Kodename 47's Knock Correction Tool
 

Disclaimer: I take no responsibility for the tables produced. Please check and confirm all data used within your ROM.

What you need:
- Your log
- @vgi's tool (Usually known as MAF scaling tool - v2.2.9 at last edit) - Github Download Page
- My spreadsheet tool: Download here -v1.07

Before you start, a couple of extra columns need to be added to your log. These base formulas are to be put in row 2 of the spreadsheet, ie. the first line of data.

dThrottle: Change in throttle. This has been tested on ECUtek where logging is done in ms. If your logging is done in seconds, remove the 100 from the formula below. Formula: =IFERROR(((TA2-TA1)/(Time2-Time1))*100,0) Note: TA is throttle angle, not accelerator angle.

KCI (°CA): Knock Control Indicator. Shows the instant any FLKC or FBKC is applied and not the whole time that the correction is applied. It also takes into account the IAM, so if the IAM drops the knock retard events will have a greater impact. This is necessary for using one of the tables in my spreadsheet. Formula: =(2-IAM2)*IF(FLKC2<FLKC1,FLKC2,IF(FBKC2<FBKC1,FBKC2,0) )

IAM Drop: Shows the moment there is any drop in IAM. Formula: =IF(IAM3="",0,IF(IAM3<IAM2,IAM3-IAM2,0)).

Copy these to the bottom of the log. Ensure they are then re-copied and saved as values rather than formulas as CSV files cannot save formulas.

To get good results, ensure you have a long log and try and drive as differently as you can testing as much of the timing map as possible. Short pulls will pull up very little data unless you're really knocking.

Copy your base and advance tables into the spreadsheet where it asks. Again, I have not tested this on RR but I don't see why it wont work so long as your copy method has [Table3D] in the upper left cell and not above and offsetting the engine load row. It will try and take into account if your two table's columns aren't the same.

Open VGI's tool and go to Log Stats. Then:
> Select log
> Set Axis
> Paste Base or Advance table (I prefer base as will most)
> X Axis: Engine Load
> Y Axis: Engine Speed/RPM

For KCI (°CA):
Data: KCI (°CA). Filter KCI (°CA) < 0
> Stats: Mean and Count
Copy the results to the spreadsheet

For IAM Drop:
Data: IAM Drop. Filter IAM Drop < 0
> Stats: Mean and Count
Copy the results to the spreadsheet

The spreadsheet will now look to make corrections where the count is deemed large enough and use the larger correction from either KCI (°CA) and IAM Drop if there are corrections in both for the same cell. It will then correct adjacent cells in smaller increments to attempt to smooth out the table.

My next step is to take the Total Timing and paste that into a tool that allows me to view it and then manually smooth out the table. ECUtek's log viewer is good at this but I know that RR is not so much. However, there is a tool here: http://www.romraider.com/forum/viewt...4439&start=135 . VGI has edited it and it works well. If that's the case, you can use the advance corrected RPM table and new base table for this tool.

Once you have checked and confirmed what it has done, you can then paste the total timing table back into the spreadsheet in the Total Timing Corrections Tab. It will then compare the automatically corrected total timing to your manual corrections and output a new base table.

Any questions, then please ask. I find this a much quicker and easier way than scouring logs.

Kodename47's Tuning Tools Page

@Kodename47 download at work is blocked so here's a quick Q :) :
- does your ss work with different scaling of advance and base tables? (NOTE: the NSFW's Tota Timing tool does not. The EcuTek total timing viewer doesn't either)
@all also, there is another ss by airboys if someone hasn't seen it for smoothing and scaling timing tables (although again i don't think it works if scales in advanced and base tables are different besides the edges - pls correct me if i'm wrong): http://www.romraider.com/forum/viewt...hp?f=33&t=5758

The total timing table does indeed take into account both differences in load and engine speed, it will base all axis on the base timing table. So yes, you can have different scaling. However, no smoothing is done by my spreadsheet.


What I have done on ECUtek is take the total timing generated table and paste it into it's own separate ROM as the base timing table. That way I can view it as one table and just use the plot view to see it and edit it easily. No need to use the ignition view.


I have just edited the spreadsheet slightly so you can set your own FBKC thresholds, I've got it defaulted to 15. Don't confuse that with 15 different knock events per cell, some events will appear in more than one row in a log. If you want to see it as individual events, you could add a FBKC column to your log and use the same formula as the FLKC using Knock Correction in place of Ignition Correction - Fine.

I made a balls up last minute and referenced the wrong table. I've since tidied up the hidden calculation screen and corrected it so give it another go mate.


As I said about ECUtek, I don't use the total combined ignition screen at all. I take the total timing provided from my spreadsheet and just paste it directly into a base timing table using the plot function on that lone table, it then works as it should as no calculations are taking place ;) If I make any manual corrections I then paste the new total timing table back into the Total Timing Corrections tab, let it remove the advance table values to give me a new base table.

awesome, thanks mate

If anyone has downloaded it recently, I recommend that you re-download it again. Link is still the same as the 1st post.

New updated file, lots of updates and improvements. Now has transient retard table and a plot of the total timing.


Direct Link


The plot can be customized for yourself and will work with loads up to 4 :)

link is down

Should be back up now, I've been playing with the smoothing on the knock correction and I had mistakenly removed the file from the folder. This should be the last update and I've left example of knock correction there for people to have a look at.

This is amazing. Thanks so much. Didn't even think this was possible

Typical that I say it's the last update and then I tested a flaw with adjacent cells with similar values. I've since corrected it - V1.04 at the link :D

This seems like a great tool man. Great job.

Would it be possible to upload an example, already edited, log?

I'm afraid I'm not really an excel wizard. So reading the instructions doesn't compute in my brain. I guess I need to see visually what I have to do.

Edit:
Disregard, it would seem it worked by magic! :D

Thanks a lot! Can't wait to redo this with a proper log!

https://s28.postimg.org/mjjqk1iwd/Sc...t_18_33_55.png

I just tried out the tool again more serious using a massive log of 35000 lines.

I am wondering if I am doing it right. Still, it's the Excell log-work that makes the most trouble. If you would take a look at the log that I got the formulas right I would be very grateful. I am pretty sure I did, as the tool provides useful data.

What I wonder, when looking at the log isolated, is that some events are not being counted. Here I sorted all -1.3 FLKC events and in the formula field (which I named FLCK corr) a lot of them have zeros. I suppose this is the purpose of the formula, because of a related field? But I would rather ask to be sure.

The full log is here:
https://ufile.io/fbc37

https://s29.postimg.org/jd6yula4n/Sc...t_00_12_47.png