LT1 Throttle Oscillation Question

[96GS007]

Have an interesting little nuisance I'm working on. Every once in awhile, I get a rapid throttle oscillation. When it happens, it occurs just off idle if I'm trying to creep the car forward or I clutch in around 2000 rpm when slowing to a stop. MAF and VVE tables all look good and the VTT seems to be pretty good as well. If needed I can provide a much larger section of the log file.

99% of the time the all is good. I've attached the current tune and a couple minute snippet of a much longer (40 min) file. About 55 seconds into the snippet is where the oscillation begins. You can see the ECM adjusting timing and throttle to try and catch it. Off the cuff I'm guessing I need to increase the spark adaptive a little more and reduce the throttle. Before I go messing around more with the adaptives, I thought I'd ask for a second set of eyes to take a look. Unfortunately I'm super time-constrained at home with some family matters so a bunch of experiments are difficult to fit in. You'll note I'm on Rev 40 of the tune...lots of past experimentation...just no time to do so currently.

Car/Mods

2019 Corvette Grand Sport with the LT1 and 7spd manual transmission.
- Ported cylinder heads
- Nick Williams 103mm throttle body
- Mamo ported MSD intake manifold
- AFE cold air intake
- LG full length headers
- G-sport (GESI) Gen 2 cats rated at 500hp/bank
- DoD deleted
- GPI SS3 LT cam (pic of cam card below)
- RPS Billet Carbon Street Twin Clutch w/Aluminum Flywheel (>40lbs lighter than stock)

Screenshot 2024-01-03 121753 - Throttle Oscillation.png

2024-01-01 R40.hpt
2024-01-02 R40 Throttle Fluctuation.hpl

[GHuggins]

Give this one a try. Basically torque is too low at idle and from what I can tell predicted too high. From what I've been seeing predicted is controlled by the DD. Your DD isn't necessarily adjustable at idle ( it is, but not really ). I also believe your torque model has already learned and is why it's not showing drastic differences. Usually you have to add to the map side to correct the offset created by the VE table being lowered. You also have all of your adaptives killed for the minor corrections other than throttle. I dialed back throttle and laid the spark into it while allowing some leeway into the external table. May not help. May not be worth a damn, but you're welcome to try it.

[96GS007]

Hi Greg!


I'll give it a shot and see what happens. Over the next several days I'll drive it so it can learn and then I'll post another log.

I wonder if it would help if I set the rolling idle a little higher? Hate to call it a "stair step" per se, but rather than dropping it to within 50-75rpm of the base map as I roll to a stop, let it drop to 100rpm more and then when I'm at 0 mph it can drop the remaining rpm. I know....band-aid hack at best....even if something like that would work.

Peeking at your changes...I can also experiment with the spark adaptive on the left side of the table....edge back closer to the left side numbers in the R40 tune. The engine does seem to respond quickly timing changes.

As always, your sharing of knowledge and assistance is Greatly appreciated!

[96GS007]

Made some VT changes and the oscillations have pretty much calmed down. Also made some tiny tweaks to the adaptives on the spark side. It drives fine and idles fine....850 rpm..., it'll coast down to a stop, LTFT vs MAF and LTFT vs VE, etc is all good, at least to my untrained eye

All that's left is one little tiny area in the 1500-1800rpm range where if I'm just letting out the clutch to get the car to inch forward and then rapidly push the clutch in, it'll die. For others perhaps reading this, the clutch/flywheel setup is >40lbs lighter than OEM. Anyway, I can generally force it to die...and I can "drive around it".

Two logs are attached and they correlate to the R43 tune (attached). I've also attached the tune I plan to try tomorrow....R44....that has changes on the MAP side of the VT tables. It looked like those were off when I compared log data (broken down in the same manner as the tune) to the VT tables in the tune. We'll see if it does anything or if I default back to R43 and try something else.

Comments and criticisms welcome! I'm trying to learn. The days of my youth were spent with Holley carbs, the old L98 TPI cars where we swapped chips, and Gen 2 LT1/LT4 OBD II cars and...."LT1 Edit".

[GHuggins]

OK, so for idle dips and stalls when coming to a stop you want 2 basic things. You want the throttle to close slow when coming to a stop first off. This will be 1600 and under lower loads of the airmass side. Lower to aid in this. I didn't do that in the below. If you have it in your cal, which I didn't look, you can also make the decel region of the pedal progression table more positive. Next you want throttle to open, but not fast or a lot and you want timing to increase with load. You address both of those on the map side by having torque increase with spark right around the idle rpms. Then you slow the opening via the adaptives. Hopefully idle torque and timing will be better with this one and you'll get back some of the dip control. You can also increase the external load and reserve to aid in this as well. This is also granted VE and fueling in general is pretty close. Hope this helps.

[TriPinTaZ]

Greg speaks truths! And he won't charge you $10,000 either.

[DSteck]

Greg speaks truths! And he won't charge you $10,000 either.

Lol. There’s guesses and there’s knowns. I deal in knowns which take time.

[GHuggins]

Lol. There’s guesses and there’s knowns. I deal in knowns which take time.

I've done lots of trial and error to come to the discoveries that I have. I've learned there are A LOT of tables in the cal that affect idle and idle related controls even some in the DD section that people have never considered and several in the background that we don't have access to. Even talked to coders to get the line of order that the code goes by for control. Honestly I would consider the torque model pricing if it had one less 0 and I knew the background players for auto's would play nice with it or if you could provide xdf's for those background multipliers. I know it takes time to get it right. Up to a week on the dyno going non stop and on top of that you have to make sure to keep certain temps in the same range otherwise it's a waste of time unless you dial those in. I happen to know how to dial those in. I spend a lot of time fine tuning cals for customers so torque and fuel stay the same at all temps. I understand the time involved better than most, but for 10000, I can't justify it especially when there are so many background players that can't be adjusted to work with the new torque model. Manuals aren't a problem and you'll never see an issue with dialing them in. Auto's are a completely different story.

[DSteck]

I've done lots of trial and error to come to the discoveries that I have. I've learned there are A LOT of tables in the cal that affect idle and idle related controls even some in the DD section that people have never considered and several in the background that we don't have access to. Even talked to coders to get the line of order that the code goes by for control. Honestly I would consider the torque model pricing if it had one less 0 and I knew the background players for auto's would play nice with it or if you could provide xdf's for those background multipliers. I know it takes time to get it right. Up to a week on the dyno going non stop and on top of that you have to make sure to keep certain temps in the same range otherwise it's a waste of time unless you dial those in. I happen to know how to dial those in. I spend a lot of time fine tuning cals for customers so torque and fuel stay the same at all temps. I understand the time involved better than most, but for 10000, I can't justify it especially when there are so many background players that can't be adjusted to work with the new torque model. Manuals aren't a problem and you'll never see an issue with dialing them in. Auto's are a completely different story.

It?s all there for autos, and I already submitted the DMR requests. There is a LOT not available to end users. I constantly add things myself, so it is definitely unfair to most users (who usually assume that what?s there is all there is).

You?re talking two weeks, easy, to do a thorough good job, and on top of that? it requires a loaded dyno. Seeing how it actually calculated everything, though, the auto trans values don?t matter much and don?t really come into play aside from back calculating from axle torque to brake torque (to determine desired engine torque based on requested axle torque). It?s all based on trans fluid temp fwiw.

For example, in a ZL1 A10, the axle efficiency is 0.970, ratio is 2.85, 7th gear efficiency is 0.984, and ratio is 1.0. Take the gear ratios out and it?s a 5% loss, and then there?s a fixed spin loss of 21.5 Nm in those conditions. Point being is that it?s all there, but those values won?t really change much although I?d recommend base lining a totally stock car to see if you wind up with deviation from the model (which I?m sure exists).

$1000 wouldn?t get me to start the process though. I?m just saying? it?s a lot of work to do it right. You?ve got 12 different engine rpm break points you need to operate at (fixing rpm and sweeping load and spark is the proper way to do it). Once you have the back calculation math to go from axle to modeled (you?ve got driveline loss and friction inbetween), it?s just a matter of collecting all the data. That?s the major time consuming part. Then, the right way to do it is run regression on only the data collected instead of having to populate a table and fill in sections you didn?t actually hit with data that looks like it fits the trend. That?s my main gripe with the tools available is that you?re forcing math to consider invalid data. It?s just not right.

[smokeshow]

It's only gonna cost that much because Dave will be in the drivers seat like a caveman doing all the work manually. He could bring the price down if he automated...well, any of that process.

[DSteck]

It's only gonna cost that much because Dave will be in the drivers seat like a caveman doing all the work manually. He could bring the price down if he automated...well, any of that process.

If it were reasonably possible…

Somebody has to make a judgement call that the engine isn’t about to explode and to let it chill.

[smokeshow]

If it were reasonably possible…

Somebody has to make a judgement call that the engine isn’t about to explode and to let it chill.

Automated estops lol. At Chrysler I'd hit 'go' on the dyno right before I went home for the day

[GHuggins]

I am curious about that SmokeShow... How many engines would y'all dyno to get the torque models and how many of those would blow? That is a lot of rpm to hold even with all of the temperature controllers.

Now I can be completely wrong with all of this, but Here's the part I'm having a hard time grasping and like you said Dave, their's a lot of error going into the math's and so on, BUT to me I get the feeling Jake worked for GM when the torque modeling was being developed (end of 3rd gen and into the 4th gen era - again could be completely wrong about that) then left to go onto better things or was let go (I worked on the wrenching side primarily during this time so I know a little about how they got rid of a lot of their main developers around 2010ish? to 2012ish? and then I was just dabbling in the tuning side of things then or rather working on going back to it ). This era to me appears to have the torque models "correctly calibrated" (drops off with spark for cams for instance).

Now we fast forward to gen V's. To me it looks like the torque model is being manipulated nearly all of the time to get the controls they're after. Torque goes way up with spark at idle for timing and throttle controls and then they may even have it nose dive with rpms to close the throttle. Now move onto the newer gen V's using the E90's. To me the torque models are getting back to reality and looking more like you would expect, but still being used to manipulate things.

This is especially why I'm curious as to how it turns out. If it'll work right plugging in the correct numbers without changing anything else, then I'm obviously doing something wrong and what I see in the torque models makes 0 sense knowing all that I know running dyno's (engine and chassi's alike)

[DSteck]

What we can log right now for torque modeling is not really helpful.

[smokeshow]

Temperature related issues on the dyno are pretty much unheard of... unless someone makes a mistake. We wouldn't run the test if we expected overheating. I was at GM when torque models were being developed...but torque models are always being developed at GM. This stuff in particular has been around for decades though...so before my time.

[GHuggins]

Temperature related issues on the dyno are pretty much unheard of... unless someone makes a mistake. We wouldn't run the test if we expected overheating. I was at GM when torque models were being developed...but torque models are always being developed at GM. This stuff in particular has been around for decades though...so before my time.

Ok, kind of confirms the impression I got. Also why I'm pretty certain the 4th gens are in actuality because well honestly they go hand and hand with what I've seen operating dyno's and doing some of my own engine research. Do you know if they do indeed manipulate things for the earlier 5th gens to get the controls they're after and then use multipliers to bring things back in line?

[smokeshow]

Do you know if they do indeed manipulate things for the earlier 5th gens to get the controls they're after and then use multipliers to bring things back in line?

I don't know what that means.

[GHuggins]

Going back to my earlier assumption of what we are seeing in the gen V's.

Idle torque - prime example - torque is WAY higher with load increase as spark increases. Pretty sure of either one of two things here as this is physically impossible...

Either A, they're only dialing in the "exact" area where it's idling (this is mainly on the map side too) and then they are interpolating from there OR

B, they are manipulating things outside of the normal seen realm to gain "other" controls such as to keep throttle and timing working how they want it to

You see this with RPM increase too. Torque models nose dive in the higher rpms then will jump WAY back up again to control rev limiters or throttle or torque management and then they "seem to" use other background controllers as multipliers to put other said things using the torque model back into check.

Just like idle. They use pedal progression, rate limits and then all of the other big players to control idle. When cams are thrown into the equation or other things that "might" lower torque output at idle, these other controllers become "big" players causing a constant "fight" to keep things right even if you lower the torque models to "what they should be". For the ones really using it, just dialing back pedal progression in the idle cells can raise idle timing 5 to 10 degrees and torque about the same.

At least this is what I'm finding.

[GHuggins]

Here are some examples. Factory 6.2 showing to make less torque than a factory 5.3 except for the map side. Dyno's would say otherwise...

Then we have the factory zr1 straight from Dave's website or even a factory LT4 as far as that goes. They nose dive bad in zones outside of the normal to control throttle.

All I'm saying is it sure seems like the models are "manipulated" to get the controls GM wants and then multipliers are used elsewhere to put things back right. Again anything using the E90's seem to go back more like the gen 4's...

[GHuggins]

To the OP, sorry about the thread hyjack. Guess we should have put this in the torque model discussion thread.