabatic compression. In general though, pre-ignition is much more likely to occur due to glowing deposits, and sometimes simply due to running too much boost for the compression ratio.

The second (note there are more!) and perhaps more common mode of abnormal combustion, often referred to as "detonation" occurs after the spark and generally results from the gas in front of the flame front autoigniting and the resulting flamefronts (and pressure waves) colliding. It's not really a detonation as such, but I won't go into the details of why not as it's not relevant. Anyway, there are many methods to control this type of abnormal combustion event, in general a safe bet is that if ignition advance is reduced then the total combined temperature due to adiabatic compression and the heat from the flamefront is lowered to below the autoignition temperature and the knock is removed. As a general rule of thumb, manufacturers opt for compression ratios which exceed the maximum possible to achieve MBT timing at maximum supercharging pressure; this compromise is necessary to maintain some element of driveability off boost (the effective compression ratio of a low compression ratio in vacuum is so low that the thermal effiiency drops so drastically that the engine make virtually no torque). Knowing that the engine cannot achieve MBT timing on boost the manufacturers opt for the least "painful" compromise... continually running on the limit of detonation and using a knock sensor to react to differing fuel quality and atmospheric conditions.

It all boils down to effective in cylinder thermal management. If one can control (by whatever means available) the temperature rise in such a way to keep it below the autoignition temperature than knock can be inhibited. As stated one mechanism is by reducing the ignition advance. This alters the cylinder pressure curve to move the maximum cylinder pressure away from 14-16 degrees ATDC, which also reduces the maximum pressure at the same time (and also the thermal efficiency). Basically peak temperatures are reduced at the expense of higher thermal input to the exhaust valves. Of course there are other methods available, such as octane boosters (either real ones which increase the autoigniton temperature or "fake" ones which act as flmae inhibitors), reducing the temperature rise by raising the standard heat of evaporation of the liquid phase constituents of the charge etc.

In the case of actually increasing the autoignition temperature of the fuel, it is possible to maintain the peak cylinder pressure at a point near 14 degrees ATDC (MBT timing) for a higher effective compression ratio or higher initial charge temperature. This is true octane boosting, literally making the fuel "better". Not many things can achieve this (to date, and correct me if I am wrong, the most effective "additive" ever found is still Tetra Ethyl Lead or TEL for short). Any additive capable of functioning in this way will show significant gains in power output.

Damn, the post is too long, gotta split it!

pat

Part II....

Sorry about the formatting it all went to pot during the cut and paste from the original post which was rejected for being too long!

In the case of the flame inhibitor, this slows down the flame front, with
less energy released by the burning mixture the charge does not get as hot
with the same ignition advance, and hence autoignition is avoided. It then
allows the ignition to be advanced again and initially it would look like
power has been restored. But there is a catch, of couse. In a perfect
world, fuel would burn instantaneously and thus ignition advance would not
be necessary in order to achieve maximum cylinder pressure at 14 degrees
ATDC. But this is not a perfect world, and ignition advance is necessary
in order to place the peak cylinder pressure at roughly 14 degrees ATDC.
By making the charge burn more slowly, sure the peak cylinder pressure is
reduced at the same advance, and the temperature may be at tolerable
levels at the "nominal" (ie timing before retard due to knock) but because
the mixture is burning more slowly, this timing will still not place the
peak cylinder pressure at 14 degrees ATDC. In fact you'de need to run
even more advance to get it there. The question then has to be asked
whether by slowing down the flamefront and allowing additional advance,
the thermal efficiency at that operating point has been affected in an
advantageous way. Or have you just allowed the engine to run with the
stock ignition map without detonation but not actually making any more
power than it did with the retarded ignition?


The increase in relative humidity (either by driving on a rainy day or
through water injection) may achieve its positive effect on an engine's
propensity to knock either due to the fact that the initial charge
temperature is lower, or due to the fact that peak cylinder pressure is
reduced due to the additional heat "removed" by the evaporation of the
liquid phase water. The water also has an impact on flamefront
propagation, and it could be argued that this has a similar effect to the
flame inhibitor, but it can be shown that the energy "robbed" by the water
can be recovered, and more by allowing greater advance (and hence thermal
efficiency). This is a halfway house between flame inhibitors and "the
real deal", that does appear to give reasonable results (but of course not
as good as real octane boosting).

An interesting point to note about the measurement of Research Octane
Number and Motor Octane Number is that it is performed at 600 and 900 RPM,
respectively, with a variable compression engine (the crank can be moved).
Sadly this is not quite as good an approximation to a real engine as
perhaps one might think, because a real engine has the benefit of squish /
quench areas. These inhibit the autoignition in the endgas either because
they allow heat to be drawn out of the charge or because they squash the
charge down to such a thin "sheet" that combustion as a whole is not
sustainable in those areas. As one varies the compression of an engine by
altering the location of the crank one also adds or removes the effects of
squish / quench. Of course one could stipulate that at no point should the
test engine be allowed to have squish and thus the effects can be ignored,
for test purposes at least. Also important in the real world is charge
stratification and the effects of swirl within the combustion chamber. It
is possible to design a combustion space capable of operating at 23:1
compression without knock if one pays particular attention to charge
preparation, localisation etc. This is well in excess of the theoretical
14:1 limit (or thereabouts) for gasoline SI engines. There's more to knock
control than meets the eye!

I'm sure that there as many many more things to take into account but it's
getting late and at this rate this post is gonna fill up a whole page on
its own

Cheers,

Pat.

kowalski

I'll get my coat...........this is so over my head its not funny

__________________
If it goes fast, I NEED it.............

Hugh Keir

Pat,

You sure know how to make an entrance, posts 1 and 2 made about the best reading I have seen on this site.

Got any more on cylinder pressure curve monitoring ?

I have seen a spark plug drilled with a pressure sensor and also a piezo electric washer that fits under your spark plug. I contacted the manufacturer of the in plug pressure sensor, but their response was it is work in progress. Autronic also claim they will have system that works with their ECU's in the next year or so, but don't hold your breath.

Some Hyundai engineers also presented a good SAE paper 18 months ago on their cylinder pressure curve monitoring test results which also looked at whether power was still increasing at the onset of detonation.

Thanks for the contribution.

nitro

"MYCROFT"
Perhaps i have been a bit strong in my point of view and trying to understand the "FOOLING" part, if so an appology is given to you.

"PAT" a great indepth thread

__________________
When only the BEST will do!!

pat

Hugh,

I did have a datasheet for the Optrand device, I got that a couple of years back, and although it looked very interesting I was put off by the fact that a) they were USD 800 each (and I needed four), b) I needed to drill holes into the cylinder head, c) they required special interfacing as they are optronic devices. At the time they were definitely in use by at least one F1 engine manufacturer, as the datasheet clearly showed (it had the team's name on it, ahem...). The situation has probably improved somehwat over the years and the devices are probably a lot more user and wallet friendly. Optrand still appear to exist, which is a good sign and perhaps it may be worth investigating this again.

As you stated there are manufacturers who make spark plugs with integral pressure sensors. If memory serves correctly, Optrand did do something like that too, and NGK/NTK certainly list such plugs in their catalogue. Interestingly the NTK UEGO oxygen sensor is still "experimental" so it may well be the same for the pressure sensing plug.

The great thing about these sensors is that effectively you don't need an ignition advance table. Your DSP can perform closed loop advance optimisation by continually keeping the spark at such a point as to get the peak at or around 14 degrees ATDC, and if it detects any abnormal pressure variations it can pull the timing back to a suboptimal position to prevent the onset of detontation.

One thing that perhaps I forgot to say in my previous post, at least not in a roundabout way, and it is perhaps slightly at odds with what Mycroft has said, is that in my view, these octane boosters genuinely DO reduce the fuel's propensity to autoignite *given the same conditions in the same engine*, rather than just disguising the detonation. If it really just disguised it then there would be evidence of det on the piston crowns and you'de end up blowing head gaskets left right and centre. However, the method used to reduce the fuel's propensity to autoignite is NOT to raise the autoignition temperature (which is what a good octane booster would do) but to make the fuel burn more slowly, effectively making the fuel worse than it was to start with, but in doing so saving the engine. You wouldn't necessarily see this effect on a knock engine but you would on a normal engine. It effectively alters the cylinder pressure curve in such a way as to delay the onset of detonation.

In short... an off the shelf "magic bullet" octane booster can save your engine, but it will not necessarily improve the power output. To stave off detonation AND make better power, you need a "real" octane booster such as TEL, Benzene, Toluene etc

It is possible, however, that the effect on the cylinder pressure curve of the "magic bullet" may have a net positive effect on power,, but certainly not to the extent that a genuinely higher octane fuel would have.

Cheers,

Pat.

pat

All,

Just a quick note re: Optrand... they now do a "CALplug" whereby they modify a production sparkplug to fit an optronic pressure sensor for continuous high speed use (they suggest motorsport applications). Their catalogue is available as a PDF download from their website at http://www.optrand.com,/ and it would appear that they have indeed made the sensors much more user friendly with 0-5V outputs etc

Cheers,

Pat.

Peter

Pat,

Very interesting posts although some flew over my head too but good stuff.

So, getting to the point of this thread, on balance, would you recommend the use of a 'magic bullet' octane booster generally? In my case I do only use one for track days when higher than normal water/oil temperatures are seen and, despite it perhaps not being altogether accurate, the det. sensor does indicate an increase in det. readings if I did not use one. I am not looking for or expecting any power benefit from it's use, it is purely as a measure to reduce the likelihood of det.

Hugh,

A few posts back you posted this:

For Apexi users like Peter, they have the option of adjusting the “ high water temperature / ignition “ and “intake air temperature / ignition “ maps to control things, but it sure is tempting to tip in one of these bottles for safety.

I've had a look through the functions available to me through the Commander and cannot find these. Are they only available using the FC Pro? The water temp screen does not allow me to change water temperatures.

Peter.