Weird lean condition I can't figure out

Now and then it commands a leaner than 1.00 Lambda while under load, and I can't figure it out. It doesn't happen often, but when it does the lambda varies between 1.1 to 1.7 lambda. It studders when it does this, and it's not a fuel cut feeling, but just like it looks, too lean. It's always about a half second after I give it gas, then it leans out for 1.5 to 2 seconds, then it's fine. It doesn't matter how much gas I give it, so it happens with just a little or it could be floored, but it only happens when I give it more gas.
Usually only one O2 sensor sees this lean mix, left or right which seems random. Today it did both, which also happens. The Desired Lambda and EQ Cmd command it lean, so it's not a fuel supply issue. It seems rpm, load and boost don't matter much either.

Another thing is it has only done this with three tunes, ever. If I load a new tune it goes away. I see nothing in common between the three tunes so I suspect that it may be a glitch because that does happen. The second tune it did this lean thing I decided to drive it for a while to see if I can figure out what is happening, but no luck. Over two weeks it only did it maybe ten times?
The only pattern I see is it almost always happens after being stuck behind a slow driver. You know, the 65 in the fast lane type. Once I get a shot to get around them, studder... Again, it's rare that it happens, but when it does, that's the condition. If I try to simulate it, it's a fail. Is that weird or what? So it would seem the existence and proximity of the inconsiderate a-hole appears to be affecting my car ;)

Basically, how/why would it ever command lean under load? Anyone know? Anyone else had this happen?

Pix example from yesterday:
Attachment 142029

This issue is happening more often as time passes. Now it does it several times a day and 1/4 to 1/2 throttle seems to trigger is most often. I should say 1/4 to 1/2 gas pedal, the throttle does whatever the hell it wants. It still refuses to do it when I try make it happen, and still greatly prefers it when passing an obstructing driver.
I've been running 50% Ethanol but changed to Gas for a couple months while trying to smog it, and it still did it.

Nobody has any idea or had the same happen? Just another one of my personal jinxes?

Probably the worst one yet. Lean for 3 seconds, peaking at 1.6 Lambda under 16 lbs boost at the 12:13:35 time stamp:

Attachment 145493

Surprised it even ran, but I kept on it just to get data, then let off when it seemed it had no intention of recovering.
I see STFT goes to 0, which only happens when decelerating and is the only time Lambda is ever commanded leaner than 1.00.
Spark freezes too, which generally only happens on decel. So it gives me the impression it's going into decel mode?

Well I hope you get it figured out as I recently picked up a 2014 S550 that I would like to tune and have read a number of your posts. I have the original MPVI Pro so I'll need to get the MPVI3 in order to get going on this car. It runs great as it is and I'm not looking for any crazy boost in power - just a nice safe bump.

maybe one of the fuel pumps are going bad

What is your rail pressure doing when this happens?

The key item is it's commanding the condition. The command is usually a random # but 1.343 is a number I've seen it settle on a few times. It's even more rare they both read the the same, which is why I took the pix. I was thinking maybe that # is a clue?
Rail pressure is good, imo. It might drop 25, maybe even 50psi under load, but that's it. Plus I can see the Injector pulse width drop when this happens. Sometimes the Desired and EQ Cmd will read normal but the Inj PW drops anyway. I thought maybe it was commanded and HP just didn't see it, but it did this pretty bad once today and STFT pegged 25 trying to compensate. So apparently it can also cmd less fuel without the command being aware, because this wasn't confusing enough already. Whatever the cause, or how it goes about it, it's being commanded lean.

I can't see how/why it would ever command leaner than 1.00 under load, so a glitch is all I can think of. Especially if it can cut Inj pulse width without the rest of the ecu knowing about it. This lean glitch only changes when the tune is changed, so however that tune is, is how it is and it only changes when I load a new one.

I also have two ECU's, but no change when I swap them. I thought maybe loading the tune was glitching it, so I've been using a charger on it, turn the key on and wait a minute before loading, then after I check for codes to see if the CAN glitched while doing it. It changed nothing but worth a shot.

A new thing I noticed is spark going negative, which I assume is for Cat heating? Oem does that all the time but I turned that off looong ago and that was the end of that, until a few weeks ago when this lean thing got a lot worse. This too is tune related, so a tune either does it or not. If so, it may be mild and only retard a couple deg, or it might go past -20. It almost always does this at or near idle, like oem, as if my fix for it got lost. That's believable because the past few weeks I've been making all my new tunes from a virgin copy, so it's easier to imagine the fix was lost or didn't take. Normally I make a tune from a previous good copy where the fixed worked, so it continues to work.
I've always had issues where I make a change and it doesn't do anything, or doesn't do what it should, so maybe this problem has always been there but now it's getting worse?

Since all this crap stays with the tune and only changes with a different tune, it basically narrows it down to the HP Editor?

Attachment 145862

Do you get misfires when this happens?

I would imagine since it's falling on its face with a lambda as lean as 1.99 sometimes.
Does it register misfire code? Never.