Characterizing the VW ECU

Goals:

Obtain a fuel map from the original VW ECU as a starting-point for ECUality
Answer questions about how the original ECU behaved, such as:
- What aspect of the O2 sensor signal does it care about? (threshold crosses/sec? DC voltage?)
- Generally, does the optimization routine change the "shape" of the map, or just apply a global offset?
- How fast does the optimization respond?

Here, I use an Arduino Leonardo to send signals to the original ECU that simulate sensor data from a running engine. With that stimulus going in, I record the pulse-duration of the injector signal coming out.

The breadboard between the Leonardo and the ECU contains:

a couple RC (low-pass) filters for smoothing PWMs into analog voltages.
a couple N-mosfets for handling a voltage-shift (0-5V to 0-12V)

Discussions

Mike Thielvoldt wrote 03/20/2015 at 08:04

An engine fuel-map:

This was obtained at an O2 input of .5V, which I found to cause the injector duration to remain constant. The air variable represents the 8-bit pwm value that was used to simulate the 0 - 5V signal produced by the air flow sensor. 255 would correspond to 5V. 0 to 0V. Linear in between.

The left column is the air variable,

The top column is the RPM variable.

the block of data in between is us/4 of injector duration. (to get microseconds, multiply these numbers by 4).

RPM:       500   695   965   1341   1863   2589   3589   5000
air: 220   2470   2470   2472   2464   2471   2464   2459   1917
air: 180   2162   2170   2160   2160   2333   2021   1597   1204
air: 140   1900   1898   1918   1896   1632   1236   943   765
air: 100   1789   1750   1618   1228   948   755   616   596
air: 80   1518   1498   1177   941   749   622   588   563
air: 60   1294   1291   944   773   636   637   587   560
air: 40   1164   1160   779   776   678   648   587   561
air: 20   1123   926   786   826   719   631   900   589

Are you sure? yes | no

First Fire

Discussions

Become a Hackaday.io Member