Try with DSLGRAD 437.5 and DSLOFS 31.25

I use the same sensor as you for my drag car. At first I had the same problem , I searched linearization of the sensor but I have not found it. Another 4 bar sensor which I've used very often but in diesel cars with BT is 0 281 002 316 and for this linearisation is: 500mbar at 0.5V and 4000mbar at 4.5V. This calibration we used to my car and the result is satisfactory.

Cool. I will give it a try. A long time ago I found specs for map sensors on bosch site and the one you listed struck me as a fit for our cars:

http://rb-aa.bosch.com/boaasocs/index.jsp;jsessionid=A33AAEE16429B168F3E582158D1DE99D.sundoro?ccat_id=110&prod_id=203

It makes sense that they would use common chip for measuring pressure (motorola freescale?) and only stuff it into different part# for specific applications.

Coincidentally, that sensor looks like a better fit for our cars since it is a 'fat' inlet from factory?

Roger... but I wanted them to match up after using specs not working them backwards from differences between pressures. Sounds like the official specs for different part# 4 bar MAP should work 100% accurately and I will check that in couple of hours.

Ok, confirmed. 437.5 and 31.25 (thanks to 99pwr for pointing me towards this direction!) which are calculated values for bosch 0 281 002 316 are spot on for this one. My pressure differential with engine off is 2mbar, that's 0.2% discrepancy and essentially dead on within reason considering there are two separate sensors measuring these...

DSLGRAD = (4000-500) / (4.5-0.5) = 3500/4 = 875  ---- 5120 hack ----> / 2 = 437.5
DSLOFS = (500) - (0.5 * 875) = 500 - 437.5 = 62.5  --- 5120 hack ----> / 2 = 31.25

These are set to whatever 5120 bin had in it. It seems spot on, 1008 mbar today, sunny here.

Let's talk PID control.

Am I right that the PID control will be all messed up with the 5120 hack? Seems like KFLDRQ2 axis is changed (most important one) but there are other parameters and they all have output in %/hpa. Since we are halving hpa change by half, shouldn't the tables be multiplied by two then?

List:

LDRQ0S
LDRQ0DY
LDRQ1ST
LDRQ1DY
KFLDRQ2 (done?)

Without Q0 and Q1 changes, the PID will have hell of a time (or fail altogether) in keeping desired boost.

I did a quote from ME manual which was an excerpt and out of context really. The outputs from tables/scalars for PID control are converted "change from %/100hPa to %/hPa" (divide by 100) with a footnote that the part I copied "%/hPa", should really read "% / 1hPa".

As such, they are unit of pressure dependent. When ECU is reading 50hpa difference and tries to match it, it calculates change in DC for 1/2 of what it should be using which is 100hpa. Hence the quotients should be multiplied by two, I assume unless I am not thinking straight (keep in mind I don't really know how this is all calculated so don't blast me!)

Anyway, my empirical preliminary pulls indicate that PID controller is unable to control boost on my 770s, once it overshoots it never pulls out enough DC to even stop boost increase. It might be my completely inadequete KFLDRL which I need to still calibrate, but I suspect PID controller is out of balance as far as dynamic control goes as well as the quotients are not scaled to reflect 5120 change (well, one is but the others are not).