In your hypothetical situation, I agree that you would expect a flat load line. This is what you would get though because as you get leaner the IPW value would reduce, as would the AFRMAP value.

Yes AFRMAP decreases - in decimal 128 is 14.7:1, 12:1 is about 157.

Agree it is better to find real airflow.

Agree IPW corresponds directly with massairflow/rev only at constant AFR. That is why dividing by AFRMAP should work, because it corrects for requested AFR.

Thoughts once more?

Appreciate the discussion on this.

 

Ahhh...I didn't realize that the decimal value decreased as AFR value increased. (see my second post above).

In that case, I think this will work fine and should be included once refined and tested some more so that we can have map tracing!


Eric

 

Bingo. I think so too. Just doing an Excel graph to post and will put the csv log in and xls with calcuations and graphs. Give me a few minutes...

 

Excellent! Can't wait to take a look.

If this turns out to do the trick and we could get map tracing working, then I can finally concentrate on proving whether the Evo ECU acts like the DSM ECU in knock correction.

Either way, we can prove once and for all when and how much timing is pulled due to knock.


Eric

 

Here I have highlighted by scaling to show the difference between my 3 bar MAP sensor reading and the calculated load = 2550 * (IPW-0.46) / AFRMAP

As expected as VE reduces gracefully at high revs (stock turbo) the graphs diverge.

Will see if I can find a section that has transients and mixed load/revs.

 

Here is the zipped folder - MalibuJack here is the .csv with the raw log in it. Also is a .xls with the above graph and you can see the formula used.

 

Look good to me.

Hopefully we can get some more testing and validation and see what we come up with.

So far so good, though.


Eric

 

Here is mixed driving, showing 3 bar MAP sensor, new load calc and 852*airflow/RPM. Notice on over-run the new method doesn't work when the injectors cut out - it goes negative because the injectors are zero, but the lag time is still subtracted.

 

Correlation between 852*airflowhz/rpm load and new load works until 852 load drops when airflow Hz clips at 1603.

 

Care to share the Data.XML file with which the csv log was taken? I saw a category for calc load in the log. Was that derived after the log was taken or is it in the Data.xml file?

 

Calc load is done after in Excel.

MAP is JDM IX only for the 3 bar MAP.

 

Better calc for US 8 & 9 is:

calculated load = abs(2550 * (IPW-0.4) / AFRMAP)

0.4ms is the estimated injector lag time at about 14.4V which it is likely to run. Abs gives the absolute value in Excel to stop it going negative on over-run - it will just show 0 load.

 

Come of you clever Americans Go burn some more hydrocarbons and tell me if it is so

 

just out of curiosity (and my own lazyness being at work), how are you calculating the IPW?

 

It is one of the parameters logged in EvoScan (request 29)