MembersPage/MiskaPeippo/AudiTQ

I'm putting MSAVR working on my Audi 5kTQ -86 with 2.2L 10 valve turbocharged inline 5 (motor code MC1). Some notes and information about the project will be collected here.

Wiring information is collected to MembersPage/MiskaPeippo/AudiWiring page. (This actually is a bit misleading, will fix it soon.)

I'm going to create first version of my ignition map based on [original ignition map] from MAC-11 ECU.


Fuel injectors inserts

For Bosch style electronic fuel injectors to MC cylinder head (and other CIS heads I guess), you find part #037 133 555A (used in mid 90's VW Golf) from local VAG dealer.

037133555A.gif

Plastic insert has same threads as original CIS injector inserts and fits directly to the head.


I think it's time to start keeping a project diary here once it's finally looks like I'd soon have some time to start install of GenBoard/VerThree to my Audi.

05-06-2004:

I received a spare ECU to be sacrificed and reused for stock sensors and wires where applicable. Since the car has mechanical injection system, new wires to injectors etc. are still needed.

06-06-2004:

GenBoard fits almost perfectly inside old ecu case. Some free space is also available to put InputTrigger/AudiTrigger board and wiring to connector inside the box.

09-08 .. 19-08-2004:

Soldering wires, testing trigger functionality, trigger problems, false triggers

I interfaced 2 original VR sensors from my engine to LM1815 circuits (some info about timings and circuit itself in InputTrigger/AudiTrigger) and everythinhg worked perfectly on the table with clean signals (from few Hz to >15kHz). However when I tried the circuit in with real sensors in car while cranking, the first impression was that I was suffering from some false triggers.

Firmware in the trigger board stored tooth number when TDC trigged occurred to eeprom. Checking the numbers showed that tooth number is drifting slowly (3..4 teeth/crank rotation). I had unshielded wire from original ECU connector to triggerboard (appr. 20cm) and I suspected some noise spikes causing some extra triggers. I changed to shielded wire, but it didn't help.

The next thing was that I made the leds blinking and checked how often the triggers are seen and it revealed the truth: TDC trigger (single pulse / crank rotation) gave false alarms about after every 3rd tooth!

So there had to be enough noise coming from sensor when engine rotates and because of only one pin is generating some real signal, circuit working in adaptive mode (pin 5 of LM1815 unconnected) soon finds something where to catch. Without a scope I couldn't verify the exact signal strength from the sensors.

The next thing was to try a fixed arming threshold instead of adaptive (pin 5 to VCC). Cranking test and eeprom contents showed a regular pattern (cam, tdc, tdc, cam, tdc..) with same tooth every time, so it looks like working now, but I'd still like to check the sensor signals with scope.


I'm using original distributor based ignition. Coil is driven by original power stage (huge transistor within coil). To control the transistor, I need either one extra PNP-transistor to feed positive pulses out or use one injector fet-driver channel. I decided to use the spare driver (5 out of 6 in use).

Also I needed to patch edis.c to get ignition signal out from non-standard place:

ign_activate(void) {
ch = ignport_map(enignct);
...
#ifdef ALTERNATIVE_IGN_OUT_TEST
    digitalout(ch, DHIGH);
#else
    I259(DHIGH, ch);
}
#endif

Deactivation is done respectively.

My h[] table looks like this:

# alternate injectors inj0..inj4. inj5 is used for ign!
h[0]=04 10 08 02 01 20 00 00

h[1]=00 00 00 00 00 00 00 00

# ignition sequence
h[2]=50 50 50 50 50 00 00 00 # ????
This sounds bad, h[2] values should be between 00 and 07

So the ignition signal is now getting out from h[0,5] = inj5. Car has now also been running for a short time with VEMS V3.1 board controlling ignition.