tephra

iTrader: (6)

Hey All,

I have been testing my Flex Fuel code, which is in V3 of my TephraXMOD!

So of you have probably seen this already, but now that I am getting more confident with it, and it coming out of Alpha, i think its a good time to introduce it.

So, here is the story, E71 in the tank almost out (second ding a few k's back), pulled into BP gas station, put 20L of BP Ultimate in (E0, 98RON):



Top Graph is Fuel Trims + Boost and AFR, you can see that I dont start the car until about 28 seconds in.

Bottom Graph is the raw sensor (blue), the averaged input (red) and the injector scaling (teal)..

Now one of the cool things with V3 (well i think so), is that you can cycle the fuel pump, without starting the car. Just hold down the Cruise-Control Cancel button, and it will spin the pump for as long as you like (and your battery holds up).. You can see that I did this from about 0:05 to 0:25.

You will notice some lean-ness after I started (0:30 - 0:46), I am not sure whats causing this, as you can see injector scaling only gets higher (thus leaner) after that point. I think it's because my cranking IPW is still E85'd so causing a misfire... anyways not to sure, on E85 startups are flawless.

So how does this all work?
1) The ECU uses a moving average algorithm to average out the sensor input, this provides two things:
a) Some safety factor, if something unexpected occurs there will be a little bit of a delay before the ECU starts acting on the new input
b) Allows for the ECU to alert the user to *something* unexpected (by way of CEL)

2) The average is the fed into a couple of interpolation maps:


You can see that each major function of the ECU is represented, this allows different interpolation curves to be applied to say timing vs fuel vs boost.. This should allow for a great deal of customization and trickery.

Right so once the interpolation percentage is calculated its then fed into the respective map lookups. Just like normal, but instead of a fixed output it varies between the "normal" maps and the "altmap #1"...

And as always, there are some extra safeties:
a) the ECU wont calculate the moving average if the load is above a certain threshold.
b) there is a sensor error voltage, but will be sensor dependent, meaning the flexconvertor sensor with its programmable curve will work well, but other products without a sensor failure voltage.. will be a bit hit and miss...

/phew, i didn't intend for this to be so long, but there you have it - hopefully it can turn into a "yeh i got flex how do i tune it.. type of thread

tephra

iTrader: (6)

reserved for more info

bellacotim

David,
On my ID1000, I looked and there are 16 steps for injector scaling sweeping between E10 to E85. I would recommend to increase the Ethanol content to at least 16 entries to maximize resolution and take advantage of these quantas.
Tim.

tephra

iTrader: (6)

Hey Tim,

The interpolation algorithm will interpolate down to the single step integer level. So by not having 16 entries doesn't mean your missing out on resolution.

Making the table bigger (originally i had it a 5 big, but settled on 9) just lets you add more curve detail, which probably isn't needed anyways..

In fact for the injector scaling, my original code just had the table output a injector scaling value, but then i realised this would be confusing to tuners because the different tables would do different things.

Clipse3GT

iTrader: (10)

Very nice work David!!!

It would be nice to use this to verify E30 or E40 for the people that don't want to run E85.

Do you know if this sensor would work and gauge combo would work?

http://www.veisystems.com/ws/v1-seri...tio-gauge.html

ANALOG OUTPUT
The analog output provides a proportional absolute output from 0.5V to 4.5V for ethanol-to-gasoline ratios of 0 through 100 respectively. If the
signal input from the sensor is too low (out of normal range), not present or not readable, the analog output goes to 0V, and the display shows
“LO”. If the signal input from the sensor is too high (out of normal range), the analog output goes to full (~4.75V), and the display shows “HI”.

bellacotim

Thanks for interpolating between ethanol %s.

I have some questions about processing load and such. At what rate are you running through your patch? Are the interp inj scale/fuel/timing/boost values placed into soft tables for ECU consumption or are you intercepting and doing the interp at time of lookup? If the former, how often are these interp tables rebuilt? Are all the tables rebuilt at once or is update done incremental? Is table rebuild only allowed during certain engine operating conditions?

Finally, can you interpolate scaling for MPG display?

Thanks again,
Tim.

tephra

iTrader: (6)

I didn't visit the URL, but judging by that description that sensor should work ok. In my patch, the axis to the 4 interpolation maps can be changed to voltage (I mean it is anyway), and so you can have it go from 0.5 to 4.5 if necessary/required.



The ethanol interpolation slope and injector scaling are done via the main loop, in my xmod hook. So 100Hz.

The table lookups are done in their respective lookups, so things won't be being calculated out of order.

Additionally I figure mpg scaling is more related to injector size and thus being like injector latency, should apply the same between pump and e85.

mantella

iTrader: (6)

if you need any testing let me know. this time the engine is in the car unlike the 1st couple revisions of aem wideband for live tuning where I had to use propane to get my readings

the sam

Tephra, I've sent you an email. Have multiple cars with different roms to test.

BoMbA_CaJu

Tephra, do you have this mods for Ralliarts????

lathiat

Nice work, fuel pump cycle was a great idea.

kaganader2

I have this gauge set up in a few cars.

Haven't been on EvoM in a bit, just saw the thread.

Another safety precaution that should be applied as it is done in Chevy flex fuel algorithm, is that the last recorded value for the ethanol content be save and used for next startup and in the rare occasion that the sensor fails or connection is lost. This will also help with that leanness you see after startup also. They also freeze the percentage/hz value over a certain load & tps value but you have done this part already.

The cranking misfire you noticed with lower percentage of ethanol but no change to the pulsewidth for cranking, with the value of ethanol that has been saved from last drive etc, this could also be used in multiplier formula injected into the pulsewidth of the crank table to make appropriate adjustments based on the ethanol %.

From the last time I looked up this stuff, I remember 50hz = 0% ethanol 100hz = 50% ethanol and 150hz = 100% ethanol from the GM sensor.

kaj

iTrader: (60)

any plans to do this for the VIII/IXs? i would

lathiat

Already available here: https://www.evolutionm.net/forums/ec...n-evo-ecu.html

kaj

iTrader: (60)

No way! I'll check that out. Thank you