Tuning with MegaLogViewer VE Analysis

Requirements:

MegaLogViewer 2.28 or greater (The latest is always recommended)
MegaTune 2.25 released version or newer - this uses xml based msq files. Non-XML based msq not supported

How to use:

Drive car through as many throttle positions and RPM ranges ans possible. There is no need to disable Gego correction, Accel enrich or over run drive under the same configuration as you would any other time you are driving the car.
Collect 15 minutes of data or more logging with MegaTune
Save current Megasquirt configuration (MSQ) with MegaTune
Open Log file in MegaLogViewer
Open MSQ that you were running on while logging
O2 selection
- MS1 & MS1-extra - if using an Wideband O2 sensor, select it from the Wideband O2 AFR menu.
- MSII - No need to select an O2 sensor, AFR will be read directly from the log file for WB
- NB All - O2 field will be used
Set the top table to the VeTable you wish to analyze
Click VE Analyze
In the dialog, click Run Analysis
When analysis is complete, the suggested table will be displayed. You can view the original value in the tooltip along with the hit count and weighting. The higher the weighting the more sure you can be that optimal value has been generated. However, if weighting is low, the value should not vary far from original as the original value carries it's own weighting. If weighting is low, try generating another log file with more driving at the cell you are viewing. Not all cells can actually be used though, these cells you are probably not to concerned with.
Using a Mouse right click you can revert any selected cells back to the original value or modify them before accepting the new table.

Advanced Controls

Cell Change - This impacts the amount of data required to move the cell value or could be looked at as the cell resistance to change. If set to Hard, the initial cell value is more respected and will require stronger analysis data to change.
Wide band delay - Adjust Wide band O2 sensor delay, compensates for delay in WB O2 sensor readings If set to 1 the AFR value from x+1 will be used when processing record x. Example: Wide band delay set to 1, when record 241 is processed the AFR value from record 242 will be used. If set to 2, AFR from record 243 will be used. Normally,
Best results are achieved set from 0-2, depending on car setup, O2 sensor, records/sec in your log file and RPM you are focusing tune. Higher RPM's tend to need less delay.
Min RPM - Data Filter, filters all records below the displayed value. To edit, click and enter the new value. This is used to filter out log records where the engine is not running. You can also use this to filter idling log values in a situation where idle AFR is not accurate due to leaky exhaust, big cam or excessive spark advance.
Max RPM - Data Filter, filters all records above the displayed value. To edit, click and enter the new value. By default this is set to high to have impact, but this can be used to filter rev-limiter hits that have impacted AFR.
Min MAP - Data Filter, filters all records below the displayed value. To edit, click and enter the new value. This filter helps to focus on only specific region or area of your VE table.
Max MAP - Data Filter, filters all records above the displayed value. To edit, click and enter the new value. This filter helps to focus on only specific region or area of your VE table.
Start Time - Data Filter, filters all records before the displayed value. To edit, click and enter the new value. This filter allows you to use only a section of your log file for VE Analysis.
Stop Time - Data Filter, filters all records after the displayed value. To edit, click and enter the new value. This filter allows you to use only a section of your log file for VE Analysis.
Min CLT - Starting with vesion 2.69, records are no longer filtered based on the warmup flag. Instead Min CLT was added as a configurable setting. This is the minimum value for CLT that will be used as input during VE Analysis.
Custom Filter - The Customer filter allows for any specialized filter you may require. Usage is a little more advanced, but with this single filter you can do what all the others do plus a great deal more. You can enter any logical expression with field references, the logical expression is executed for every record to determine if it should be used as valid input.
Example: [RPM] > 2000 && [TP] > 80
All records with an RPM over 2000 and TP value over 80 will be filtered from VE Analysis and not used for calculating the resulting table. More information on usage of the Math Parser can be found here.

For the curious, How it works:
The formula used for correction is in the MSGraph.properties file:

AFR_recommendedCellValueFormula=(cellValue*[Gego]/100*[AFR(WBO2)+WBafrOffSet]/afrValue)+(cellInterpolated*[Gego]/100*[AFR(WBO2)+WBafrOffSet]/afrValue)/2

You can likely read through it an see what is happening, but a couple of key points:
- 4 VE table cells will be analyzed for every record not filtered.
- Each of the cells will be assigned a weight that is equal to it's contribution to the end interpolated VE value. This is the same weighting that is displayed as a tooltip when viewing a log file with your table displayed.

In the above screenshot the current record has
MAP=48
RPM=1325
This means the the actual VE value used will be interpolated from the 4 cells highlighed in yellow. The 45% in the tooltip above represents the weighting for the MAP=50, RPM=1500 cell. The total weighting for the involved cells will equal 100%

- Back to the formula, There are 2 primary approaches to calculating the optimal value being used and then averaged. By breaking the formula down, we can look at both components independently.
The Cell Value approach in the 1st section:

(cellValue*[Gego]/100*[AFR(WBO2)+WBafrOffSet]/afrValue)

Takes the specific value of each of the 4 cells and corrects it using Gego and AFR. In other words, if the record being analyzed appears 10% lean, a recommendation will be added to each of the individual 4 cells that is 10% higher, with the associated weighting.
For Example using the above 4 cell diagram, the top left cell value of 23 would have a recommendation of (23/0.9) 25.555 with a recommendation weighting of 0.35

--
The Interpolated value Approach for part 2 of the formula works a little different:

(cellInterpolated*[Gego]/100*[AFR(WBO2)+WBafrOffSet]/afrValue)

This uses the weighted interpolated value of the 4 cells as the seed value for all 4 cells. In the above picture, the weighted interpolated would be:

(23*.35)+(35*.45)+(30*.09)+(32*.11) = 30.02

So using this formula the correction for the record will be applied to the interpolated value (30.02) and that same recommendation will be added to all 4 cells with the cells contribution weighting. Using the same example of that record appearing 10% the top left cell would have an added recommendation of (30.02/0.9) 33.35 with a weighing of .35

--
Both of these methods were used alone in some earlier versions of MegaLogViewer and each method has strengths. The Cell value approach is maintains the cell relative value even when VE Analysis only views it from one side, meaning there are log entries where a cell is the top cell in the 4 cell interpolation, but no records where it is the bottom cell. However, the Cell interpolated approach aids in smoothing the table.
With the help of many in testing, I found a hybrid approach to work best. Thus the above full formula calculates each cell both ways and applies the average with the associated weight. This seems to give the benefits of both methods.

This formula is read from the MSGraph.properties file, so if you wish to experiment, you can edit it there. applies the interpolated value to all 4 cells with the impacted weight