RomRaider

These are the values that I use.. I vary the data collection value from 100 - 300 depending how long of a drive I've been on.


So now, if the system logs over 100 lines of any point within 100RPM and .5 psia of the cell value it will make a correction.
If the result has only been changed by <1.5%, then it won't bother applying to the output table.. Saves jumping back and forth between +1.5 correction and -1.5.


Two other things..I am curious though if closed loop corrections line up with VE corrections.. Does a 5% correction in CL mode equate to a 5% change in VE?
When I suggested that feature, I was thinking more along the lines of ignoring logged data that registered < X total correction..

More importantly..
Also, is it possible to change (or have the option to) go from a correction multiplier based on the average values per cell , to one using mean?
Averages are much more effected by outliers like tip-in/out, where as mean will give a result based more on the commonly occuring values.

Are you finding the tool to help you in close loop?

I had to manual go in and adjust stuff that the tool would increase when there was no need too.

The log attached if of a car being started for the first time and cold. When it does, it cuts out. Previous tool runs of older versions of data stated the areas highlighed needed reducing.

Are you using a wideband?
What are your expectations from this tool?
If you're having running trouble (I'm not entirely clear if you are) this sheet is unlikely to help. This is really more for tuning rather than diagnostics.

This tool isn't going to work with a log from a cold car. I'd recommend warming it up first.

In that log, you've only really got a single cell of good logged data, it's going to be impossible to do any meaningful manual changes if you've only got one cell to work with.

With your "Max (%) throttle change to apply Closed loop adjustments" set to 1 and "Minimum throttle angle for OL calculation to be applied" set to 0, it's ONLY going to make changes for open loop and idle. If you do any logs while driving, you're going to need to change those settings.

If you're problem is with the engine running, maybe start another thread in regards to that, once you've got the car running, then come back and use this tool.

No I don't have a problem and can tune manually with out it. But what I was surprise of how in either selecting the OL or CL, the changes that occurred. Otherwise its a great tool.

Did we find a way to make the spreadsheet work on mean corrections, opposed to average?

Excel's AVERAGE() function takes the arithmetic mean. The filtering performed in this sheet *should* remove *most* outliers. But I see your point.

Coincidentally I have been planning on using some statistical analysis in a spreadsheet for semi-automatic ignition timing tuning.

I used the technique for my honour's thesis and it worked quite well for single data points, I'd just need to scale it up.

Maybe I could use the same approach here.

Watch this space!

There seem to be an issue with the Max Distance +/- from RPM table..

If I set to any normal value (Lets say 50, 100, 200, 500, 1000 etc.. It's cutting out a LOT of data.
If I set it to "-", LOTS of my logged data come back into the counts of data points.
However this sets the entire changes multiplier to 1.0000

Any ideas?

Nvm, Fixed.

Sorry I missed this, what did the issue end up being?

The datalog was using 0 for OL and 1 for CL instead of the typcial 7 8 and 10.
Had to just change the forumla to except the new numbers

I added a multiplier to reduce or increase differences, and added a VE Smoothing table from another speadsheet i found on here(i think). i dont recall the name of the user but ill credit him when i find the post.

smennie can you clean up my modifications or see if they're useful in your sheet to others.


This has helped speed things up for me. i dont have to spend so much time in ecuflash multiplying and interpolating the VE table to smooth it out and switching through spreadsheets.


Hope this helps

.

I like it!
I've actually got a smoothing spreadsheet as well which I should have shared...or maybe I have. It has a couple of 3D plots in it as well to visualise the difference.

Just a quick one gentlemen, I've recently switched over to full time OL, and the spreadsheet is still ignorint a huge ammount of data. I can't even see any differences between data it determines is fine to apply a correction multiplier and data it doesnt.
This was my drive home from work, so there was quite a bit of idle time in traffic.. Yet it's saying that I've done hardly any low end driving (Even with the filter settings set absurdly loose)

Photos attached
Any insight?

i286 has done a huge amount of work on the sheet, and he picked up a lot of issues like this, and has increased the range in which open loop calculations are done.
He sent it through to me last week.
I just need to do a final test before we release it.

Yeah still an issue with the RPM filter, puting a simple "-" in that cell brings back all the data.
Drama is it no longer applies the correction.