DIYEFI.org Forum

my total time is 1.6ms (from the logs), and the opening time is only 0.6us. So, 10us is 1% at most. What are your open times?

I'm not sure the errors stack the way you think they do.

None the less, high accuracy spark at extreme power levels are more important than going from 1% to 1/10th of a percent fuel accuracy at idle, which will have much larger errors anyway.

You know, in case we want to build a perfomance motor.

Trust me, there is nothing I care about more than squeezing [dr evil]One Million Horse Power[/dr evil] from a 1.0 litre 3 cylinder daihatsu with bad compression on two cylinders :-)

Everything will be as accurate as possible, but reality (whatever the calcs that lead up to it, and whatever physical phenomena there are) means that fuel is more important.

Perhaps gearhead can chime in here, I know he has done some measuring of the actual accuracy delivered by MS2 on the ignition front.

Admin.

All that is not to motivate which is more important, I don't want to "win" that fight. But it seems a lot more effecient to bitbang all the fuel delivery, so that strange engines don't become special cases.

As has been pointed out before, those individuals with enough skill and a strange engine will be more than welcome to knock out corresponding strange code. It seems like shooting the foot for no reason to code everything for the lowest common denominator when at the same time you reduce everyones efficiency.

I just measured the first half of the runtime of a very basic ISR (see diary thread) at 150us

double it for the other half 300us

8000rpm
4000cycles/m
65cycles/s
each cycle has 6 injection events
each injection has a start and end
700 ISRs per second from a max of 3000 with nothing else running. Add in the 240 interrupts per cycle for wheel input, and...

Starting to get the "its not fast enough" picture? :-)

Admin.

http://www.diyefi.org/forum/viewtopic.p ... p=574#p574

cough-sorry-cough

Still, I don't see the point in sacrificing accuracy on fuel for spark. Having said that, it looks like it will be about a factor of 8 difference in accuracy between the two styles which is still potentially a good thing for trim on V8,10,12 engines if they want to run sequentially.

Interestingly, I *think* gearhead said 10us jitter in the spark outputs on ms2e, which looks about right with our 6.4us and 40MHz vs. their 24MHz.

EDIT : At 8k RPM one degree timing is 20us so we have 0.5degree timing accuracy (or better) using that bit banging. we *may* be able to go more accurate with some cunning padding etc, but I think that is a feature best saved till much later :-)

Admin.

Good, good. 1/2 degree is awesome, if it's at all consistent.

So, how much accuracy, in time, do we need for fuel? Let's throw some realistic numbers at this, since by my set up, 10's of microseconds is fine.

It's not just the accuracy, it's also the jitter.

if you think its at a 10us mark, and its actually varying 10us each side of that, thats 20us error, when you are only opening 100us on large injectors for a small motor, it becomes an issue.

Try to give me an idea of these 100 us opening injectors. The only point of reference I have is my own 550cc/m injectors on a 450cc cyl, with two squirts I still get ~1ms opening times, and that's at 18:1 AFRs.

I think 100 us is unrealistic. For my motor, that would be a 5l/m injector.

8InchesFlacid wrote:Try to give me an idea of these 100 us opening injectors. The only point of reference I have is my own 550cc/m injectors on a 450cc cyl, with two squirts I still get ~1ms opening times, and that's at 18:1 AFRs.

I think 100 us is unrealistic. For my motor, that would be a 5l/m injector.

What is your opening time? Is 1ms what your log shows? That log includes opening time AFAIK. I think you are the example :-p

Besides, a high performance app might have 1000cc injectors on a 500cc cylinder. That is not uncommon (in "tuning" circles) (and if you are ever reading this "Jim" please absorb it)

Admin.

My injectors report ~1.6 ms at idle, with a 600 us opening time.