NefMoto

In taming an intitial overboost situation (A6 2.7T w/ FTs), I've been pruning back KFLDIMX.  My tune is pretty mild, 16.5 psig or so.  (That's referenced to atmospheric pressure of 800 mbar/11.8 psia since I'm at 6,200'/1900 m.)    I'm to the point where I've taken KFLDIMX to 30-40, and I probably will need to go a bit lower before this is completely resolved.  My N75 is not stock.  It's an 'H', although I fitted the orifice from the original 'B'.  WG cracks at about 8 psig.

I hadn't seen DIMX reduced this much in other tunes, so my first thought was a torn WG diaphragm or hole in a pressure line feeding one of the WGs.  Nope, all good.  Another possibility is a hole in or popped off reference line from one of the turbos.  It doesn't seem likely, though, since control is otherwise pretty good once the overboost is corrected via PID.  (No sense dialing it in better until the initial overboost issue is solved.)  The last possibility (that I can see) is the N75 itself, maybe in combo with the tightened WG.

Since testing for a bad reference hose is a fair amount of work, and it's dang cold right now, thought I would ask if it is common or reasonable to see KFLDIMX need to be reduced to this degree given my set-up?

What is your WGDC like?

KFLDIMX should be set to the wastegate duty cycle which is needed to run target boost.
However, your KFLDRL might be out of whack.

You might need to use LDRAPP to re-linearize the PID.
Many factors, but a good idea is as always to start with some logs demonstrating the problem! :)

I'd agree with prj and guess your KFLDRL needs attention.

Yes, of course.  Please see the attached.

(http://i1176.photobucket.com/albums/x330/dakota70440/551R_20121226_074950C1_zps0085b0ad.png)
(http://i1176.photobucket.com/albums/x330/dakota70440/551R_20121226_074950C2_zps4f3c8f32.png)

Until I can figure out what photobucket did to change their auto-thumbs, here are full-size images.

You need to linearize your boost control.
Start from scratch.

You need to set b_ldsafw to 1 in CWMDAPP.
After this make pulls with LDRAPP set to various values.

Make a full pull with each value in 3rd gear to redline, and also make a pull until 3500 or so RPM in a high gear, and start as low as possible. Basically floor it almost from idle (1000 rpm).
The values for LDRAPP would be 0, 10, 20, 30, 40 and so on.

Ramp them as high as possible, obviously you will have to take care with the higher values. But I think you can go up to 50 on your setup if you make sure that fueling is adequate.
After you are done, you need to plug the values into an excel sheet, and get KFLDRL and KFLDIMX. As you will have two runs for each WG preload, pick the higher boost values.
The cells that you are not able to visit (higher duty cycles) you can just fill 1:1. So the 80 column will be all 80 for example.
In an ideal world you would want to do that on a steady state dyno, but we do not live in an ideal world, so two range of pulls will have to do.

I attached the excel sheet I have used on some cars with high wastegate preload...
It's not all fancy, and you might need to edit it a bit (axis values and so on, and maybe the KFLDRL higher than 60-70 if you need to), but you will get the idea.
If after that boost control is still under- and overshooting, then you will have to adjust the PID itself, not only the linearization values.
But the linearization values you will have to get either way, so you are not wasting anything.

HTH...

Can you describe a bit more in detail about what is going on in your spreadsheet?

Thanks.

Download it and have a look.

Also read this topic by elRey:
http://www.nefariousmotorsports.com/forum/index.php?topic=517 (http://www.nefariousmotorsports.com/forum/index.php?topic=517)

No sense for me repeating what has already been said.

More backstory:  I had actually had it all dialed in very satisfactorily (relying on elrey's work among other threads) and then the N75 crapped out and I switched to the other version, frankly, to save $20.  I assumed that a bit of PID tuning is all that it would take to regain lost ground...

So from the direction you suggest I go, it is indeed not kosher for KFLDIMX to be set so low?

Thanks!

Thanks. I missed his very nice followup. Added to s4wiki.

Not sure where you're located, but I have an n75 I'd donate to the cause (at the very least to rule it out).

^^^^  Very generous offer, thanks.  If I don't figure this out shortly I'll bite the bullet and buy the correct version.  I'm outside of Denver.  Need to wait for the roads to clear, so no heavy throttle for a few days at least  :-\

No worries! Good luck, and PM me if you need it.

Closing the loop, no, it is not unreasonable to have to tune KFLDIMX down to the 30-40 range depending on build specifics.  Ended up with P-term at 1.8 also, quite a bit lower than I had seen previously (except for RS4).  Linearization was mostly fine, P-term and DIMX were the major issues.  A couple of minor tweaks is all that's left.

(http://i1176.photobucket.com/albums/x330/dakota70440/551R_20130110_073835B_zpsbe65ff2c.png)

@PRJ, your excel sheets generates IMX directly of the measured values, isn't this supposed to be backtracked tru KFLRDL?
Am i missing something?

That's correct, I posted an updated Excel a long time ago.

Ok thanks, missed that somehow! ;D