Ok, I did some internet searching and info is hard to find but found this info on 4bar sensor 0281006060:

200mbar - 0.215V
3408mbar - 3.803V
4000mbar - 4.461V

Plugged both ranges into formula from page 352 and they both came out to 894mbar/V or 447mbar/V (vs 430 as calculated based on 2.5bar specific scaling info extrapolated to 4bar) when putting it into bin with 5120 hack in.

That's still doesn't account for my 80mbar difference at atmospheric pressure between two variables especially that they both are calculated using DSLGRAD and DSUOFS... something (missed in 5120 implementation) must be altering  pvdks_w   
 

Ok, 4-5mbar POSITIVE (looked at several idling logs), but not 80 negative as in logs posted above... but this is irrelevant to this discussion since the behavior you're bringing up would add to the MAP pressure... and since it does add a few mbar, the real gap is even wider.

So... any thoughts? Some minimums kicking in for MAP sensor maybe? I would log B_lde, z_lde (recalling names from memory, whatever they are called in the section calculating sensor pressure reading) and several other variables to narrow this down but have no location addresses.

Both psu_w and pvdks_w use DSLOFS in calculations and psu_w appears above 1000mbar for sunny weather I have here today. Again, the problem is that using the same translation table and offset, both variables have much different value when idling / at ambient pressure (plus few mbar as you pointed out) so something is changing/disregarding/correcting pvdks_w in its path probably due to some minimum we missed.

psu_w?

udsl_w * DSLGRAD + DSLOFS, gets filtered to pvdksf_w, and compared to pvdkmx_w.
If pvdksf_w > pvdkmx_w then pvdks_w = pvdkmx_w and B_edsl is set.

Else, pvdks_w = pvdksf_w and B_edsl is cleared.

The only changes required in this path are DSLGRAD, DSLOFS, and PVDKMX.

The filter doesn't need to be touched.

PVDKMX shouldn't cause the symptoms that you're seeing.

That would explain it. I didn't catch this subtle difference when looking at PDF... so this is missing from 5120 XDF then.

Anyway, looking at your XDF from a while ago (auto generated), DSUOFS is zero... My DSLOFS is -4.10hpa (as per bin posted earlier). That's still only 4.1mbar difference between values so unless location of DSUOFS at 0x13DEE is not right...something else is at blame.

Please advise.

Edit: can this offset be as high as ~40-50mbar? If so, taking inaccuracy of DSLGRAD value, it can be made to work.
Edit2: NEver mind edit #1, changing DSLGRAD would keep the gap as wide anyway.
Edit3: just noticed that both variables don't share anything, DSUGRAD vs DSLGRAD got me too...

You're must be right. I will first address DSLGRAD to have actual proper value and see how much I am off. It has to be the offset afterwards. I was confused, yet again, and not noticing that the two varaibles I am comparing are using two completely tables . Dyslexia hits: DSLGRAD  vs DSUGRAD

Here's an updated XDF with more axis tables for Boost Section items fleshed out and linked.
General trend has been to take axis definitions and move them into the Axis Sub-Section.
Additionally:  DSLOFS, DSLGRAD, DSUOFS, DSUGRAD have been cross-checked and verified to be correct versus PRJ's auto-m-xdf.

I'll continue to update this as we find changes.
Our goal here is to get a completely simplified list to do the swap with little effort.

Summary:
DSLGRAD: 0x13DEC
DSLOFS:   0x13DEE
DSUGRAD: 0x1BE68
DSUOFS:   0x1BE6A

I solved the problem with leaning out while taking off some throttle, KFVAKL map was at fault, I reverted to stock one few weeks ago and it was a mistake and large injectors mine anyway) require it to be essentially zeroed out to not cause issues when transitioning to falling loads...

So cross that one off.

I also adjusted DSLGRAD so that both pressures are equal at engine stopped. I ended up using 461 hpa/V for DSLGRAD and 3.91 hpa (positive) for DSLOFS.

Here are my theoretical calculations per data found on internet and me 7.1 manual:

Sensor linearization:

200mbar - 0.215V
3408mbar - 3.803V
4000mbar - 4.461V




Stock example:

             delta y       (2500 hPa - 200 hPa)
DSLGRAD = ---------    ---> ----------------------    = 541 hPa/Volt
             delta x       (4.65 Volt - 0.4 Volt)


4bar MAP:

     4000 - 200
   -------------  = 894.95     ----> 5120 hack -----> / 2 = 447
   4.461 - 0.215
   
   
     3408 - 200
   -------------- = 894.09     ----> 5120 hack -----> / 2 = 447
   3.803 - 0.215
   


DSLOFS:

Stock Example:
   pvdkds = udsl(200hPa)*DSLGRAD + DSLOFS ---->  DSLOFS = pvdkds(0.4V) - udsl(200hPa) * DSLGRAD = 200 hPa - 0.4 V * 541 hPa/V = -16.4 hPa

4bar MAP:
   DSLOFS = 200 - (0.215v * 894 hpa/v) = 200 - 192.21 = 7.79     ----> 5120 hack -----> / 2 = 3.895



I was still off by some 20-30 mbar after this so I asdjusted DSLGRAD until the pressures matched, I didn't bother with properly scaling DSLOFS since it would only change by 1mbar if that, no point touching it. Sensor's measurement error is much higher than this.