FastHatch

I would like to discuss how a autotuning feature should be implemented in FC Tune. The result should be some kinde of pseudo code.

Here are my ideas:

The fuel correction map can be used as a target AFR map, that is what the software is aiming for.

o2 feedback has to be disabled.

A wideband must be connected to the box.

The current injector pulse width and injection correction coefficient can be monitored.

Only the fuel base maps are modified.

Does every wideband have to be calibrated individually or would it be possible to make a pull down menu with default values?

jamespond24

iTrader: (13)

Autotuning is the way to go. Keep us updated if you have something going?

87GTR

iTrader: (61)

I think the auto tune should use really safe timing too. if the user wants to get more aggressive with timing then let them make the adjustments.

Nzo

Fasthatch have you decided on a good way to do auto ignition timing or is this only for AFR?

FastHatch

Only AFR, timing has to be done on the dyno. Street tuning the timing never gives you the best result. Also for timing i have not the feedback like for AFR.

Also the name autotuning isn't really correct, you tell the software the desired target AFR and it just trys to find the necessary injector pulse width.

Autotuning the ignition actually means the software has to find the best value.

fritts

iTrader: (5)

Since there isn't a good way to tune timing. Are you going to create a table that can retard timing. It probably would have to be based on the change in reading of the knock sensor. Would need to be adjustable because my 40 max reading for knock can somehow be the same as someone elses at 120. Also if possible have it ignore shift points. I don't know how you would do this possibly look at throttle change or maybe use a length of time the knock has to be at a specific max level and ignore the value.

TailHappy

Whew, this would really be a cool feature, but requires a tremendous amount of consideration.

I personally would want to limit how much the software is allowed to mess with the numbers. Such as any particular cell can only be changed, 5%, for example. I would also like to be able to review what changes it has made. I'm not sure how all this is going to work since you said you would be using the correction map to tell it what target AFR, and a lot of us guys like to normalize our correction map to all ones for ease in tuning.

Also, there's so many correction factors applied to the maps (air temp, water temp, etc), it seems like the software could find itself chasing it's tail, so to speak.

Another idea might be to lock it down to a particular portion of the map. For example, I don't mind it messing with cruise or idle areas, but leave the boost areas to me.

The biggest concern is that our fragile rotaries can't handle any mistakes. Leaning it out just a little too much could quickly mean the end of that engine....

FastHatch

@fritts: Please post other questions in the FC Tune thread. Thanks.

So you would like to have a different map beside the fuel correction map, say a virtual fuel correction map? Do you want to set a overall limit or for each column/row? Same limit for enleanment and enrichment?


The engine must be at operating temperature and there must be no throttle movement while the software collects the samples. That way the correction factors are minimized. The software can monitor the current injector pulse width and injection correction.

This could be done, cells or portions of the map could be selected and then marked or unmarked.

That shouldn't be a concern if you set a upper and lower limit what the software is allowed to change.

87GTR

iTrader: (61)

sounds more like self map mode or auto map build mode

RotorJoe

Great topic, I'm listening and very interested.

TailHappy

I would be satisfied with just an overall limit, though if you really wanted to get fancy I suppose a complete extra map would allow the most versatility. I just have different comfort factors with what I would want to allow it to do based on engine load.

So you would pretty much need to go out and do specific runs to get it to autotune, correct? Because during normal driving I'm constantly moving the throttle, other than at cruise. However, air temperature would still be a factor if your correction tables weren't right (which I don't think they are from what I've seen)

Perfect!

Exactly, but that's why I think limits are a necessity. I can't afford to risk it removing 20% fuel from my boost area because my wideband crapped out or something...

Sounds exciting! Nice to see something new again for a 10 year old car.

Howard Coleman

iTrader: (6)

you are on the correct path as to auto tune...

i suggest a 5% limit which will work well as it does cap the change. further, it is common to move a fuel cell 10% or so but since autotune will be enabled more often the 5% won't prove a problem.

OTOH, if it wouldn't be difficult, perhaps we could stipulate a max change /cell.

congrats on your efforts and i look forward to moving ahead w FC Tune.

twokrx7

iTrader: (1)

I would suggest a mode where you make a run or runs with auto-tune off and ask the computer to suggest a change in the maps to achieve target AFRs, then you accept the changes or modify them. Perhaps with some helpful functions such as a menu driven maximum percent change or percent of suggested change etc. the manual auto-tune can be made in baby steps safely. Before accepting changes it should provide an easy to read table of AFR values per cell including average AFR, number of samples, as well as min and max AFR reading. Another screen would give the average AFR, target AFR and the seggested change to achieve target. This would take out some of the risks of improper interpretation of the AFR data if auto-tune changes values while you drive.

TailHappy

Another vote for twokrx7's suggestions.

jpandes

I'd like to see this in action! Maybe on someone else's rotary first

neit_jnf

iTrader: (17)

I tried the same topic some time ago and nobody seemed to like the idea

https://www.rx7club.com/engine-management-forum-37/why-doesnt-anyone-make-self-tuning-stand-alone-ecu-452003/

How come they like it now??

TailHappy

I like the idea now because for one, we actually have someone capabable and offering to do it so it's not just a pipe dream, and also it's being created from scratch so we can all put in our input on how exactly we want it to work.....