Running full boost on a public highway looking like some sort of half arsed astronaut with fuel pipe coming out of your ears and into the engine bay, might attract a little bit of attention. Especially from the old bill.

There is a more scientific way of doing these things you know.

Graham.

Graham - would love to hear of a more scientific way!

Fuelling seems to be pretty easy to sort, ignition timing is more tricky, any light you could shed on this would be great!

_________________
Kris
95 ST205 - slept for 10 years, now waking up...
Avensis
http://www.gtfours.co.uk
_______________________________________
Turbocharging - the replacement for displacement

A laptop plugged into the appropriate sensors would be a good choice.

Anyway that apart, lets consider the following scenario.

You are on full throttle in third gear running through 4.5k rpm at 18 psi boost and your det cans detect pink. What are you going to adjust? Ignition timing, duty cycle on the injectors, fuel pressure on the regulator?

Boost air temps will also have an effect on how you determine a solution. How are you going to monitor this without a laptop?

Graham

Graham,

I think the plan is to do this 2-up . . one driving, the other with det cans and laptop in passenger seat (or at least thats how we did it when mapping Penny)

Yes you get strange looks but Mr plod can't rightly do much if speed etc is ok and it is passenger fart arsing about with mapping

_________________
Loosey
Loosey@gt4dc.co.uk
Storm Blue JDM ST205
Super White JDM ST205
http://www.gtfours.co.uk

Some people are like a Slinky .. Not really good for anything, but they make you smile when you shove them down the stairs!

Graham - yes I agree datalogging is good

I would've thought if your afr is good, say 11.5:1 and you get det then I would've thought that you would take some timing out?

Since the afr is in the right region then altering either the duty cycle or FPR is not going to be the best method to reduce knock?

The whole monitoring of knock via a laptop is intriguing. This is what the Power FC can monitor and display, which is certainly useful, but I am inclined to believe that it's usefulness above say 5k rpm is limited?

I very much like the idea of a laptop spewing out a number for knock as seen by the KS, but listening for knock seems to be the prefered method?

_________________
Kris
95 ST205 - slept for 10 years, now waking up...
Avensis
http://www.gtfours.co.uk
_______________________________________
Turbocharging - the replacement for displacement

You know Loosey, I never even considered how the job might be done with two people involved.

I guess trying to do things alone is not always a good option.

Graham

Maybe for det , knock limiting or control of the ecu would be ok if the system used fuzzy logic.

The main reasons would be once det has happen its to late.

using FL would allow this input to be monitored and allow for a pr-emptive output, before the det accrued, this way, one could use it for tuning as it would be based on the cycle before, using look up tables, thus the calculations could be coded and processed within time.

Using analogue is not the way for sure, its not accurate, well that’s what I been told, with my limited understanding.

I use a stethoscope, its old school, and works very well, nothing to date can out process the human brain, it gets my vote for now.


The noise thing is a major problem, even using a filters active or not, therefore would require a much more complicated circuit to sort the harmonics etc to achieve a true and clean output signal, that could be converted into logic , thus being processed , within the small window for control of the ecu before it would be required, this would allow for tuning.

Would this make sense?

Worth logging, all the inputs, for later analysis, this can help; it’s making all the data work for you, which seems the drawback.

jon

All IMHO of course

Whilst det cans are fine for tuning purposes (aside from the cyberman issues lol) they are never ever going to work in the real world. It's going to be hard on track when you need to wear a helmet etc. etc.

In fact as far as I can see the times when det is most likely (very few of us are ever going to be able to map under trackday or other extreeme stress conditions) are precisely the times when det cans are a right royal pain in the proverbial.


Which leads to ongoing knock detection

IMO you should be able to predict det in either analog or digital circuits.

However, in the digital world it is very simple to tweak when you realize you forgot a parameter etc. while in the analog world it gets "messy"

Plus you need memory IMHO. At some point det will happen. Bad gas, nutrino strike etc. The hard part, as you allude Jon, is what the hell just happened. Why did it det?

You need to go through the logs and learn what conditions precipitated det and make a call if this was just an act of god single event or a trend based on evaluation of previous visits into a given load cell

Then and only then can you begin to make any kind of guess about whether knock is on the cards

Then comes the other problem. Without a self programmable (software, not maps) ECU what the heck can you do about the DET you know is just about to happen? I guess you have to break into (purely for eg) MAP and IAT so that you det detector can tweak either timing or fueling

All of this is going to require quite a lot of CPU power to do correctly IMO. You'll need to start running FFT algorhythms on the knock sensor output. Plus logging of fueling, ignition timing, map, air temp, TPS etc. etc. I thing the IO requirements alone would put this beyond PIC territory
In fact, my current "best option" plan is to use something along the lines of a megasquirt switched purely into monitor mode acting as a piggyback onto the main ECU (whatever it may be). You simply rewrite the MS code to build a knock monitor and use one or more of the outputs to drive the main ECU.

Interesting thoughts Steve.

Of course corrections for IAT, water temp etc all seem pretty standard for a decent aftermarket ECU.

Providing that these corrections are setup correctly then alterations away from the optimum tune conditions should be catered for. I suppose the problem then becomes when the "other" corrections to the main map are not correct and don't e.g. pull some timing when the IAT rockets.

A digital monitoring method for det I think would be bloody marvellous, it's just how to achieve it! I think I have a suggestion for an analogue monitor for det...but this requires some degree of proving yet...

The output of the KS looks something like this...

_________________
Kris
95 ST205 - slept for 10 years, now waking up...
Avensis
http://www.gtfours.co.uk
_______________________________________
Turbocharging - the replacement for displacement

I don't suppose anyone knows how the stock ECU retards the ignition timing in response to the knock sensor output? Perhaps it is triggered via a threshold set on the output voltage of the sensor ...... or am I being too naive?

_________________
Don
GT4DC Chairman
1994 Toyota Celica GT-Four ST205WRC JDM 269bhp @ 0.9bar
1994 Toyota Celica GT-Four Special GT 590bhp @ 1.8bar
1989 Van Diemen RF88/89 Formula Ford 1600
2008 Nissan Patrol GU 3.0L ZD30DDTi 154bhp

I would guess it's something pretty simple like that yes although presumably the KS output goes through some signal processing first

I would make an educated guess and say that a voltage that is generated from the KS would inject an offset to the output of the pickup coil in the distributor. The offset would vary in accordance with the voltage that the KS produces. In a 205, the pickup coil might be a hall effect transponder on the inlet camshaft, on a 185 it's in the distributor. I dont know, but either way, the offset would make the ecu think that the cam position is more advanced then it actually is by a varying degree. The net result is a retarded spark.

Just a guess, but it's the way I would do it.

Graham

When the KS retards the stock ECU ignition on an ST205 it is quite noticeable ...... no subtlety involved. That's what made me think it was some sort of threshold setting? All this is my Mk1 'bum dyno' so I could literally be talking through a hole in my posterior

_________________
Don
GT4DC Chairman
1994 Toyota Celica GT-Four ST205WRC JDM 269bhp @ 0.9bar
1994 Toyota Celica GT-Four Special GT 590bhp @ 1.8bar
1989 Van Diemen RF88/89 Formula Ford 1600
2008 Nissan Patrol GU 3.0L ZD30DDTi 154bhp

Hi.

I believe after reading the toyota specification on Denso ecu

It constantly monitors the signals from the knock sensor, but only functions within a window set from factory ie rpm Ne G1&2, kpa, tps set points.

Depending on the signal amplitude received will determine the
Amount of base advance angle retarded in varying degrees,
Until the set level has been reached, then ignition timing will return to normal, until the next event, this is usually measure in time, with, ign &, input signal cycles.

The mega squirt like most other ecu’s offers a user settable range of input knock over ignition pulses, if theses are met, then the base timing is retarded to user set levels, or inclement etc, unitl no further knk inputs. after a pre set time period,it will return back.

Apart from, the logic, which counts the events, within ecu clock cycle routine, the amount of processor usage is very low, taking in to consideration that the ecu code has been assembled for a very clean routine, not clumsy like windoz .

Nice scope image Kris, definitely look slike you need a form of FFT. If you wanted to play about with FFT maybe you woudl be best to look for a program like "MatLab" that has inbuilt functions for FFT. You woudl have to use a laptop to run the program etc so is less than ideal for a permenant setting but usefull for testing.

_________________
JP
GT4DC Treasurer

Grey St185 RC
Silver BMW E46 M3
Multicolour yawn Honda VFR400
Silver Honda Civic Type-R