Simple "quick'n'dirty" Tuning Guide

[Hentai]

Ok so here is my partial way I tune off line.
First set your light loaded areas, anything that doesn't really need any enrichment going on, cruise, mild acceleration, idle, overrun, etc to lambda 1. In the Picture below we can see I have setup lambda from 100KPA and under. In my locale baro is around sea level atmo. I would suggest starting point be at your locale barometric pressure. You can also see I start richening it up from 1 lambda to .9 lambda then to a final .8 lambda.

Another reason you want to tune it at stoich = lambda 1 is because generally no matter what wideband controller you are using this should be the most accurate area compared to being rich or lean of stoich. Thats why if you can you tune for stoich.




This also requires that the data from the wide band and the data log are matching, make sure that is correctly setup in your firmware as well.

Get your engine up to operating temp. I will generally highlight a selection of cells ( depends on software ) and change them globally to the same value to an operating range I can at least drive around in and data log from. Now just drive and data log, if you come into an area that isn't over or under fuel, try to mass change those cells until it can run properly. Once you have a running setup, then start a new data log and go drive around for a while, high revs, low revs, high load, low load, etc. If you are new you can also use the brakes to increase load while keeping the revs in the same location. Also note at lower airflow the wide band reaction to the gas will be slower so there is latency issue to worry about, so if you have some data points but the EGO reading is up and down and bouncing a lot, I wouldn't use those, but if it smooth and stable then I wouldn't mind using those data points.

Gather the data from your drive, try highway, in town and cruising. To note you want steady data, you don't want sweeps or acceleration data to tune based off if possible.

Now head back and grab Open Log Viewer with your favorite drink and a tablet for writing.

We care to see MAP, RPM, VE, Lambda, EGO from the data log at least

on your tablet
you want to write down

Code: Select all

RPM | MAP | VE | EGO | NEW VE

Now go though the data log and try to find areas that match the load and rev cells ( example 3300 revs and 60 map )

so if target lambda is 1 and we see EGO is reporting a steady .898 and the VE is currently say 80 at 3300 revs and 60 map
We take the VE and multiple it by the current ego value, 80 * .898 = 71.84

like wise is we have say 2000 revs and 80 kpa with a VE of 70 and EGO of 1.06 with a commanded lambda of 1

70 * 1.06 = 74.2

So you basically pick out of the data log cells via rpm and load that are close to the ones on the VE map
If the data log shows 59~ kpa and 2400~ revs and the map is 2400 revs by 60 kpa then I wouldn't mind using those points.

For tuning cells that aren't at lambda 1 you take the current lambda and divide it by commanded lambda in that cell

so say current is 1.1 lambda and commanded is .8.
1.1 / .8 = 1.375
You do 1.1 divided by .8 and you get 1.375 correction value.
From there you can take the current VE cell value and mutiple it aganist the correction value to get a new VE value to plug into the VE table.

This also why tuning at lambda 1 is easier as you don't have to do this extra step, it also the same for tuning with AFR you have to deal with doing the same thing to find a correction value, if you were using lambda its easier to get the tuning done faster.

Something I gonna hit on quickly is injector non linear area, this is an area of injector usage that as mentioned won't have linear rise of output. You to be careful as going up or down ve here may cause different than expected output. On my current ID1000 injectors generally under 2ms of pw I know I am in this region so if there is no way to compenstate for this region my ve table may look odd at these points, don't be alarmed.



With this way you can do most of the map tuning without having to be in the car and fumbling with the PC, esp if you don't have a partner to help you out.

[Fred]

Mostly good advice! And simply put, thanks.

One point, no need to mess with your lambda table (which is YOUR target, NOT a target for the ECU) if it's well setup before you start, which is essential. The above method works fine with any lambda values. Both default curves in the firmware are very usable, and kinda similar. Take a look and tweak to suit yourself BEFORE you start doing anything. Before first start, before first firmware load, etc. Lambda == 1 is good advice for an initial tune between say 70 kpa and vacuum across the entire RPM range. Once tuned you can lean it way out in vacuum if you wish. Obviously higher kpa values should be richer than lambda 1 for best power/torque and safe operation. You can still use this method on those sections, though :-)

Another thing, once done with the changes be sure to smooth out intermediate values that didn't get tweaked to match the changes that you made. Otherwise it'll be rough and bumpy.

A final thing: Hentai alluded to this, but didn't explicitly say it. Your log that you base this process off should be a contiguous log, all of which is warmed up and without any tune changes in the middle. If there are tune changes, truncate the log and only use the part after the last change.

Fred.

[Hentai]

Gonna update this with picture(s) and include more info for tuning cells not at lambda 1.

[Hentai]

Updated my guide

"Another reason you want to tune it at stoich = lambda 1 is because generally no matter what wideband controller you are using this should be the most accurate area compared to being rich or lean of stoich. Thats why if you can you tune for stoich."

[Hentai]

I met emstudio author in the flesh and it mad me realize I need to add some more additions to the tuning guide for others thats have less experience tuning ecu's.

[Hentai]

updated about injector non linear region

[Hentai]

Provided that the injectors are modeled correctly in the ecu software
Generally there is things to note about the model of the VE table based on if we are a normal engine ( no wild cams, or lack of vacuum for SD )

going up in KPA to atmo results in increasing VE, you should never see a part where VE value is numerically less than the VE value in the cell below it.
Now this isn't the same as you rev out, you can have the VE go down as you rev out due to engine setup.

Where this become an issue is when you use speed density for setups idle with very little negative pressure and setups that don't have injectors modeled correctly or can't because there isn't a place to. In these cases the VE value in the cells around it is generally used set a very close PW and VE values will end up being odd looking.

I should probably add pictures to help ilustrate my points.

[Hentai]

To speed up offline tuning you can use also use an approach of gathering two tuned datapoints for the same rpm area at different loads ( eg, 3000 revs, 45kpa, 3000revs 80kpa ) make sure they are tuned to hit your target lambda, ( eg lambda 1 ). with those tuned and hitting the target you would run the formula high ve - low ve = delta ve, high kpa - low kpa = delta kpa. Delta ve divided by delta kpa = slope. From this slope you can figure out at the different load points where to start at.

This will help dial in a rpm range offline without having to collect tons of data points to begin with.
This also requires to have proper injector data at lower airflows where it inj pw maybe non linear.