Have you ever wanted to own an electric car? or have you ever wanted to drive a Porsche 911? I wanted both, so I decided to build one. This blog is here to document my journey as I convert a 2002 Porsche 911/996 Carrera 2 from a gas guzzler to a completely electric vehicle - not a hybrid!

Monday, March 25, 2013

Driving to work

I drove the car to work today, but ran into some trouble about half way.  The transmission failed into limp mode and stayed in limp mode, so I had to drive to work in 2nd gear only.  This limited me to about 45MPH at 4500 RPM.

I drove over to Imola autosports after work and they cleared the TCU for me which got me out of limp mode and allowed me to use the full range of gears again.  On the way home I also noticed the in-dash RPM gauge spiking from 2000 RPMs to 8000 RPMs and back down to 2000 RPMS for brief moments.  When this would happen, the transmission would release the current gear, I'd have to stop the car, put in park, turn off, then turn power back on, put back into drive before the TCU would re-engage any of the gears.

So, this is a show stopper.  I need to fix the RPM glitches or its not really drive-able.

I also got these error codes from the TCU before clearing it -- any input on how I should tackle these??

-K

4 comments:

  1. You noted noise induction was causing issues before - that'd be my first guess always. (Especially, as i"m sure you appreciate far more than me, anywhere analogue meets digital!)

    I assume you've not changed much (if anything) around how the Porsche monitors the wheel speeds though? The engine you will have had to.

    The "implausible" sounds like defensive coding in the ECU/CAN code - and it's observing spikes or wildly differing (i.e. 100MPH on the left, 10MPH on the right) speeds between the wheels.

    Can you get a read of all the values off the CAN? I know various people have successfully hacked up CAN sniffers in the past. Might give you some clues.

    Shame you've not got access to a rolling road or something where you can safely and easily test out runs while having gubbs of electronics hanging out of the door!

    I thought about you while ductaping my 1996 MX5 mk1's split radiator tube last week ;-) Enjoying the regular posting :)

    ReplyDelete
  2. Or, I suppose, it's vaguely possible the ECU and / or other components along the CAN aren't getting sufficient voltage, or a clean supply ... ? (Which I guess would classify as noise.)

    Perhaps the clock frequency of the ECU is getting stuffed by bumps on the input? Presumably all that is ultimately driven off the same supply as the new engine? (I.e., the battery packs.)

    You could probably rule that out fairly easily with a half-decent volt meter and/or revving the engine up while in gear vs coasting along in idle.

    ReplyDelete
  3. Can you verify that your arduino circuit is still receiving and sending a clean tach signal?

    ReplyDelete
  4. *Clean* is the key word there. The hall effect sensor input comes in very pronounced, and the output is as expected, however, there is a significant amount of noise on the input, which causes the output period to shift around, and there is additional noise coupling to the output, which I think is causing false triggers on the ECU. I did learn through experimentation that the ECU will report the correct RPM if I pipe in a clean square wave centered at 0V and 5V swing, so its a +/-2.5V square wave. The key learning point here is that it appears I don't need to replicate the speed sensor wheel exactly which had 58 teeth with 2 missing for a total of 60, it seems to work fine with a continuous pulse train of square waves.

    ReplyDelete