It's not often I come here with a serious puzzle, but...

kb58 wrote: Once the "thing" happens, the car stops accelerating and the rpm doesn't increase much after that.

I'm just saying that without data AFTER the event, its hard to figure out what is going on. I don't mean ride it for a minute, I mean don't instantaneously pull your foot off the throttle when it happens (just an extra second or two). How the system responds to the issue is just as important in troubleshooting as what happens leading up to it, since at the end of the day we don't know everything about whats been done programming wise to the ECU.

From the data you have provided us and how strongly "built" the motor is as-described, unless there is detonation, I sincerely doubt you are going to "hurt" anything, but that decision is up to you, just trying to help

9s should be plenty cold enough. Gapped to no more than .025 I hope, preferably .020 with the cylinder pressures you're running. Cold plugs don't tolerate wide gaps, the "cold" comes from shortening the ceramic insulator, and when you get to this range, surface discharge across the ceramic gets to be likely, so you need to make the gaps smaller.

Proper feeler gauges are on the way - and boy are good ones expensive.

Yesterday I took the plugs out before I couldn't find the gap-measuring tool, so they were reinstalled into the same locations along with the same coils, and the gap wasn't touched. I added cam-related data to the list to be logged, temporarily raised the rev-limit to 8500 rpm (and eliminated all "pre-limiting" factors as well). I also raised the boost target from 170 to 210 kpa, since that's about where the new spring pressure is, and did a run today...

The problem was much worse! MAP still has the same drop-off, only now it happened at 6800 rpm. Apparently the ECU didn't like trying to control boost closed-loop with the stronger springs, using the old boost-vs-duty cycle table, and the PID loop was in need of tuning. Aside from that mess, the logs showed that none of the cam variables were acting up, so cam timing is off the list.

Another run was made late in the day, switching off closed-loop boost control and running solely on the wastegate spring. The rev limit was still set to 8500, and...

The rpm drop-off anomaly was gone! Yup, you read that right. All excited, I lowered the rev limit to 8000, and it came back at around 7200 rpm, so there's still something strange going on, maybe a sw bug. More testing is in order to determine which limit variables are causing the problem and how, but at least I finally found a variable which affects the problem directly.

I'll post a question on the ECU user's forum tomorrow but no doubt they'll claim it's perfect unless I load a cal and logs. Once it's established what's going on, looks like someone won a book!

Interesting

YAY PROGRESS!

It's always something "dumb", generally because all of the "smart" options were exhausted. I had a real "problem child" at work that I also solved yesterday. Was running very poorly when hot. Everything checked out okay. Ended up bringing another car of that model to the place and methodically swapped everything over. We ended up swapping out everything but the license plates and nothing was solving the drivability issue.

An extremely long and painful story short, it was something (on the surface) completely unrelated...

I'm beginning to have some doubts. I woke up this morning thinking that if the rev-limiter really is root cause, the problem should be present in second gear, and it never is. The rev-limiter isn't gear-specific, so is it something like an intermittant connection on an ignition coil that I disturbed during plug removal? This is what I get for messing with more than one thing at a time.

My usual methodology is to make a change, do a run, check the results, make another change, then do another run. If it's an intermittant problem, or something that "moves around", it could really mess up any conclusions.

I'm going to have to do more runs, maybe three at a time, to confirm that the problem is at least stable and repeatable. After than, adjust one rev-limit parameter at a time and see what happens, because right now it doesn't fit the facts very well.

The logger is always running - I just have to take notes on the setup for each run. The tunes and logs are going to be needed because I'm pretty sure that the ECU manufacturer will claim that there's nothing wrong. (That already happened once when I reported that a pin assignment in their manual was wrong - "sir, no one else has ever had this problem." Uh huh, and it was very satisfying to hear them admit it was in error.)

Sounds like something isn't working like you (or they) expect. Keep us posted on the final fix!

You come up to the limiter too fast for it to be noticable.

Knurled wrote: You come up to the limiter too fast for it to be noticable.

potentially...

Whats the sampling rate of the system?

Fueled by Caffeine wrote:

Knurled wrote: You come up to the limiter too fast for it to be noticable.

potentially... Whats the sampling rate of the system?

Which is why it would be nice to look at some data to see if that gives us some clues.

The manufacturer hasn't stated the sw loop time though I imagine it's probably 100s per second. The datalogger sample rate is anything I want it to be within reason.

I asked the manufacturer to explain their rpm limiting in more detail. They have PID values associated with it, but it's not clear when they're enabled. Users can "assume" it's during the spark and fuel cut window, but it isn't explained fully... which reminds me...

The ECU manual is clearly written by people who know the system intimately, and that's a problem. I've all but told them they need to hire a writer who's never used their product, and have HIM write it. Only then do users get all the basic background information that the pros take for granted and don't bother mentioning. Over and over I see people asking, "how do I do this", only to be told of some secret handshake/commands that no one knew of because it's not in the manual, but I digress.

In reply to kb58:

What you need is a strategy book. And I'm sure the maker is afraid of giving that out for fear someone copying them OR someone finding out that they are being copied.

From all I have read about your experiences this new ECU company sounds very "special"

In reply to kb58:

Is the changing of the drop-off point making you think that the problem is related to boost control? I have had a similar problem behaviour wise that was due to hitting the limit of possible maf scaling ( about 300 g/s) on the 16 bit Subaru ecm. But I have a hard time thinking that is the cause here.

At the moment I suspect a rev-limiter value (the ECU has several) isn't working as expected, or maybe I don't understand how it works (quite possibly). As mentioned above, the PID values are suspicious because depending when it's enabled, it "may" be affecting engine speed a fair distance below the stated rpm limit. I'd expect the PID loop to only be enabled at and above the rpm limit, but it's not clear when that's happening.

I do wonder a bit whether this is another red herring. Since the engine rev-limiter is such a central part to protecting the engine, I can't see how I can be the first to find something wrong with it.

My plan is to generate two tunes and two logs, one showing the anomaly and one not, with the only difference being the yet-to-be-identified rpm limit parameter.

So earlier I speculated that the ECU mfg would claim that their rpm control is perfect. Sure enough

No, there is not a rev limiter bug.

It's nice to have an incentive to find out!

No engineer, software or hardware, who has any experience whatsoever, will EVER say with 100% certainty that their product/design is perfect and infallible. The engineer-brain knows that there's always going to be some rogue, unconsidered, therefore untested situation...

"There is no bug" means you're talking to marketing, or tech support, not an engineer. That, or the rule of exact language is in play: it may not be a "rev limiter" bug but a different bug altogether

Some engineers DO proclaim infallability in their design. At that point they have ceased to be engineers and have become zealots, because they have closed their mind to the idea that their baby isn't 100% perfect.

Absolutely agree - I was in software for 10 years, and 5 yrs of electrical design before that, and No engineer would ever say that unless he'd sold his soul to Marketing. Seeing the reply was priceless and I really looked forward to knocking the attitude down a notch. However, for this anomaly they may be right. I loaded in the original rev limit settings and about half the time it worked great, and the other half of the time it had the usual problem. I have no idea whether it's always been this way and I just happened to catch a few bad ones, or if me messing with the coils half-way fixed it. Regardless, I received the feeler gauges tonight and will get the coils tomorrow. The plugs will be re-gapped and the coils swapped out, changing nothing else. A few drives with that setup should be enlightening. If it keeps doing it, I'll adjust the rev limit upward again. After that... yeah, well, the next time you see a homeless guy muttering to himself about "those dang PID loops", that'll be me...

Oh, and I added more logger parameters to show whether the ECU was limiting spark or fuel at any time. The only time they showed activity was when it worked fine and properly hit the rev-limiter. What a puzzle!

Why the rev limiter has a pid loop is a bit baffling. What is it controlling with it?

Rev limit = driver is stupid, stop engine, now. Seems simple.

Could be not many people actually hit the limiter. I set them for the mechanical limits of the setup, usually quite high.

Coils arrived. Once it cools off (it's dang hot here) I'll swap them out.

About the rev-limit, I imagine because they use PID loops everywhere else, it was free to use. They say it's so that the rev limit can be "softened" so it doesn't hit so hard. Maybe it's important to drifters used to just keeping the pedal to the floor? I set it on the low side mostly to protect the engine from going nuts, like, say, the throttle cable sticks and I shift to neutral. It would give enough time to reach over and kill power.

kb58 wrote: They say it's so that the rev limit can be "softened" so it doesn't hit so hard.

And I think this may be the source of your issue. It is probably hitting the soft limit 500rpm before the hard limit.

So it is "not a bug", it is working as designed. Therefore it's a PEBCAK - silly end-user wants access to all safe available RPM.

And of course, there's probably no way to turn this feature off, because We know better than You what You want...

(is the ECU company German by any chance?)

The rev-limiter does have a soft limit threshold which is user-adjustable. The soft limit was set to 7800, and set to be completely limited at 8000.

Last night's logs showed - when the engine did successfully pull to the rev-limiter - that it started limiting at 7800, so I can't point fingers.

Oh and maximum injector duty cycle was at 75%, so injector sizing (1650 cc) is about right.

Nope, they aren't German; they're right here in Southern California, that should narrow it down.

The "we know better than you what you want" attitude is a very German one, is why I asked. The concept of a "power user" is nonexistent.

Back from the drive.

Drive #1, with the re-gapped plugs, new coils, and:

Rev Limit Fuel Cut: 8000

Rev Limit Spark Cut: 8200

Fuel Cut Start Window: 200

Spark Cut Start Window: 200

Did two runs, with the first one failing short at 7700 rpm (without fuel or spark being cut). The second run succeeded, with fuel and spark being cut at the points specified.

Drive #2, with rev-limits moved upward.

Rev Limit Fuel Cut: 8200

Rev Limit Spark Cut: 8200

Fuel Cut Start Window: 200

Spark Cut Start Window: 0

First run: succeeded, with rev-limiting starting at 8000

Second run: failed at 7700.

Third run: failed at 7770.

Next I'll try bumping rpm limits a lot higher, maybe to 8500, to clearly see if it's a contributing factor or not.

After that, I'm going to pretty much throw my hands up if it doesn't change anything.