Annoying Saturn ECU Behaviour
 

I haven't posted in a while, but the contest announcement has inspired me to do so. :-)

For those who've followed my 2001 Saturn Mod thread, you know that I've installed kill and restart buttons. The restart button is for convenient restarts at lights (at speed, I use the clutch partial-release, press, release technique another poster suggested for smooth restarts with minimal wear).

Anyway, to get to the point, one problem I've had ever since installing the kill switch is that after using it for a day or so, the ECU starts to get the idea that the car is prone to stalling. What it does to compensate is keep the idle high, sometimes very high. I can be driving along at say 1800 rpm, and when I release the gas and push in the clutch to shift up or down, the engine revs up to 2500 or so. This works out reasonably well when downshifting, but it's annoying when upshifting.

I spent a long time scratching my head, reading various Saturn forums, replacing my TPS (although the readings I'm getting from Torque suggest all sensors are working properly), all to no avail. As a software guy, I finally realized the problem is a bug, or rather an unintended consequence of making the engine operate in a way the ECU designers didn't account for.

If I stop using the kill switch for a day or so, the problem goes away, as the ECU relearns that the engine is fine.

Has anyone else seen issues like this? Found any solutions?

First question: does it throw any DTC codes?

I don't think so. It certainly hasn't turned on the SEL (and I did have a different problem in the summer, P0410 caused by a clogged pipe, that did turn on the SEL; there were no other codes besides the one describing the problem). I'll scan it again when I get the chance.

Well that’s certainly no good, but I can imagine that being a result of making a kill switch... perhaps alternate between the kill switch and cycling the key?

The problem with cycling the key is that in the day time, the day time running lights turn off when I do that, so oncoming drivers think I'm flashing my lights at them. Cycling the key also disables the airbags for about 10 seconds, and there's the risk of turning the key too far and locking the steering wheel. Those are the reasons I installed the kill switch in the first place.

Turning the key to the accessory position will kill the engine without letting you lock the steering wheel. Not sure about it’s effect on lights or airbags. Kill switch is definitely safer though. :thumbup:

But as I said, there's a danger of overshooting the accessory position and ending up in the off position. Been there; done that.

Turns them both off on my car.

Just a weird thought. Could the IAC be partially open upon shutdown? This could be causing the high idle. ....
Try removing it, extending it fully then reseating it, then leave it unplugged.
Just a thought.

Does the ECU control the lights? How did you wire your kill switch? Assuming the lights are just on with the key on (through relay and whatnot) and you just kill power to the ECU it should turn the car off while leaving the lights on and an electrical circuit will keep you from overshooting the acc. position. It doesn't solve the airbag problem though.

I'm wondering if there's a push-button way to simulate momentarily moving the key to a lower position that will kill the engine.

I have an L not an S series however I have noticed when coasting out of gear my idle stays high based on vehicle speed. It stays at 1500 or so above 30 mph. As I coast down to stopped the idle returns to normal. I have also found if I kill it and restart it while rolling everything comes back except the cruise control. I do alot of fuel cutoff decel but very little engine off coasting due to loosing the power steering in this big heavy car. When I drove a Metro the engine was off quite a bit on my commute.

Should be able to cut the power/ON wire going to the ECU...it won't think that the engine stalled this way, and should keep normal idle after restarting.

Don't know what drawbacks there might be from doing this.

Thanks all, for the replies above.

I'm pretty sure the IAC isn't the problem, since the high idle even happens during later gear changes, not just immediately after a kill-and-restart. The ECU would have reset it to where it wants it long before that.

I don't think the ECU controls the lights, but the same switch position that shuts off the ECU shuts off the (daytime) lights.

The kill switch is wired to a normally-closed relay, whose contacts are in series with the +12V feed to the fuel injectors (very easy to wire in, as my circuit just plugs into the dedicated fuel injector fuse's slot). The ECU isn't "aware" of this mod at all. As far as it's concerned, the engine just stalled for no apparent reason.

I guess I could kill the power to the ECU instead, but as you've mentioned, that doesn't solve the airbag problem. It also means that Torque will disconnect from the OBDII port, since the ECU will have vanished. I also wonder how the ECU would respond to waking up in a car that is already in motion.

One other data point. When the car is in this high idle state, if I stop at a light and don't turn the engine off, the idle eventually settles down, dropping to about 1000rpm (which is still a bit fast) after about 30 seconds of high revs (which feels like forever, and burns a lot of gas).

I’m definitely going to watch this thread, for it stands to reason that if you’re getting a high idle with the kill switch in your S series, I could end up with the same unintended result in my ION...

thought about this for a while: I think the problem is that the CPU is looking for a tach signal which it won't get with the engine dead and not spinning. Obviously, empty cylinders/ engine compression during coast will act as a coast retard. So: your kill relay will need to add some sort of PWM signal to the ECU tach input that looks like the engine is running at some RPM at or above idle. the ECU should tolerate some stumble or momentary lack of signal, and with throttle off during the kill process just think it is coasting. I know the process in my F250 is: ecm doesnt care what rpm the engine is at during coast, it doesn't add fuel until idle drops to about 1000 rpm (because it "Romps"), taken from the TDC cam sensor. perhaps a 555 timer circuit running as an oscillator?

Yes, put a piece of vacuum tubing and a drywall screw to expand it in the idle valve intake port, its the small hole (5/16 dia) on the bottom of the throttle body bore. This will lower the idle. You might get an error code, I just ignore it. May need to give it some throttle after cold start till it warms up a little.

I guess what I really need is something whose speed is triggered by the fuel injector pulses as soon as I turn off the power to the fuel injectors. Something that pretends to be an engine based on the fuel the computer thinks it's feeding the engine. I'll need to do some diagnostics to determine the relative timing between the FI pulses and the tach signal, and then whip something up to fake this out.

How much you actually save fuel using a kill switch?
Is it really worth the hassle?

I'd rather all Saturns came with DRL kills.

In city driving, probably a lot. It's less clear cut for my commute, but there are about 3km (out of 45) during my commute where I don't need the engine. With the engine idling, I burn about 2.5L/100km on those parts. With engine off, 0. Averaged over the whole trip, that's about a 0.17L/100km saving. Averaged over a year of driving, around 18L of gas saved.

EDIT: That 18L/year is assuming I only do my commute (rural highways). Once I throw in the city driving, the savings become much more.

You need a dpdt relay and a 555 timer, the board can be bought on ebay.

Switch the fuel pump relay and a relay to the crankshaft position sensor. Switch off the fuel pump and the CPS and switch on the oscillator to the PCM CPS input.