DIYEFI.org Forum

I've put this thread up because I thought of a couple of cuts that SHOULD be performed that aren't the norm recently.

Obviously there is RPM based fuel and/or ign cut. These shouldn't be able to be turned off, just raised up to a level that won't get hit.
For boosted apps there is excess manifold pressure based fuel and/or ign cut. Likewise, always there, just tunable to an unreasonable level.

Less settings, simpler, better, easier to understand and diagnose.

But additionally:

If duty cycle exceeds some percentage, then the engines operating conditions are no longer under control, we're playing Russian roulette with your engine's AFRs, which is BAD in a boosted application. This hysteresis could be done on time, or on requested duty or perhaps something else. I'd quite like to make this compulsory such that anyone with a marginal setup is forced to upgrade their equipment (or just increase the pressure a little) to avoid hitting it. Allowing engines to run leaner than they want to will allow engines to blow up. I'd rather keep the number of user engine deaths down in situations where they could turn around and try to blame us. With this, they would have no choice but to acknowledge the deficiency in their system and correct it. I personally know people that have cooked engines running them on 100% duty. I myself ran my truck on 100% duty for a long time, but knew the risks, and ran with it. More ignorant users may not even be aware that they have a fuel supply issue unless you forcably bring it to their attention. At the very least, the default should be ON, and MAYBE the ability to over-ride it (without code change, someone determined could always disable in code...) could be provided in some advanced tab or similar.

If sync loss occurs, and is regained too quickly, then you can get the case where some fuel events are missed out, and others aren't, ditto for ignition. This can create a lean condition without a hysteresis, and I think it's unreasonable to leave it to the decoder authors to handle this, it's a common-to-all-decoders issue. in the event that you fire more than one injection of fuel per cycle (all non-sequential applications) then you can easily go lean. Say you're running a super rich 10:1 AFR in boost, and you miss one of your two injections in a cycle due to brief sync loss, you'll get a lean, but (barely) combustible mix of 20:1 in the chamber at massive pressure, and quite possibly damage the engine. The situation is worse if you have 4 injections per cycle and miss only one, then you might go from a healthy 12:1 to a devastating 16:1 which will definitely ignite and explode and destroy something. Such hysteresis could be done with time or with engine cycles (using last known RPM and time). This one should not be able to be switched off, as it's a design thing, not a user setting. The time limits should also not apply to initial sync which should be instant if possible. Only to "was running, now isn't" situations. We could have a minimum level that users cant go below, with a more aggressive default to alert users to their issues. That way they can make it more benign if they want to live with it, but can't put themselves in danger of lean conditions due to part cycle sync loss.

These have been thought through pretty well, but feedback is still welcome. More so, though, what ELSE needs a similar treatment?

For all of these, it will be done through scheduling or not based on conditions. IE, for the sync loss one, the decoder could regain sync, but no outputs would be fired due to nothing being scheduled. This is nice and clean.

Fred.

There is an example. This is setup for Sean's LT1 V8 with Semi-sequential injection, the engine would receive 50% fuel on 4 cylinders for one cycle each. Not good in boost.

You can see the missing ignition pulses too, but they don't hurt, really. The joined together ignition pulses are just a function of resolution of the log and decreasing non-dwell period at higher engine speeds. IE, not real.

Fred.

What about wall wetting contribution? When you do a fuel cut, shouldn't you perform some sort of enrichment when resuming injection similar to an acceleration enrichment and for the same reason?

Also, are you sure that you have an issue if you go really lean? A combustion lean of peak will be colder than peak (a bit lower than stoichiometric, if I remember correctly) and simply have less power. There is this misconception that running lean will always result in a melted engine but this is only true if you always stay on the rich side of the peak (or close to peak on the lean side). If you lean it out far on the other side of the peak, you simply lose power AND lower combustion temperature.

Jean

jbelanger wrote:What about wall wetting contribution? When you do a fuel cut, shouldn't you perform some sort of enrichment when resuming injection similar to an acceleration enrichment and for the same reason?

Probably a good idea, but such things are currently not handled at all! And, amazingly, all engines so far have responded very well without it. Who ever would have thought. Oh, wait, me! :-) Don't tell motofab, though. When we get around to adding transient corrections and/or wall wetting stuff, it can be handled at that time. The same thing would need to be done for rpm cuts as well, so I don't really see how it's relevant to the specific topic (items requiring a cut, not how to handle post cut conditions), but it's not out of place either, really. One could argue that the appropriate thing to do would be to use post start enrichment again, as it's been completely dead. Let's start a distinct thread if we want to get into such semantics, though. This one should stay nice and narrow.

Also, are you sure that you have an issue if you go really lean? A combustion lean of peak will be colder than peak (a bit lower than stoichiometric, if I remember correctly) and simply have less power. There is this misconception that running lean will always result in a melted engine but this is only true if you always stay on the rich side of the peak (or close to peak on the lean side). If you lean it out far on the other side of the peak, you simply lose power AND lower combustion temperature.

All true. There are two issues, heat is one, detonation is the other. In any case, with the two latter cases (large paragraphs) you can't be sure exactly where you'll end up (AFR wise), and therefore it could be in a bad place. Hence I believe such circumstances warrant a cut. Some oems do an AFM based cut in a similar way and likely for a similar reason too. If the AFM is maxed out, any amount of air could be flowing through there, and thus the ECU doesn't know, and thus the car is vulnerable.

Fred.

Fred wrote:There are two issues, heat is one, detonation is the other. In any case, with the two latter cases (large paragraphs) you can't be sure exactly where you'll end up (AFR wise), and therefore it could be in a bad place. Hence I believe such circumstances warrant a cut. Some oems do an AFM based cut in a similar way and likely for a similar reason too. If the AFM is maxed out, any amount of air could be flowing through there, and thus the ECU doesn't know, and thus the car is vulnerable.

Agreed about the two issues and that not knowing where you end up is bad so cutting for a full cycle (or multiples of it) is a logical solution.

However, I'm not sure that protecting a user from his own stupidity is really the job of the ECU but if you feel strongly about doing this, I don't see a real down side to it (other than having to implement it and maintain it). Natural selection is not a bad thing...

Jean

For me it's mostly about software correctness - ie, if i don't know whats happening, i should have a conservative output - nothing. There is the secondary issue of reputation damage by idiots ruining their cars by their own hand and then blaming the box.

As for full cycles, for a second I thought you were right! Then I realised that if you're delivering two injections per cycle, the implication is that you're delivering that to two oppositely timed cylinders. Thus if you start firing on one, then one cylinder gets a full breath of fuel, and the other gets half, and if you fire on the other it's reversed. Which raises the question of if we should even bother, as we're still going to get a single cycle of non-optimal fueling at the beginning of scheduling starting again. The answer is yes! And the reason is that with sched loss you can get a repeated loss situation and feed it a consistently lean mixture to some cylinders, normal to some, and nothing to others, rather than just a single cycle at the beginning (which if aggressive enough would mean turbos had lost spool etc too). Thanks for being a good sounding board, though :-)

Fred.

Good point that you can't stop fueling for a full cycle for a specific cylinder if you don't know where the full cycle starts for that specific cylinder which is the case for anything other than full sequential. And I agree that it shouldn't stop you from doing what you can even if there will be one cycle with potentially lean cylinders.

I was thinking that you could keep a timer of the last sync loss (or whatever triggered the fuel cut) and increase incrementally the duration of the fuel cut if the condition reoccurs before some timeout. Again this might be slightly off topic but may be something to think about.

Jean

jbelanger wrote:I was thinking that you could keep a timer of the last sync loss (or whatever triggered the fuel cut) and increase incrementally the duration of the fuel cut if the condition reoccurs before some timeout. Again this might be slightly off topic but may be something to think about.

Sounds like a nice potential feature to keep in mind. I don't think it would be a desirable thing on intentional cuts like the rev limiter, though, as many of us like to bounce off of those for extended periods for various reasons :-)

If you did that with sync, and it was a fixed time out, then the cut time would quickly ramp up with a persistent problem and you'd have a non running engine. If the time out decreased proportionally with the increase in cut, maybe it could work without disabling your engine?

Can you explain a situation where you'd want that behaviour and why? I'm struggling to think of one. It seems nice to think about, though.

If you run out of injector capacity under boost you'd prefer not to keep hitting the fuel cut and keep the lean condition. So an increasing fuel cut period would likely help and not cripple the engine.

And you're right that you wouldn't want that with the rev limiter and you also wouldn't want this to completely cripple the engine. I must say I haven't thought this through but I thought it might be an interesting thing to consider.

Jean

I like un-thought-through ideas :-) At the very least they force me to rebutt them, which is always healthy.

Provided the hysteresis was sufficient a boost cut that removed fuel for X and then reapplied would be fine. Only without hysteresis would it be an issue. RPM limiter while in boost is the same, and the OEMs do that without issue.

Fred.