Log markers and instant autotune

[Jadzwin]

Hi!

Some information about new features (version 1.110)

One of the new features is log marking. Anywhere in the client the user can press M key (could be redefined).
Then on the log marker is placed, so it is easy to come back to analyze later.



There is also an option for instant autotune. The idea is that the user is able to auto modify cell / cells based on current lambda(afr) and target lambda (afr). For future releases there will be more sophisticated autotuner based on the log file.

First of all Follow cursor icon on VE table must be selected.
The following cell(s) turn into blue. If the cell(s) change color to red it is possible to autotune it.
Just press K key and the VE value(s) will be adjusted. Remember to autotune when there is no transients.

Here are some examples:

No autotune allowed


Autotune allowed (best case)


Autotune allowed

[astra vxr]

do you have to press "k" every time you want to use it or do you just press "k" and drive and it will do everthing for you ?

[Jadzwin]

Press K key when you need to adjust the cell.

[astra vxr]

then thats really going to help a lot you might as well sit there and ajust the numbers yourself

better version would be turn it on and go for a drive and it would add or remove fueling based on afr table and you dont have to touch a thing just drive this is what megasquirt has and to be honest it was bloody brillent of lower end of the table tuning if your own your own ( never trusted it high end or boost side always had second person for that or put it on dyno ) same as the log viewer just load up log and fuel table and if wanted it could change the fuel table based on log
all be it 2 small little things but they do make a difference

your going the right way with it though im not knocking it keep up the good work

[ATS_Scott]

Yes! Thanks Kuba!
Now I know what the red and blue mean!
You mention AFR and Lambda tables, so I assume its active in all fueling modes?

[bmwmaster]

Really good to know. As already wroten, it would be nice to tune automaticly while driving. Maybe the changes could be done to a spare VE Table and then manualy transfered to the real one.

Now you always need two person to tune on the road, but when you don't know anyone who knows how ecu works...........
Also letting drive someone the car is not always sucessful when the person don't know what is important.

[samvel]

Really good to know. As already wroten, it would be nice to tune automaticly while driving. Maybe the changes could be done to a spare VE Table and then manualy transfered to the real one.

What You describe is not exactly autotune, but rather learning. Learning map compares real vs target AFR/lambda or saves closed-loop correction values to map. User can define conditions which let save only true corrections and then transfer them to fuel base map. Some ecus have this functionality, but i find autotune much easier and faster to use, You tune base fuel map directly live.
Also for me it is veryvery important that every function is easy to use, with as less tables and parameters setup etc as possible.
Here is an example me tuning Autronic ecu with autotune steady state on Dynapack. The "middle" of the map can be autotuned in just 10-15min.
https://youtu.be/qMsGrHamJSs
And on this video autotune is quite slow, because i have used Innovate lambda here and tuning accuracy was setted quite high.

[samvel]

One of the new features is log marking. Anywhere in the client the user can press M key (could be redefined).
Then on the log marker is placed, so it is easy to come back to analyze later.

Very good, but can be even better !
For me it is good when many things i need happens just as i want automatically, without pushing keys, without scrolling and searching needed part from log.
Have to just let define, what conditions start/stop logger automatically. I dont care about unnecessary data, dont need this stripchart logging all the time and making my work just harder. It works so nicely in Autronic, Link/ViPec, probably some others. With every dyno pull logger starts/stops automatically and only this pull with previously selected parameters are just in front of You in the lower part of the tuning window. Now we are talking about controlling data and effective tuning.
I am just on tuning trip in Poland and have to tune ecu, thats logger using is hard as hell. It`s just wastes a lot of time.

Edit: 1 more thing. When have nice logs coming pull after pull, very useful (dyno)session logfile can be saved.

[RHD]

Yes i like logging limits too I use that on on other ECU. It would be very handy to have extensive options like rpm, temperatures, load, TPS and one for AFR outside "x" error from target etc. Very handy for final fine tuning. Then only the very areas that need attention would be saved. In a perfectly tuned situation you would come in after a race or a test drive and there would be no logged data. Usually they throw a little flag at the start and stop of every segment to identify a new piece of data. Looking forward to the external logger dongle for this too! very handy to perform spot checks on the tune without sifting through miles of streams etc

The start of the auto tune function is is a long awaited piece of development and I like the direction its heading but i must also agree with previous comments that having to use the keyboard is not ideal. Automatic two stage indexing will be best, starts of with big increments when the error is large, (first cell color) Then switches to smaller increments as it approaches the target (second cell color). That way if you are road tuning or if you dont have a dyno for steady state running the fuel map will quickly get roughed in before being refined to a more accurate level. Once a cell is within the required tolerance it automatically locks (third cell color) so the function doesnt try to keep retuning them over and over again each time the cursor passes through? Any cell should be easily toggled back to un-tuned if you want to redo any.

ps heres another vid of some autronic autotuning https://www.youtube.com/watch?v=L7I4Zn0sbFc

[Wuniek]

Hi,

today i tried new auto tune future but I'm a little confused now, when i use "K" key my VE is fill with 0
short movie how it is working on my setup, before i start rec i changed all VE map to 50 and I'm working on second table set : https://drive.google.com/file/d/0B0eBwU ... sp=sharing

the question is what im doing wrong ?

[Kisielisodd]

It works only in advanced mode with lambda at this moment. I will be expanded to AFR target in short future.

[Wuniek]

It works only in advanced mode with lambda at this moment. I will be expanded to AFR target in short future.

ups i did not save config before i on/off power
i will check it afternoon ...

so i rescaled my AFR to lambda but i will be so much easier if we could just divine whole table by 1.47 ( or i miss divine operator

[Wuniek]

ok i checked it once again and it works

however, it is still hard to comprehend when you simultaneously try to drive, it will be relay great if we can do it off line working with log.

[Jadzwin]

Divide = Multiply by 1/x So just multiply by 1/14,7 I add division for next release.

[Adamw]

Both are welcome features to me.
To give my opinion on the autotune feature, in my experience this is the best way to do it. To the others stating "it is not easy to do while driving", it is not designed for that scenario - it is designed to speed up the fuel map tuning using steady state load control on dyno. You hold the engine at constant load and speed in the centre of a cell, hit the K key and it is done. Move on to next cell. Motec "quick lambda" or "Q" key is very much the same thing and it makes work easy for us tuners (if you have a dyno). I tune a lot of Autronic and Motec and I find the Motec approach of hitting the Q key to tune a cell is much better than the autronic implementation where it will automatically start modifying a cell and sometimes you are not quite ready for it.

In my experience the "tune while you drive" or "fuel map learn" features that some other ecu's offer rarely work well. There are just to many effects going on in the background while driving that distort the measured lambda - Gear changes, transients, decel fuel cut, revlimits, misfires etc, that always end up producing a very messy fuel map.

There are some implementations of "autotune from logged data" that work well - for instance Autronic and ViPec. With these you can select a suitable range of the log to take the data from (where there are less errors), then you have configurable settings to filter out transients and some other conditions and once the data is extracted you can even choose which cells or area that you want to apply the corrections to.

So in my opinion an "autotune from logged data" feature would be nice one day - but don't even bother with "learning map/long term trim" type autotune features.

[Wuniek]

I agree with the above I checked the auto tune in only few places on the map and not much deviate from what was already there, it will handy to have greater interaction while working with the log ( replay run from log ) because not everyone has dyno

[tmus]

Looking forward to the log autotune feature in the future and the advance strategy using AFR defined...

At the moment i haven't tried the advanced fueling as i got my cruising fuel OK now, and the power fuel looks quite good. so kind of just enjoying to drive.

having new features means this is a never ending game! I guess i'll be forever tweaking the car

[Jadzwin]

We do not plan to add AFR for advanced strategy. Lambda is the correct measure and it takes 10 minutes to change thinking from AFR (what is fuel dependent) to Lambda what is fuel independent.

This is the very firs version of "quick tune". We will work on it and there were 3 option, manual quick tune, automatic quick tune and autotune from logged data.
Personaly for me the 3rd option is the best, because I can fine tune the VE table and have full control over it.
Every autotune algorithm is sensitive for local minimums and maximums what could lead to make inappropriate VE table.

[astra vxr]

thats how we do ours useing log and the program (ve analyzer) load log ,afr , and fuel table and i would alter to suit the afr based on log and it was bloody good for us only doing 10sec passes not ideal in the real world but it worked spot on and for us auto tune will never be quick enough for this if its for road or dyno mapping then it will be fine but if we had to pick one it would be log over auto tune just need to get my head around the lambda now

also when using full auto tune you have to turn off everthing like any enrichments otherwise it just messes up the fuel map and some times that can be a pain log tuning for the win

[PSI-Motor]

Quick guide for people using AFR.

14.7 = 1.00
14.0 = 0.95
13.2 = 0.90
12.5 = 0.85
12.0 = 0.82
11.8 = 0.80
11.5 = 0.78
11.0 = 0.75