Ignition advance

[Q45tech]

http://carprogrammer.com/Z28/P...E.pdf

Please don't post till you have read and understand all 207 pages

[PoorManQ45]

I have read all 207 pages, but I don't understand.

Actually, that's a very interesting book

[Q45tech]

Wouldn't expect a non engineer to get much from those papers.Just for you! some reading to tide you over.There will be a quiz!

http://home.planet.nl/~monstrous/tutcar ... rmulas.xls

http://delphi.com/news/techpapers/http: ... ers/other/

http://www.eng.ed.ac.uk/~mzais...s.pdfh ... .g....html

[IvoryJ30t]

with a motec m800, it would be nice to have individual cylinder PPP correction.

i was reading another paper that outlined an ionic spark advance controller with CAN bus communication capabilities.

i wonder if a capacitive discharge ignition would interfere with the ionic readings derived from the spark plugs.

i like the capability of detecting knock via premature ionization instead of listening for specific frequencies via a microphone [conventional knock sensor]

i like the ability to program a very conservative spark advance schedule, while the controller optimizes the actual advance. decreases the risk of blowing the motor to pieces while establishing base map/tuning.
Modified by IvoryJ30t at 7:12 PM 9/28/2004

[Q45tech]

They make pressure sensors that look like a washer with a tiny shield wire for use between head and spark plug.................a knock sensor for each cylinder, but way more as you can read the internal pressure rise shape and amount...........this lets you set individual cylinder advance precisely.

Commonly used in labs and on fornula racing engines

[IvoryJ30t]

i found a company that sells this system as a standalone ignition with CAN bus communication [comm with motec], plasma spark, and AF ratio cylinder balancing.

base price is $22,500

i guess ill be going alot more research, and try to construct my own. that ignition costs almost as much as my entire planned project.

that system was geared towards racing teams [deep pockets]

i could probably get it done for a few hundred. all i really want is the knock detection. i have a few years to play with the concept.

[Q45tech]

Because of runner imbalances and fuel injector variations [oem specs are 3% at best] not unusual to see 5-7% or more AF variations between cylinders. worst to best or 3% from mean.

A fine crankangle sensor can provide 1 degree spark accuracy up to some rpm

A cheaper way is the individual injector balance box that lets you add time [up to 15%] in 0.1 millisecond [1%] increments. Harder to get right but almost as effective............use individual exhaust temp sensors per cylinder on headers...........build an 8 bar graph to tune.

Fuel density can vary so you must measure your fuel and adjust for fuel temperature and viscosity between batches. Common source of error as a ecu fixed injector time cannot be right without some feedback to correct for INTERNAL variables...............outside the normal MAF correction.

Unfortunately EXACT ignition point to result in 16 ATDC pressure peak is a rolling number BECAUSE of the homogenity of fuel.....the mass may be the same but it won't always be equal distributed in the air.Studies have shown than the ignition point per cylinder may scatter 1-1.5 degrees either side of ideal.........varies with air temperature

[IvoryJ30t]

on the VH, are the knock sensors tuned to the engines resonant frequency, or does the ecu actively filter out noise?

[Nismo_Freak]

Damnit... that just makes me want pressure sensors in my chambers!

I understood most of the points, and overall the concept of it but honestly it seems like a written cry for greater engine management lol. I don't forsee a common man being able to apply any object from the thesis. The only thing that anyone could use is a deduction from pressure, being the expelled exhaust temperatures as a function of ignition timing at a given rpm. This already is a common tool for tuning the engine's efficiency by your everyday person, with or without a greater understanding of why or how there is a curve in the exhaust temperature in relation to timing and the a/f ratio. This just seems to chalk it up as using EGT's to tune the engine is like trying to draw a map blindfolded, as the EGT's will vary in accord to ignition timing, a/f ratio, chamber dynamics, and the condition of the engine as a whole (crevice flow, blow-by, etc).

This makes me feel hopeless lol... so ... who wants to make some pressure sensors

I do feel empowered by the negation of part of his "dilema". The lack of due regard for the emissions output

[IvoryJ30t]

emissions output can take a walk.

i would like an ionic detection setup that could send a variable resistance or voltage signal to an analog input on a motec ecu to trim ignition on an individual cylinder or bank of cylinders.

the hard part is setting up a microcontroller that can decide what is the magnitude of abnormal combustion present, and send the appropriate signal of correct proportions so that the ecu will pull the minimum amount of timing out without affecting overall power due to excessive retard.

[Nismo_Freak]

emissions output can take a walk.

i would like an ionic detection setup that could send a variable resistance or voltage signal to an analog input on a motec ecu to trim ignition on an individual cylinder or bank of cylinders.

the hard part is setting up a microcontroller that can decide what is the magnitude of abnormal combustion present, and send the appropriate signal of correct proportions so that the ecu will pull the minimum amount of timing out without affecting overall power due to excessive retard.

On the Motec you can set range... so you wouldn't need a microcontroller.

[IvoryJ30t]

i would still need a controller to decode the ionic current readings from the plugs.

im playing with the motec software [have been for a few months] and i would set an ignition compensation table based on rpm and the amplitude of recieved signal.

i havent figured out a way to control the individual cylinders based on an input signal.

if i cant have individual cylinder knock based ignition control, i might as well devise a bandpass filter and amplifier to control an ignition compensation table based on the knock sensors.

basically, if im gonna put this much time and effort into this project, i want the best control scheme i can possibly configure.

[IvoryJ30t]

without designing an entirely new ECU, this doesnt seem feasable.

i cant figure out a way to control individual cylinders on the fly. the entire map can be trimmed on the fly, and individual cylinder fuel/ignition maps can be used, but not adjusted at run time.

since the ionic feedback systems work on a per cylinder basis, it would not be wise to use individual sensors just to make overall ign map changes.

the motec corrects for intake air temp, fuel temp, coolant temp, ect.

those maps correctly setup, plus my own scheme should give me the control i need.

now, beyond the internal ECU corrections, i have a plan.

i would run the knock sensors to a controller of my design/build. i would preamp and filter the incoming piezo signal.

after the signal is bandpass filtered according to the resonant frequency of the motor, i would have a voltage signal output according to the amplitude of knock.

the amplitude of the signal will be adjustable [sensitivity control].

it will interface through a peak and decay [forgot real name] timer.

the motec would have a correction table based on input voltage.

the controller's timer would output a voltage referencing knock once the recieved signal crossed a threshold. the peak and decay timer would taper the output voltage towards zero providing no more significant knock is recieved.

in effect, that would retard timing, and add the timing back as long as knock is insignificant. there would most likely also be a fuel enrichment table watching the input to the ecu, so that the mixture is richened slightly as timing is pulled.

it sounds a lot more complicated than the circuitry actually is.

the sensitivity of the dectection circuitry, as well as the decay time off the timer, would be adjustable in order to tune the system.

a headphone output would be nice.

in addition, there would be a dash mounted potentiometer that would have direct control over a second ignition compensation table.

this sounds overly elaborate, but if im going to continue to save the amount of money needed for this project, i want absolute control over what my $10,000+ engine is doing.

[IvoryJ30t]

now, the big question.

are the knock sensors tuned, or full range?

if full range, anyone know the VH's resonant frequency?

[Q45tech]

~~ 6164 Hz based on 900/[pi x cylinder radius in meters]= 900/0.146

Most knock sensor system have a difficult time differentiating knock from engine sounds at high rpm........thus the need for at least 2 on a V8 and very sharp filtering and a narrow time gate window.

The frequency will vary from cylinder to cylinder so 5500-6800 Hz multipole band pass filter, etc

The following data sheet on an IC designed for the purpose should help:http://focus.ti.com/lit/ds/symlink/tpic ... sensor.htm

As you see 5900, 6120, and 6370 Hz are common

Google "knock sensor frequency" = 10,100 returns.............pdf files =3420

[IvoryJ30t]

wow, you always find the best files. i could [and do] sit for hours reading the stuff you post.

i think its funny that the middle report has a hardware layout almost identicle to the layout that i posted above off the top of my head.

i didnt think of referencing RPM, because i would be using analog circuitry [op amps, comparitors, a 555 timer ic for decay, ect], and the motec would be referencing the signal against a map with rpm as the x axis, and signal level as the y axis.

i can layout circuitry, etch and drill a pc board, solder components, but im limited in what i can do because i cant write code. that rules out microcontrollers, unless i get some help on coding.

i like the design of the thielert controller, but i cant easily replicate the increment/decrement function without knowing how to code.

my setup would basically be a dumbed down version of the thielert, unless its somewhat affordable.

[IvoryJ30t]

that TI chip looks promising. would bring the component count down considerably.

i have to read the entire spec sheet. skimming through it, it looks good.