So, it's another fun issue from the car I'm working on.
Mostly stock RS6, converted to MT. I've been working on LDR PID when I noticed in logs that sometimes msdk_w is all over the place. I checked ps_w - and for some weird reason it's also reading lower than it should. Could unmetered air cause this?

MLHFM, MLOFS, KFKHFM, MSNWDK/WDKMSN etc are all stock. For shits and giggles I tried the tune in question on my own car (modded TVUB/KRKTE to make it work with my injectors, also set EBC to zero, everything else left alone) - and made a few pulls. Couldn't make a solid log since it's so cold & slippery here - and I boost ~1.3 bar on wastegates, but it -seems- to behave somewhat decent. At least I didn't catch such crazy oscillations. This means, it's more likely to be a hardware fault.

I'm posting a couple of ECUxPlot screenshots of the problem, along with a log I made it with. Don't mind the "not-so-good" LDRPID, it's still a WIP.
I'll update the thread as I progress through the problem. The next thing I have in mind would be logging all "components" involved in forming of the msdk_w and figuring out which "step" causes this behavior.

Anyways, input is appreciated!

delete please

EDIT:

For those that care, I have mostly fixed this by making msdk_w and mlhfm_w go closer together. Apparently, ECU reacts to them being far away from each other by these crazy msdk oscillations, being unable to filter the alpha-N output.

HOW did I fix it:
I have logged ps_w and pvdk_w, isolated WOT regions, then removed remaining "transition from vacuum" - basically filtered ps_w over ~1030 mbar and modified KFKHFM accordingly - in such a way, that ps_w is just slightly below the pvdk_w. It also automatically brought mlhfm closer to msdk. Now when I think about it - a more "appropriate" way would have been to also tweak KFURL/KFPRG in some regions, because AFRs are just a bit (3-5%) off in some areas, and I had to fix those with FKKVS and KFLF. I will most likely tinker with it later.

Now, before attempting this, we have verified that the problem is not hardware: re-checked MAF seals, swapped both MAFs from one side to another, tried changing out MAF inserts for the spare one. Also, we carefully logged and watched AFRs (WBO into 5v input).