Ive tuned a ton of cars and i feel ashamed for not knowing this, but how is the timing actualy advanced or retarded? I know you can do it with adjusting the distributor, but when you enter in values in tuning software, and nothing inside the dizzy acualy moves, how is timing being advanced or retarded. Is it in the duration of the rotor when the spark is released...as in the position of the rotor in corelation to the post in the dizzy. Cause goin through a full sweep of 16* at idle, up to 46* and being able to retard for boost or what not, it doesnt seem like the rotor would cover more than 15* or so of dureation. Im not sure how to look it up on google or something along those lines. I hope someone actualy understands what im asking.

 

That doesn't really tell us anything about adjusting timing under s300 though.. it just gave an explanation in general about the module that controls the ignition.

 

You know how the rotor is like 60 degrees wide? That gives the ECU roughly 60 degrees of adjustability as the rotor passes each point. It knows where the rotor is from the CYP, CKP, and TDC sensors present in the distributor.

For the record, the ICM is basically a big transistor that takes in low power square wave signal from the ECU and grounds the coil accordingly.

 

This is the kind of answer i think im looking for....

So given that the sweep of the rotor covers potentialy 60* of rotoation, the ECU basicly tells the coil when to fire the spark and the rotor has that time its sweeping past the post to fire? Thats what i came up with in my head, seems like the only logical theory.

 

I'm going to guess here, but I think there is probably a table somewhere in the ecu that takes the RPM of the camshaft along with the data input of each of the 3 crank sensors and determines the angle of the camshaft at any point, based off pre calculated values of angular speed.

At 8k rpm the camshaft rotates 133.3 times a second... it's pretty amazing it works the timings involved to get .25 degrees of timing advance is nearly zilch!

 

It's not so much a table as much as it is a scheduling formula. Basically you've got your CKP with 12 teeth per cam revolution, TDC with 4, and CMP with 1.

Once it is synced against the CMP, then it can (theoretically) rely solely on the CKP sensor.

Everytime the CKP tooth edge is seen, it looks at how long it was before the last time it saw the CKP tooth, and then how long it was from that. Based on those three values, it can calculate angular velocity (as well as acceleration) allowing it to effectively "schedule" timing events. It has a master clock (or free running timer more technically) that can run at the chips operating frequency (assume 10 MHz). With that in mind 133Hz is nothing. At that RPM you'll have ~75,000 main clock ticks to schedule timing events against. So in theory you can have timing resolution down to 75,000/360 = 208 clock ticks per degree.

Unfortunately, timers in these chips are still only 16-bit so you can only schedule a timing event 65,535 ticks into the future. In order to make this work, they slow the free running timer down (usually divided with a prescaler) so that at maximum RPM, the time it takes between teeth doesn't exceed the capacity of the timer. (So you can schedule a timing event even right up to the point before the next tooth is seen)

Bear in mind that there are probably far more complex formulas for scheduling timing events, but this will likely satiate your curiosity at least for a few minutes

 

Ahh that makes sense... It sounds like you potentially do something in the electronics industry like program AVR's or write code, or perhaps assembly.. But I follow it.

I still think its pretty slick how this technology is pretty old.


It makes me wonder why Honda never used a distributer'less engine earlier.. Obviously with 32 and 64bit cpu's even a 32bit ecu would be ridiculously more accurate.

The things I have fiddled with in the past, sometimes the "Kiss" "Keep It Simple Stupid" methods still are probably the best for problem solving..


I think one thing engines is missing now is a pressure sensor per cylinder, similar to that of a knock sensor but mainly for avoiding pre-det. You would be able to determine MBT. If this is the case software like neptune could be used to log ignition timing vs cylinder pressure. We could then figure out peak ignition timing values for every rpm and make some pretty amazing ignition maps.

You could do something similar to this now.. just with no sensors.. run the car on the dyno look at torque at a specific rpm and adjust ignition timing till torque is peak then move onto the next rpm range... although this would be pretty expensive to do unless you had your own dyno.


--Aaron

 

that's why you rent time on a dyno if you don't own one

 

Funny guy

 

I've spent my fair share of time inside of IDA, K-series ROMs specifically. I don't particularly care for AVRs, I'm more of an ARM guy myself.

 

Ahh very cool. Maybe I could email you some assembly code to make sense of :O.... I just need a couple bytes changed... I just dont know where to change them hahah..

--Aaron

 

there are combustion pressure sensors out there, but at high cost and with packaging issues. the sensors must be calibrated often. they are called optical pressure transducers, google it and read up. there has been lots of R&D done to develop this technology. i have also seen strain gauges fitted to head bolts that measure strain. when you discover this unique solution we can market it as "unobtainium" and save millions of gallons of fuel!