[Green-Jimmy]

Can someone tell me how to arrive at the correct values in some of the tables? I know its a pretty basic question, but I want to make sure I'm not making wrong assumptions.

For example:

FN071 - Load Normalizer for Spark, EGR, etc..
FN904A - Spark Sealevel Table

If I have a load of 5 the Load Normalizer for Spark shows that I will be using row 0 for the FN904A Sealevel Spark Table. Now, if I have a load of 6, will the normalizer function put me in row 0 or row 1 of the Spark Sealevel Table?

The second part: Once I know what row in the Spark Sealevel Table is being used, I can use my RPM to find the value. Now, if the RPM value is 1100, then I know exactly what column ( since 1100 is one of the columns). But, what happens when the RPM value is 1200? Are the values of the 1100 RPM column and 1300 RPM columns added and then averaged?

Thanks,
Jeff

[cgrey8]

To answer the 1st question, I believe the EEC extrapolates between cells for values that are between rows. So for instance if you have 5 in the 10% load row and 10 in the 20% load row. A value of 15% load would result in a 7 (half way between 5 and 10).

The same extrapolation is done between columns as well.

[Cougar5.0]

Yep, the ECU does linear interpolation between adjacent cells in rows & columns.

[sailorbob]

Just to add to what's been said above;

Taking a simple two column function like the A9L's 'Idle Adder for ACT' you have this;

254 72
226 72
200 48
176 0
-256 0
-256 0
-256 0

In this example if the ACT is say 188 degrees then the output value is 24 rpm, this is nice and easy and you could guess the result. But if the ACT is 198 degrees it gets more complicated (not a great deal with this example but that's why I picked it!) but the formula the eec uses is a very simple interpolation;

value out = lower output + (((upper output - lower output) x (value in - lower input)) / (upper input - lower input))

So for 198 you get;

value out = 0 +((( 48 - 0 ) x ( 198 - 176 )) / ( 200 - 176 )
= 44 degrees

For tables, the eec first interpolates the columns, then the rows and lastly between the two results.

[Green-Jimmy]

The linear interpolation makes sense in the case of the single table references. But in the case where a normalizer table is used first and the output of that table references another table is where I'm still not sure of.

Like in my example above where it load is being normalized, I assume the output of that table can only be a row ( integer value ) and not a decimal value. Maybe I'm not understanding the normalizing tables and how they are used.

Thanks,
Jeff

[sailorbob]

The row / column output values you see in your editor for the normaliser functions aren't what the eec has stored in the binary. They will be typically 256 or 16 times greater.

[Green-Jimmy]

So the normalizer FN071 - Emission and Spark Scalar for Load is displayed like this in the editors:

.....LOAD.....ROW
6....200...........7
5.....75............7
4.....60............6
3.....10............1
2.......5............0
1.......0............0

You are saying there are really more datapoints ( 16 or 256 times ).

So that table could look something like this ( but with even more datapoints )?

.....LOAD.......ROW
18....200...........7
17.....75............7
16.....60............6
15.....50............5
14.....40............4
13.....30............3
12.....20............2
11.....10............1
10.....9..............1
9.......8..............1
8.......7..............1
7.......6..............1
6.......5..............0
5.......4..............0
4.......3..............0
3.......2..............0
2.......1..............0
1.......0..............0

That being the case, if I made changes to the table displayed by the editors, are the editors creating all of the in-between values that at not being displayed?

Thanks,
Jeff

[sailorbob]

For your example, this is what is actually stored in the bin:

6...65535....1792
4...24576....1792
3...19661....1536
2.....3277......256
1.....1638.........0
0..........0.........0

This enables the eec to deal with the interpolation between rows or columns without having to uses fractions or decimal places.