How much can it take?

My good friend Jesper Karlsson has helped me abuse one of our v3.1 boxes in his Mercedes 190 2.3-16. The car is normally aspirated and is producing around 230hp at this time. That is around 50hp more then stock and most of that come from the VEMS install. This car is very special as the stock Bosch K-jet really limit the engine. On most normally aspirated cars you would be VERY lucky to gain 10hp by only fitting a VEMS v3.

As most people choose to install the ECU inside the car as recommended the abuse started by installing the ECU next to the battery in the engine compartment, the well veltilated area has rain water running through it and the battery supply adds some acid fumes to the air. The winter here in Gothenburg has been perfect for this type of testing with 5 months of ice, hail,rain,snow, salt and generally shitty weather. The car has been driven hard every day to and from work and on all work related trips during this time.

Over this time the car has only experienced some trigger problems that made us even more sure that VR triggers are highly unsuitable for use in cars. There has been occasional false triggs which has given the engine a rough note. The P259 chip that is used for the fuelpump and a few other outputs was killed during installation after some confusion with the grounds. This chip died silently and left the fuelpump on. We left it that way for the entire test period. During the startup of the car there was an episode with the ignition coil, the MB stock coil had VERY low resistance and almost no inductance. Just like a high performance CDI coil. After missreading the stock schematic for the car it was assured that there was a fuse for the coil and we reused that. After running the car for a while in the garage it suddenly stopped and after opening up the ECU a terrible smell spread in the garage. The trace from the IGBT had blown up, we quickly cleaned it up and added an airwire. Now we found that there was no fuse for the coil. After adding one we found that even at 0.2ms dwell or so we blew a 15A fuse instantly. Some application of ohms law told us that the current the IGBT and board saw was somewhere between 30 and 40A. We now know that the IGBT's are extremely rugged.

From the testing we have found that placing the ecu next to the battery is as smart as it sounds. The EC connectors was heavily corroded inside, (remember to plug the unused connector positions). The PCB inside was also very corroded and it started to behave unpredictably. After we had scrubbed the PCB with cheap electrictronic cleaner and a fine brass brush everything worked well again. By cleaning a section of it with Chemtronix Electrowash PX we found that you get what you pay for, the Electrowash PX washed away the oxides without us having to use the brush while the cheap electronics cleaner needed several minutes and help from the brass brush to clean the board.

We had now fooled around with the VR trigger and applied all the filters and corrections that we could come up with and we still didn't like the way the engine ran. It was a bit hard to know what to expect as the engine has pretty hot cams but Jesper reworked the trigger install and I changed the trigger input to accept the signal from the Honeywell 1GT101 sensor. As usual when fitting the 1GT101 we heard that the engine was totally different even after a few seconds of idling. The engine is now very civilized considering the cams and it also runs stronger on all rpms.

With the engine running properly we found that the VCC on this v3.1 board had minor oscillation on the vcc lines. We plan to add capacitance in the 100-500uF range to cure this. This problem comes from the high precision voltage regulator used on the 3.1 boards. It holds the exact right voltage but it is more sensitive to oscillation and needs MUCH more capacitance then the voltage regulators used on the v3.2 and v3.3 boards.

As the car was now working nicely we decided to stirr it a bit by upgrading to 1.0.36 firmware and by removing the broken P259 chip. I also cleaned it up and tested the new Electrolube ULS UltraSolve? cleaner and the Electrolube HPA (what?) conformal coating on the board. The HPA coating is easily washed off with the ULS cleaner.

The upgrade went far from smooth, mostly since I was replacing subframe bushings on the car I was going to use on a combined FIA racing license course and trackday which started 9 hours later. In any case we forgot to move the fuelpump to one of the injector fets, this prevented the car from starting. The instructions I yelled from under the car (damned bushings!) failed to mention that a new megatune was needed for the new firmware in the ECU. This seemed to upset the configuration quite a bit as uploading the config.txt and tables.txt didn't fix it. This was late night logic as we found later. The battery had started to go flat and we connected the charger/start help. A terrible racket emerged from one of the relays in the car, we could trace it to the fuelpump relay that was now connected to one of the injector FETs. It was obvious that this couldn't be good but as it's likely that someone else would fuck up like this we let the charger be connected for a few seconds before we started to search for the problem. Unfortunately the time was short and we couldn't find anything in the middle of the night. 4-5 test applications of the charger later we gave up and I started changing tires on the car I was going to the track with.

It turns out that the charger must have killed the I259 logic chip, the only possible reason for the oscillation of the relay seem to be that the transient protection diode snubbed the incomming high voltage for each pulse from the charger (100Hz)but wasn't fast enough to turn off before it had removed all power from the input of the ECU which made the ECU reset (and turn off the fuelpump relay). When the diode recovered it was soon tripped again by the charger and it started over again. So far the only damage from this abuse seem to be the dead I259 chip and the car is now back on the road again.

Where was the charger connected ? To the battery ? Is there a chance that the peak voltage at the ECU supply was above 18V ? It normally shouldn't be higher than 15..16V.

Jörgen Karlsson