Right. I set all of the DFCO and transient settings back to stock '02 Silverado and went for a drive. Off idle transition is now acting the way it should and everything else seems fine except for one thing.
Transitions on and off the gas pedal in stop-n-go situations is harsh. Both ways, taking my foot off the pedal and then pushing it lightly like you do when going slow like around 10-15mph. Any words of wisdom on how to correct for that?
Jim
Need the latest toon file and a log. No sense in guessing.
I'm thinking it could be in the timing. That's the one thing that jumps out an me from the datalogs. And I did mess with the adaptive idle settings when first trying to sort out the original problem.
So let me guess: "Go back to the stock settings and try them" right?
Jim
Judging from that, yeah the spark delta could make a tip in pretty harsh. It ramps from idle spark to main spark when you get on the pedal, so the bigger the difference between the two tables, the more torque bump there will be.
smokeshow, I don't think it's been explicitly mentioned in this thread but this is not a LS, it's a P01 retrofitted onto a single plane/4BBL TB Buick engine.
I was wondering what it was with that displacement. Is it still sequential injection?Originally Posted by blindsquirrel
Yep, set up with the LS bits to use the '411, all GM sensors and such. Plus the weird tone wheel mounted on the damper and cam sensor in a cut down distributor/oil pump drive. But open chamber and smaller valves and ports and a sizeable dish in the pistons.
I went through the entire tune and everything that wasn't specific to this engine and car I set to '02 Silverado specs based on the tune I downloaded from the junkyard ECM. I've uploaded that file and I'll see how it runs tomorrow. Not really expecting everything to be right, but we'll see what it does.
Jim
I may not have answered the question. The intake manifold is fitted with port injectors and a throttle body. The original Silverado 26lb Delphi injectors (new) and COP is being used.
Jim
Well that was a bad idea. All the old problems came right back and I'm really not willing to go through all that again. So I loaded the tune I'd saved with all the Transient and DFCO settings reverted to stock Silverado and that ran very well.
I also now have the wideband fluctuating about equally across the 14.7 line and will continue to fine tune that.
Hot restart is too rich, shouldn't be hard to correct that.
Idle is too high by 200-300 rpms with the IAC dropping to zero and only 4.5 degrees of advance so I'll need to see what can be done about that. Might have to drop the advance even lower. Apparently all this tuning is making the engine run easier so it's running faster.
The attached screenshot is with the Silverado tune, not for any analysis benefit, just to show how wacked out it is for my engine.
Jim
Last edited by Jim Blackwood; 02-19-2022 at 09:28 AM.
Please help me to understand the Transient settings.
First here's a datalog screen print of four startups beginning at 60 degrees and up to 81 degrees. All worked acceptably well until the last one which was unsuccessful. Note also the last two, at 70 and 81 degrees showed a spike in the injector PW. Now I may be wrong, but I see the ugly hand of the transient settings at play here and would like to fix that.
So, next I have a screen print showing 5 things, please take note of the settings. As far as I know these are typical numbers but I'm inexperienced:
-Fuel From Wall Stabil: 1.00
-Fuel On Wall EXP Decay Mult vs Airflow
-Fuel Boiling Time vs Coolant vs MAP
-Impact Factor Gain vs Airflow
-Fuel To Wall Impact Factor vs Cool Temp vs MAP
So first off, am I looking at the right area? If so, first question:
-Is the Impact Factor Gain vs Airflow relevant for a Speed Density tune with no MAF? If not should those numbers be set to zero? 1? some other number?
-Is the Fuel On Wall EXP Decay Mult vs Airflow relevant and for the same reason? If not, what multiplier should be used there?
-Is the Fuel From Wall Stabil correct? What is the effect of say halving that value?
-That leaves the last two, Boiling Time and Impact Factor. Are these two table basically the inverse of each other? In other words Impact Factor showing the percent of the shot that hits the wall and Boiling time showing how long it stays there?
-It would seem that these last two would be where my problem could lie since they both are temperature based. Would the correct approach be to lower those values in the columns from 70 degrees up?
Sorry for all the questions and thanks ever so much for your advice.
Jim
Chris has detailed GM transient fuel very well in this post: https://forum.hptuners.com/showthrea...ll=1#post77975. Anyone wanting to learn about GM transient fuel or diagnose a related problem should get familiar with that.
Do not adjust anything until you understand what it is you're adjusting. These can be extremely sensitive - I have seen broken intake manifolds from pushing these settings too far unnecessarily. You also risk washing cylinders and diluting your oil. Not good.
To explain the risk in simple terms...this is what you had for your TF impact factor settings a while back:
tf.PNG
Impact factor is quite literally the weighting factor for how much fuel impacts the port wall and adds to the fuel puddle. 0.5 means 50% of the fuel injected at a given temp and MAP will NOT pass the intake valve. The algorithm compensates with a longer pulse to make up the difference. See the cell I circled from your table...that cell suggests more than 100% of the fuel stays in the manifold and intake ports. So it responds by trying to compensate with an infinite amount of fuel, as the fraction not passing the intake valve per that calibration means dividing by zero. Its not really infinite, but the ECM makes the attempt to compensate anyway. The result is a fire hose of fuel into a cold cranking engine that will contaminate your oil and can cause intake fires.
Hopefully that explains why these cals are not to be molested. Read Chris's post for more info.
Thanks for the link and the info. I'll get right on that.
The original was not my work and I reverted to the Silverado table as soon as you told me, which helped a great deal but apparently still has issues. Most of the values in that table are .5 or less. I did try reducing the columns above 68 degrees x.95 but that caused a high idle and so did x.98 so I'm very close to the limit there. Closer than I'd like to be. So I changed it back and tried decreasing the evaporate time x.95 but don't remember now exactly what that did. No big help clearly.
I'll read up on it in that link and then come back to this. A big help would be if I can learn the symptoms that indicate too much or too little in each table.
Jim
OK that looks like a good start. The main concern seems to be, at least for my tune, the lean spike on decel and to a lesser degree the rich spike on tip-in and I'll get to that eventually. There seems to be enough information there to be able to tune at least some of that out.
For now though I think I need to zero in on the so-called "Stomp Compensation" or Warmup Transient Correction, described as being "used to add additional fuel to transients during engine startup conditions. This is to adjust for conditions of inlet port and valve components not being up to satisfactory temperature. ie. extra fuel puddles on a cold intake." It seems to me that this could be where my problems with startup lie. These are all the stock Silverado values, in fact everything in the transients area now is.
BUT, something is close to a limit. This is obvious from the behavior of the injectors on warm restart where they spike essentially wide open for an interval that is more or less dependent on temperature as far as I can see. When I did a wholesale reduction of the transient values this interval decreased but other problems arose (the raised idle, up to 2500rpm). No real clue what the interplay was that caused that but clearly something told the ECM that the IAC needed to be opened and I'm just guessing that this mechanism is way more complex than anything I want to consider delving into. Best to just avoid it if I can. Which means no more changes in the transients table until I find something else that affects this issue. Have to look elsewhere.
May be time to run a search on "stomp compensation". Certainly I need to read up on it and see if I can get a handle on what changing those parameters do.
Jim
How can that be true. 100% injector duty would show in cranking log and cranking would flood and possibly hydrolock the engine.
I think if you put 1.05 you can get 105% of the fuel total. Multiplication IS division, but where do you see a zero to divide by?
I appreciate you and I want to see more posts from you explaining things. Its excellent work, I really know you are helping everyone, even me.
But in order to appreciate you and read what you write. It means I will also think about it and ask for clarification. I do a bit of computer programming myself and I am only reflecting the mathematical relevance of what you stated in hopes of better understanding it.
I have a lot of math experience (math tutor for 10 years & I've taken almost every math course offered at university including engineering maths) and in my experience if you have an equation which treats 1.01 as infinity then the number 1.00 would also cause the same behavior. In other words, you could approach 0.9999999 come close to 1 but anything 100% is 'infinite' then so is 1.01, 1.02, 1.03, etc.. 1.00+ Any number 1.00 or greater
and clearly that is not what is happening here because they put 1.00 in alot of those boxes. I'm also aware of the bit holding problem but that is only seen when approaching numbers larger than the capacity of declared type, and in this case we are talking numbers near 1.00 so I don't see any issue holding the declared variable with numbers larger than 1.00. It may be possible in the programming that numbers larger than 1.00 get treated AS 1.00 -> sure that could be true. But it won't go 'infinite'.
Last edited by kingtal0n; 03-04-2022 at 02:18 PM.
>=1.0 question, this seems to be an issue and should be resolved so we know more about what we are doing, and there's an awful lot of this type of ambiguity in the software, which just drives me bonkers. In a perfect world we'd resolve all of these ambiguities and know exactly what happens when we change a number, and know when we hit the limits. But is the limit 1? Is it zero? Is it 256? 100? 10? Or who knows what? But I'm more of a practical guy and I realize that this may be asking too much. Ideally the range would be from 1 to 10 for everything but that isn't happening anytime soon. Love it if it did, but what a surprise that would be. You gamer dudes can appreciate this I'm sure. Us engineering types make stuff way more complicated than it really has to be and deep down we all know it. No reason why the user side can't be in the 1-10 range and then scaled for the numbers required in the calculations formula. So much for philosophy, now back to work.
The problem here is that at temperatures above 70 degrees ECT I have to double crank to light off the engine. Sometimes I have to floor it (flood clear). The two conditions would seem to be mutually exclusive. Double crank adds extra prime pulses therefore more fuel. Flood clear is to get rid of excess fuel. So which is it? Is it both? How can it be both? And why isn't it also a problem under 70 degrees? This just goes around and around in my head and gets nowhere.
Somebody please make a suggestion and I'll try it. Not getting there on my own.
The one thing I haven't tried is to splice the Silverado transients table with the old "Country Boy 2bar SD smoothed" transients table since Country Boy seemed to work fine above 50 degrees and Silverado works fine below 70. That may be the most practical approach. I'll take a look.
Jim
Of course it can't...yet despite the limitation, it still tries. It's just how the math works. Chris's post details it all.
I see the math in the post
From the post:
Fcyl = Fdirect + Fevap
and
Fevap = Fwall(e-boiling time) ... an exponential decay
Fimpact = Fdesired * impact factor * gain
Fwall(new) = Fwall(old) + Fimpact - Fevap
If you put a 1.05 instead of 1.00 into impact factor or gain or whatever... it just mutliplies by 5 more percent... I don't see the zero or infinity or anything even close. Its just straight up multiplication. And um... addition. yeah adding
In any case. Good call on the 2007 post explaining stuff we should have known for 15 years already but still somehow fail to grasp completely.
Well, me anyways
Last edited by kingtal0n; 03-04-2022 at 04:43 PM.
I had a similar issue that turned out to be two things.
1. The fuel pump prime would take longer in the hot climate (3-4 seconds) than cold and the ECU only primes for like 2 seconds. So your initially could be cranking without any fuel pressure. I bet you don't have a fuel pressure gauge (i dont hehe). Try key on, prime, key off. Wait a second. Key on, prime, key off. Wait a second. Key on, prime -> Cranking.
2. The air from the table "park position airflow" needed to be doubled from 10g/sec to like 22g/sec to be more comfortable with my fuel prime pulse ranges
Its alot easier to fire an engine with a nice open IACV and a bit extra fuel than it is to fire one with a closed ass IACV and a bit too much fuel.
I bet your cranking fuel is bit too much, the airflow is a bit too low, and when you first try to crank the fuel pressure could be low which makes it seem like no fuel is going in even though you keep increasing the cranking fuel parameters.
Excess impact or too little evap factor will drive the transient fuel controller unstable. Feel free to experiment for yourself. You've been warned though
Excess impact or too little evap factor will drive the transient fuel controller unstable. Feel free to experiment for yourself. You've been warned though
If it was a real time tuning ECU I would have long ago elucidated. But because it requires a upload period to perform changes, I am far too lazy to perform 'experiments' #(@*$ annoying to make 1 change at a time then upload it and so forth. Setting up ecu bench isn't an option I am working on experiments for school. I have 20 oscilloscopes available ironic I'm too busy compare in excel data serial whether its brown color intensity from a stained slide section or injector duty from a transition event. With the bench you can hold conditions steady without concerns for differences in road conditions and observe how changes affect injector outcomes repeatedly confirming the existence of phenomena.
I bet our friend here feels the same way.
It doesn't take alot of experience or even awakeness to make a change and look at a gauge to see whether you went the right way or not.
It does however take alot of time and effort when each change you make can take 10 or 20 minutes plus has a risk of completely frying or damaging the ECU and making the vehicle undrivable because a cat chewed a USB cable or something.
If it wasn't for the cat factor... maybe
In any case I've put a few 1.05's into those gain factor and so forth without any ill effects. My tune file has a bunch of them for many years now, at very low airflow values though (near cranking perhaps)
I got the transitional behavior I wanted and forget that it exists like everything else.
I do believe you , that a number can unexpectedly "blow up" when logarithms and similar are in play. He mentioned an exponential decay but thats a decay so its going to rip out the fuel faster if it blows up which is fine atm as we are investigating richness. But none of the equations posted list a log value or base for enrichment so idk what do with invisible maths
Last edited by kingtal0n; 03-05-2022 at 06:34 AM.
Reply With Quote