Subpage of MembersPage/MarcellGal/PowerAndTraction, the implementation of PlugAndPlayMotronic on Audi 200 quattro 20V turbo
We used 6+2 style ignition output (6 power IGBT-s + 2 logiclevel ignouts) for all PNP ECUs (and econoseal ECU-s made for PNP-adapters for these 10VT/20VT enginse) so far. This way, using the 4 stepper outputs, there is a quick migration between using
- 5 logiclevel outputs: 4 stepper + 1 ignchx (audi power output stages work with 510 Ohm but using stronger than standard: 510 Ohm/2 protection resistor might be a good idea to allow more picky coils, like the webshop VAG COP to work, though that actually seems a few mm too big to fit into the 20VT head )
- 5 power outputs: to leave out the power-output stages and drive the coils directly from genboard
- 2 x AMP 3pin male and 2 x AMP 4pin male allows a clean reversible PNP bypass of the 2 power output stages that drive the 5 coils.
- note that it is extremely important to use proper fuses for protection. Coil consumption is appr 1A/cyl at 9000 RPM, so using 2..3A/cyl seems good. Eg. 3A for COP, 4A per 2 cyl when using wasted spark coilpack, or 7.5A if you have only 1 supply pin for a 2x2 coilpack. It is not a too good solution to use 10A shared by all 5 coils: try to split to as many parts as possible !
2+6 might also make sense (2 lines to clamp instead of 3) if one is certain he'll use logiclevel coils/modules (the ones factory setup, or other) later as well.
The differences between
- AAN
- we had a miswiring (project delay) from a circulating xls that swapped ignition control lines for cyl1 and cyl3. Took some time to find out, but was very easy to fix in configuration. In fact, on AAN, spark control (logiclevel ignition outputs go there) cyl1 is motronic55pin1 and cyl3 is motronic55pin23. Maybe there is a difference between AAN and ABY with these pins ?
- in the AAN, sensor GND was motronic55pin30, VR-trigger GND was motronic55pin48 (for both primarytrigger 135 tooth and secondary trigger crankhome-VR)
- 3B
- custom harness based on Econoseal-36 + Econoseal-18
- or small adapter from EC-xx to AAN or 3B
However, there are a few details that is nice to have.
Tach freq-signal (AANpin40, 3Bpin6) to drive the factory cockpit tachometer gauge is simple:
- some used 0/12V stepper outputs with success
- (others reserve stepper outputs for ignition =>) a p259 output with 510 Ohm or 1k pullup to 5V (EC36pin4 has onboard provision for a 0805 component!) also verified to work (with tach_divider=0x10 reading seemed correct). Probably a weaker pullup (eg 4k7) would work just fine, we didn't try
- this means that a standard logiclevel ignch_x should work perfectly too
Trip computer boost display
- now AAN motronic55pin32 connected to s259pin5 via 1k*10uF RC filter.
- Sami configured iac_sol_channel=16 for this (s259-1) and found it linear: mdi50 (80/256 duty) => 0.9bar boost; mdiff (~100% duty) => 2.5bar boost:
Vehicle-speed AAN motronic55pin50
It would be awesome to log this. With only RPM, MAP and TPS (vs time) in logs, it involves heuristics to find out the gear. With vehicle-speed data (either from this signal, or from a GPS which has elevation and climb data as well), road-dynoing becomes easier to automate.
- Emil and Andrey suggested to capture the vehicle speed with the free input signal when PS2 not used.
in the Audi 100/200/S4 ? Likely the same in the 80/90 chassis.
Trip computer fuel-consumption display
Now AAN motronic55pin31 connected to s259pin6 (s259-2 that is GenBoard/Manual/DigitalOut/Table 0x26) via 1k*10uF RC filter.
The manual shows that a potentiometer<variable resistor> described as a fuel consumption sensor in the Fuel distribution head/flow meter measures the position of the plate, just like a TPS. It's a 3 wire sensor that connects directly to the gauge for its power, ground and signal
The gauge cluster has some adjustment for fuel consumption trim between 0 and 15% is described in the manual using some switches or a resistor on the cluster itself.
I can take measurements of the Sensor with the throttle plate closed, at idle and at WOT if they are useful. => not really useful.
- Better to map the fuel consumption the same way with iac_sol_channel and mdi..
- note that vehicle speed is needed for this (that the dash gets directly from the gearbox)
- so Marcell cannot measure this (his dash shows intermittent unreliable vehicle speed)
- Be careful when mapping this during driving ! Take care.
Speculation and experience
MembersPage/EmilLarsson knows the bmw's do some strange calcs inside cluster, for fuel consumption: the gauge starts moving when hooking up tachometer.
MembersPage/SamiKorhonen : some guy used 0-5V signal what his old KE-jetronics MAF gave to his computer.
In fact the tach signal, and a 0..5V signal should be all that is needed. We just have to find out if the signal is rather proportional with MAF or MAP or injector-pulsewidth, or something else ;-)
CLT and MAT on an Audi AAN (20VT) engine
Measuring against motronic pin 30 (sensor GND, mislabelled in our s2central doc), our DVM can see the CLT (pin45) and MAT (pin44) sensors from the motronic55 connector. Since the sensors are grounded through ECU, using other ground pin for measurement will not see the sensors.
CLT (coolant) is measured at a few points
- rear end of the cylinder head => sensor for the ECU
- water manifold sensor with 2 blade connectors
- water manifold, other sensor (AMP type ? should look around in photos, hard to see after mounting. Maybe a small camera ?)
- coolant radiator bottom => this switches fan, not related to ECU
- we also see a (likely temp) sensor (with 2pin AMP connector) in a thick (appr 30mm) tube, probably hot coolant for cabin heating system
It is strongly recommended to use lower MAT pullup resistor value than the factory 2k7. I used 1/(1/2700 + 1/510)=429 Ohm (430 Ohm would be perfect as well. If your ECU is clamped, you can change by removing endplate only, or use an 510 Ohm pullup to 5V outside that becomes effective 429 Ohm together with the internal pullup, as the above calculation suggests). This "stronger pullup" gives appr 2x as good resolution. I uploaded my PTC MAT sensor calibration to EasyTherm/SensorTable
From 1.1.29 the firmware uses 11 bit sampling internally, so the airden and MAT reading is smoother for the steep PTC curve.