Audi 80 1992 1.6lt

[settra]

hello forum.
my goal is to convert my carbiratur audi 80 (with computer controled ignition) , to Electronic fuel injection.
i will use parts from a latter Audi 80 , which had fuel injection, BUT i want to diy the ECU. i will use, an arduino microprocessor, an other various Atmel Chips... ( for start, i will be controlling only the EFI, not the ignition)

i will be starting next month..
if some one has any suggestions,or heads up, i will be glad to hear , as i have no previous experience!!

[kb1gtt]

$14 stem discovery f4 is much faster than arduion

[settra]

so, my car has only ONE sensor, relevant to my engines speed... one, that is located at the distributor...
i was able to monitor the distributors signal, by manually turning the engine, by the camshaft....
turns out, that , the signal is constant 12v (or you can invert it. same results), AND goes LOW, 4 times, inside every camshaft rev..
so 4 pulses, equal cam rev, or 2 crank revs...

so... by that signal alone, i suppose i cant get EFI to work, becouse, i will know, that one cylinder is before TDC, BUT , i will not know, which one... correct????

i will need to figure a way, to know, the 1st cylinder's TDC seperatly??? any good way to do so???

[AndreyB]

You said you were turning the engine by hand.

I suggest you get a cheap logic analyzer and record your signal while the engine is running at constant speed. There is a real chance that the widths of these four signals is different and by noticing the difference you would know which cylinder is which.

You said you have a distributor? I guess that means stock ECU does not need to differentiate the cylinders so maybe this would not help, but that's worth a try.

Also what about slightly newer audi's - can you take some parts from a newer audi which would maybe have more sensors?

[settra]

yiap. i could use a logic analyzer. maybe the widths are different. that is a good idea. the widths where very "big" btw...
the Stock ECU does not need to differentiate the cylinders. it just produces the spark on the ignition coil. the spark, goes to the right cylinder, by the distributor!! (so there is a big change the pulses are the same).
i was thinking, of building a disk, with like 36 small wholes, and place it, on top of the cam pulley. and use a led , optical sensor, to count the wholes, (each whole would add+ 10 degrees etc), like the newer car's do...

the problem would remain, to separate, the start of the cycle. but i could do that, with maybe, another disk, with just one whole..
(i know, that in OEM ecu, this is done , by one missing tooth/hole, but i haven't figured a way, to make this happen "Reliable" code)

[AndreyB]

but i haven't figured a way, to make this happen "Reliable" code

http://sourceforge.net/p/rusefi/code/HEAD/tree/trunk/firmware/controllers/algo/trigger_decoder.c

[settra]

you are forgeting, i wount be using stm :/...

[AndreyB]

you are forgeting, i wount be using stm :/...

Which line of this file says 'stm'?

[AndreyB]

I do not want to sound disrespectful, but it might be worth considering stm just for the fact that this code is available for stm. No pressure Maybe just a little...

It's just that the whole ECU is a large project and in order to produce something good we need to join efforts.

[settra]

arduino code is way different... believe me i considered it, but i cant simple jump microcontrolers so easy i have a certain lack of knowledge, on the engine part arledy, and some on the electronics too... cant boost it, with a new microcontroller which i am not comfortable with (remember, i plan to build ALl the electronics of the ECU...)

I cant understand how its done , on the code you posted, but , i know that, in general, the ECU , counts pulses, and their duration, and when it sees a "longer duration than ussual" , it supposes, it has lost a teeth, and starts the new cycle..
but what longer duration actually mean?? double duration between signals?? is it "safe" enough that way??

[hasse.69]

I suggest you get a cheap logic analyzer and record your signal while the engine is running at constant speed. There is a real chance that the widths of these four signals is different and by noticing the difference you would know which cylinder is which.

Or this one , half the price:
http://www.banggood.com/USB-Saleae-24M-8CH-24MHz-8Channel-Logic-Analyzer-Latest-Support-1_1_15-p-74101.html

[AndreyB]

I cant understand how its done , on the code you posted, but , i know that, in general, the ECU , counts pulses, and their duration, and when it sees a "longer duration than ussual" , it supposes, it has lost a teeth, and starts the new cycle..
but what longer duration actually mean?? double duration between signals?? is it "safe" enough that way??

http://rusefi.com/docs/html/

Yes you are right, it's the issue of finding the right coefficient - on the one hand teeth have different widths, on the other hand RPM is not constant. I have this coefficient configurable via tuning software and I guess the only way to figure this our would be trying it on a real car.

[settra]

mm... well, asuming that all the teeth will have the same distance from one enother, then , if you say something like :
pulse distance = 1.7 than previous, then the problems would be :
if the car engine decelerate so fast, that the rpm drop in half, (no change this is gonna happen)
if the car revs up so fast, that even though the teeth is missing, the pusle distance, becomes the same with the rest (which i suppose is like "doubling the rpm in dt time. ) and can happen during cranck??...
i think , that having enother sensor, that pulses once per cycle, is easier... >_>

[AndreyB]

I do not have the answer to this yet. I am sure that you cannot go from 3000rpm to 6000rpm in one revolution - and we are interested in one revolution, but cranking is interesting. I am planning to gather some data on this.

[settra]

cool