PorKen

Unfortunately, I barely understand it myself, hence the vague language.

If you have followed some of my various rambling threads you may notice that once I begin to understand something completely, I start to describe the topic in much simpler terms.

I was hoping that those with more experience with the LH code would describe 'O2-adjust' and 'Adaptation' in more detail, in a way that I would understand as well, but it looks like I may have to figure it out myself - and make a new thread for it.


I literally looked at that small piece of limit code above for about week before it started to make sense. One problem is I also believed that it could not be so simple.
Then I tested the effect of changing those limits, then adding enough fuel (by changing a injector constant) to overcome the bias of the 'sleeping' or 'frozen' O2-adjust registers.

GregBBRD

This discussion about code and fuel really confuses me.

Seems like an engine running at 6,000 rpms with 12.5 to 1 mixture and 25 degrees of timing should make the exact same power, regardless of which "code" is used.....

Am I missing something here?

Hilton

Of course what you say makes sense - the tricky part is making sure the 12.5:1 is consistent and predictable.

If the engine only determines the adjustment variable while its running in closed loop, and only runs open-loop above 3800rpm, then there's the risk of target AFR in cells in the sharkplotter being o2-adjusted to 14.7, which will skew the long-term adaptation parameter, and thus will over time, cause drift in the AFR for cells where it runs open loop (at high rpm/load - where it has the potential to hurt more).

I think what Tuomo is asking is, "exactly which cells on the sharktuner/sharkplotter map will the LH force to be 14.7 AFR during closed loop operation?"

By knowing explicitly which cells , so that the target map can be set that way, you can minimise the influence of the longer-term adaptation on the higher-rpm/load cells' AFR, where its open-loop, and thus ensure consistency and predictability of that 12.5 AFR at 6000rpms. Or you can bypass the whole thing and run your car like a non-cat car, with a CO pot, and no O2 adaptation

Of course, I didn't go to school in a country with first grade, let alone 3rd grade, so I might just be exposing holes in my understanding, or misinterpreting Tuomo's questions, or just reading tea leaves

(apologies if I've got the terminology wrong - my laptop has my ST2 software on it, but I'm too lazy to go find it, turn it on and look at button labels)

PorKen

Yes? One of the subjects at hand is the transition point to open loop.

In regards to your confusion - I have recoded my LH to make it easier to go open loop at part throttle and then make it stay there until cruising again. At the same time I change an injector constant so that instead of 15:1 AFR, it is around 13.5, for much improved torque (both by the extra fuel and higher ignition advance this allows).

This has nothing to do with full throttle at 6000 rpm, where many folks never linger. (Although I have also written a lot of code to make WOT more consistent.)

GregBBRD

That makes sense....although that seems like it would really kill the fuel mileage, for a subtle change in partial throttle torque.

I wonder if people that are cruising just below where the O2 sensor normally "kicks out" are concerned about a little more torque or if they would rather have another few mpg?

I know that the people who own my street strokers are absolutely "giddy" that they get 2-4 more mpg, cruising down the freeway, than they did with their stock engines.

PorKen

These are better left for another thread I reckon, but to your questions...

The tests are easier but still above normal cruising. The torque difference at 13 vs 15 AFR is quite noticeable, pedal response is much improved.

I have actually logged better freeway mileage, possibly from not having to change gears as often (as with your stroker engines, no doubt).

ptuomov

I just want to avoid the following scenario:
1. I tune every cell for best power, say for the sake of argument 13.1.
2. Then, I start motoring on the highway.
3. The LH running in closed loop mode and adaptation on, and learning, interprets this as the MAF etc. being horribly worn and starts pulling fuel everywhere like crazy by making the adaptation adjustment very negative.
4. I decide to flatten my eyeballs by pressing pressing the pedal to the metal for 10 seconds.
5. I run lean in open loop mode at 6000 rpm and 1500 SCFM, because the adaptation correction is applied everywhere to pull fuel.
6. There's a hole in the piston.

I can avoid this by turning off adaptation. Except I don't like turning off intelligent features.

So... I would much rather know in which cells the LH 2.3 learns the adaptation level, then tune those to 14.7, tune the other cells to 13.1, and only have the adaptation correct me back to 13.1 if the MAF sensor starts drifting.

For me to do that, I need to know under what conditions (i.e., in which fuel map cells) the LH 2.3 learns the adaptation correction and not ever tune those for best power.

What are those cells in which adaptation learns?

Can I change that rules about in which cells adaptation learns?

jcorenman

Who are you asking these questions of?

ptuomov

Anyone who's both willing to answer and informed enough to answer correctly.

Given that this is the ST2 thread, I am making one general point.

Instructions how to use ST2 and a detailed explanation of how the LH 2.3 works aren't separable. You have to understand how LH 2.3 works before you can use ST2 correctly. I would like you, John, and Niklas to put more of your knowledge in the ST2 manual regarding how those features of LH 2.3 that are relevant to currently available adjustments in ST2 really work.

In particular, as an owner of ST2, I would like that manual to contain more information about how exactly adaptation is applied, how adaptation is learned, and what exactly triggers closed and open loop modes.

It's a great product and I think this would make it even better.

Hilton

It seems from discussion above, that the thresholds are 3500rpm or 280 "maf units", which would correspond to the cells in the big red rectangle below.

If that's the case, then..

makes sense, and as I understand it, all the cells inside the rectangle should thus be set to 14.7 when (shark)tuning 928's which will run with an active O2 loop?

jcorenman

Tuomo, the reason I asked is that you posed a classic straw-man question:
No one ever suggested that you do that. The contrary, in fact-- the "cruising down the highway" part of the map needs to be tuned for stroich (if you are going to use an O2-sensor) and the high-load part needs to be tuned appropriately for that.

I understand your question, I tried to answer it above but I can't give you an exact range of cells. I also don't believe it is an issue: I've run Sharktuner for many thousands of miles, I've never seen a situation where it would have screwed up adaptation based by getting into high load but still running closed-loop. High RPM and light load, yes it will run closed-loop-- so those targets should stay at stoich.

And sure, it would be great if the ST had another five tabs of parameters to play with, and the documentation covered a lot more of the inner workings of LH, but seriously-- how many folks are interested enough to spend the time learning? This is a small piece of a small market.

jcorenman

No, by 260 you are well into WOT and 14.7 is too lean. The WOT switch will shut off the O2-loop but it is not always reliable. The default map (shown) probably gets rich too early but by the 210-230 row you should be shooting for a richer mixture. For us, the 160-200 area is sort of "no man's land", there is some data there but you don't hang out for long.

The LH will also run in closed-loop mode (O2-sensor active) at high rpm and light load, so I would suggest sticking to the row-by-row targets shown in the default map.

RKD in OKC

When does the adaptation stop, isn't this asking when the WOT map comes in, and isn't the adaptation effecting the WOT area why there is a WOT map in the first place?

PorKen

Looking back through my logs, 280 is too high. (An even 4 x 46h - should be a logarithmic function)

250 may be a better number. (the MAF threshold value 46h (70) * 3.6 = 252)

I use 3Bh (59) which kicks in north of 210.

ptuomov

Two issues.

First, even if we know when the system is in the open loop and closed loop modes by logging the data (ST2 has the option to record the O2 adjustment, which goes to zero when LH switches from closed loop to open loop), we still don't know in which cells and under which conditions adaptation learns the adjustment. The cells or conditions in which adaptation learns the adjustment may be all the time when the car closed loop, or a much smaller subset of conditions.

Second, this is certainly not how people actually use ST2. They tune a lot of the cells in that rectangle to much richer and leave adaptation on. I don't understand the consequences of this practice.

The key to knowing how one should tune with ST2 is to understand when adaptation learns, if one leaves adaptation on.


It's an extreme example, but not a straw man. How do I know that I am not tuning a cell to be richer than 14.7 that the system then perceives as an error and causes an unwanted compensation?

Since I had the now-fixed fuel heating problem back in the day, I've seen all kinds of expected and unexpected (to me) behavior of the adaptation adjustment. This got me to think what exactly does that adaptation do...

Adding new parameters may or may not be worth it. Depends on demand. However, I think that fully documenting what the existing features do is something that the existing customers would expect. Consistent with this, I'd really like the documentation to tell me under which conditions adaptation learns. Being able to change those conditions would also be nice, but less critical.

I am very supportive of this ST, ST2, and SP effort. Overall, I think the product is awesome and everyone should buy ST2. This is just some constructive criticism of how to make the product better.

One useful Sharkplotter feature would be to plot the percentage of time that was spent in closed loop mode in each cell. Then, one could as a short cut tune those cells to close to 14.7 that are predominantly closed loop, before knowing the exact rules by which adaptation learns.

Adaptation has two parts. First, when is the adaptation adjustment applied. Second, when is the adjustment learned. My understanding, perhaps incorrect, from this conversation is that adaptation adjustment is applied always. My understanding is that we don't exactly know when adaptation learns, but logic dictates it is some subset of those cycles that are spent in closed loop.

Are you talking about moving from closed loop to open loop or when adaptation stops learning? I assume former. If so, you're saying that your chips move to a open loop (and for practical purposes to richer fueling) at lower load levels.