This pic off the web may give you a better idea about how the timing is varied. kPa is MAP reading. Similar to TPS. Consider 13 kPa as closed throttle, 110 kPa as WOT.

Well, I have just completed my graduation in B.E. (Electronics & Telecommunication) earlier this year. Right now, preparing for M. Tech. I have always been interested in cars/bikes! Have been tinkering with electronics since long. That's about it!

Arresting knocks has been my task for over 2 yrs on my engines now.
My Drag Fiero has a static engine compression of over 230psi and at dynamic, its shoots really high and there are hardly a few ways to arrest engine knocks. I wasnt getting supply of aviation fuel, and speed 97 was still not sufficient.
Anybody else would have thought of dropping the compression to keep the engine safe, but I never like to go backwards. Instead I went ahead and redesigned my combustion chamber enabling more swirl, reworked on the intake and went aggressive on fuelling. Meanwhile I bumped 10psi more static compression. My target was to run 100psi over a stock engine (stock Fiero/RTR is around 150psi) and no fancy fuel nor any fancy ignition. I am still running the stock CDI on this one however with the cam and other works. My head design mods worked the and the engine now handles such crazy compression with 97 octane. My head gaskets only last 3-4 runs in full blast
I have had blow-ups and what not with knocking in the last few years on many engines. I have ended up with melting pistons too
The best state of tune is when the engine is at the edge of knocking.
Today my street spec R15 runs 200psi and does not knock(runs on pump 87 oct gas). My RTR (air cooled) runs 195psi and runs on normal pump gas (87 oct) with a stock exhaust. There is a lot to tuning!

--SNIP-- -Edit: That was so wrong. See below.

And good to know you're an electronics guy. Being a software guy, I can get great things done, but even though I know some bits about electronics it a bit out of my league.

Hehe, you're madman! These are the things that really show commitment to what you're doing. Great stuff!

One question - Did you make any mods to the crank/bearing/con rods etc to support the increased power? Clutch plates/spring too? Basically the bottom end?

At lower rpms, we have higher vacuum. Hence the pressure readings will be low. Hence the 13 kPa. At WOT, there will be virtually no vacuum. Hence the 110 kPa reading. That is the absolute pressure reading not vacuum reading. And I think that is a NA engine!

So, we retard the timing at higher loads. Also, at idle it is the lower left corner I think!

Some more searching online, tells me you're right. My bad. Better I remove some part of the previous post I made, to avoid spreading confusion. Thanks, seems like I'm slowly ending up with a better understanding of what's going on.

Thats where the RTR/Fiero shines. I'm running stock engine bearing, con-rod, lighter crank, stock clutch, springs, gearbox.
If it was any other engine, it would have packed-up by now. The best part is, I have blown my Fiero engine long ago on my street spec Fiero 200 with some 230psi compression and the valves and piston was just mangled. The bottom was still perfect and all I needed to do was a clean-up and rebuild the top-end and work on the rest.

I remember you quoting on some other thread that ignition timing will need to advance with higher rpms. I hope you remember what I said.

Yeah, that I do!

Would you mind repeating it? I may not be very clear on what goes into engine tuning and all that, but I want to learn.

And I found this article seems to contain a good deal of basic info: http://autospeed.com/cms/title_Getti...2/article.html

Joel and I had discussed about ignition timings and the way it is advanced at higher rpms. But he said for the Apache, timing was not advanced much at higher rpms. The reason was that the port geometry and head design was such that it induced a great deal of swirl/turbulence in the air-fuel mixture at higher rpms, thereby speeding up combustion to a large extent as rpms increases. This showed that the Apache has a efficient head/combustion chamber design.

And nice article that one! Thanks for sharing.

So let me do short recap of what's come by in the last number of posts.

• Ignition timing depends on RPM and load.
  • Higher RPM, more advance.
  • Higher load, less advance.
  
  
• RPM is simply measured at the crank/flywheel.
• Crank position is not measured at TDC, but at some predefined point before TDC, like 45 deg BTDC
• Load can be derived from MAP or TPS, both are a reasonable indication. With MAP being more accurate of course.
• Ignition timing at high load (less manifold vacuum) needs to be retarded, because the pressures in the cylinder are higher and thus cause the flame speed to rise.
• More swirl in the cylinder also needs less advance since the turbulence causes the flame to spread faster as well.

From this is seems that if we have a micro-controller (uC), we could easily measure RPM and throttle position. The RPM can be measured by using a timer/counter that gets copied and reset at every revolution. It would count the number of microseconds in each revolution, it can easily be mapped to an RPM number. The TPS we can read simply by using an A/D converter, and making sure that the TPS provides voltages between say 0V and 5V.
So if we make sure we measure the TPS during the time we're not busy calculating the ignition timing, we should be able to setup a routine that takes a known amount of time to calculate the ignition time. Knowing that, we should also be able to calculate how long to wait before firing the spark. (Another timer could be set to provide accurate timing.) One could use software on a PC to precalculate the wait times, and not have the uC do the hard work, this would reduce it to a relatively simple program.

One thing I'm not sure about is if the internal oscillator of an ATmega uC will be stable enough with the heat it gets exposed to, it's probably better to use a crystal.

Of course the actual implementation would be a little more involved then what I noted here, but it seems doable. Making sure the software and hardware to run properly isn't really that difficult. It's not hard to build a small bit of hardware that simulates the crank signal and TPS signal. And if there's a scope handy it's easy to see if the pulses are coming at the expected times.

The only other concern is getting the timings right. Which has to be done on a real engine.

Yeah, and a 8 Mhz crystal is a must for this sort of precision job. And the recap is very concise and clear.

So what do you think of the idea for the algorithm, any inputs on that?

Well, step by step -

1. Wait for a signal from ADC1 (connected to crank hall sensor)

2. ADC1 reports a rising voltage which has a peak and falls gradually.

3. At a certain voltage level on the rising wave, the time is noted. Timer is started (Timer2). Again at the same voltage level on the falling wave, the time is noted. The mid point of these 2 times gives us the peak time. This is the hall sensor trigger time. Offset of this time to the timer start time is calculated.

4. The rpm is now calculated from previous time stamps. The rpm timer, timer1 is restarted.

5. Get TPS value. Look up is performed. Get then time to wait. Incorporate offset value from step 3.

6. Wait till timer reaches calculated value.

7. Fire ignition coil transistor/SCR

8. Stop Timer2.

9. Go to step 1.

I see. I was wondering for a moment if the ADC would be able to keep up, but it seems it's capable of sample rates of over 200KHz, which should allow for plenty of headroom.

This also assumes the signal from the HAL sensor is symmetric, which is should be isn't it?

The one downside is that it means you are constantly needing to read the ADC value.

As I see it, it's just a simple program that needs to be written. The logic for calculating the wait times can be done on a PC. Are you considering storing the timings in the EEPROM memory? That way you could simply upload a new set of timings, without having to rewrite the program memory. Or even go as far as having a serial interface that allows replacing timings on the fly while running.