RomRaider

This has been tested on 32-bit ECUs, but I do not know if it will work on 16-bit ECUs.

The factory ECU has two methods of determining how much fuel to inject, known as closed-loop and open-loop fueling. Both methods start with the ECU calculating in the amount of incoming air, either with the MAF sensor, or via engine speed and the MAP sensor (speed-density). Given the amount of air coming into the cylinders, the ECU can calculate the required amount of fuel to achieve the desired air-fuel ratio. In open-loop fueling, the ECU simply injects whatever amount of fuel these calculations indicate. In closed-loop fueling, the ECU takes into account two additional factors.

First, it considers difference between the current AFR and the desired AFR, and it adjusts the fuel injection by an amount that proportional to the difference. You can log this parameter, it is called "AF Correction #1." Second, the ECU considers how much correction has been needed in the past to achieve the desired AFR at the current MAF rate. There are four or five stored values which correspond to the different MAF rates, and you can view these with a logger, they are called AF Learning #1 A, B, C and D, and some ROMs also have AF Learning #1 E. The learned value currently being applied can be logged as well, it is called "AF Learning #1" (with no letter after the #1).

The AF Correction value is also sometimes called "short-term fuel trim" because it is adjusted rapidly (many times per second), and the AF Learning values are sometimes called "long term fuel trims" because they are adjusted slowly.

Some ROMs also have AF Correction #2, and AF Learning #2, since they have an O2 sensor for each bank of cylinders. To the best of my knowledge, only the H6 motors have this second O2 sensor and second set of parameters.

Closed-loop fueling allows the ECU to achieve the desired AFR in spite of differences in fuel and other variables. The target AFR is chosen for a compromise between best fuel mileage and lowest emissions. Also note that in cruise, the target AFR will fluctuate slightly rich and lean of 14.7, to facilitate two different chemical processes in the catalytic converter.

The drawback to closed-loop fueling is that the fuel trims (both short and long term) can hide sub-optimal values in your MAF scaling and tip-in settings. It's generally beneficial to have the ECU applying these trims while you drive, however if you are tuning tables related to fueling - especially transient fueling issues like tip-in and load-change enrichment - the trims just make your job more difficult by obscuring the AFR that your settings would achieve on their own.

So, I typically disable closed-loop fueling while tinkering with such tables, and re-enable it when I'm done. Some people prefer to run open-loop 24/7 @ 365, but I am not one of them - being a bit of a tree-hugger, I like to give my catalytic converter a bit of help.

Contrary to popular belief, closed-loop mode does not necessarily need to target the stoichiometric ratio (14.7:1 with gasoline), and open-loop mode does not necessarily need to target the very rich values that are typically used in high boost (10:1 - 11:1). It is definition possible to have the ECU use open-loop mode with AFRs very near 14.7 in open-loop mode, as I will explain below. It might be possible to have the ECU target rich values in closed-loop mode, however this seems like a bad idea until the rich/lean fluctuation I described earlier can be tuned out. At the time of this writing, the tables that govern that fluctuation are not included in the ECU definitions.

To disable closed-loop operation:

Set all of the closed-loop / open-loop delay values to zero. This will cause the ECU to switch from closed-loop to open loop as soon as the open-loop conditions are reached. The open-loop condition that we're most interested in is the one that tells the ECU to switch to open loop when the AFR specified in the Primary Open Loop Fueling table is richer than the "Minimum Active Primary Open Loop Enrichment" value, which is contained in a single-value table in the "Fueling - Primary Open Loop" section, which brings us to the next step...

Set the "Minimum Active Primary Open Loop Enrichment" to something quite close to stoich. I typically use 14.65. Now, whenever the AFR specified in the "Primary Open Loop Fueling" table is richer than 14.65, the ECU will use open-loop fueling rather than closed-loop fueling.

So, find all of the cells in the "Primary Open Loop Fueling" table which are leaner than (larger numbers than) 14.65. Typically this just means the set of cells in/near the top-left corner of the table, whose values are 14.7.

Save these changes to a file, flash that file to your ECU, and enjoy. In my experience, the difference between 14.6 and 14.7 is so small that there is no noticeable change in gas mileage. Also note that the target AFR used in closed-loop mode is typically slightly richer than 14.7 anyway (you can log that with the "Closed Loop Fueling Target" parameter) so this trick might even increase your gas mileage by an infinitesimal amount.

To re-enable closed-loop fueling, I simply change all of the 14.6 cells in the "Primary Open Loop Fueling" table back to 14.7. If you prefer to use non-zero CL-to-OL delays, you'll want to revert the changes in that table as well.

(Pictures of these tables will be added later.)

(This space reserved for future updates.)

Just FYI, NSFW - there's another trick on the GR's (not sure if it's a quirk, or also on other models). If you set the Closed Loop compensation tables to zero, you run in open loop permanently (that's why if you intend on not using these, you need to put in 0.0001 or something similar).

That might work on my LGT as well, but I haven't tried it. Merchgod put a comment on that table that says "Lean compensation in this table will potentially force open loop during normally closed loop fueling conditions." The "potentially" part kept me from trying it, though, because it didn't sound as dependable as I'd like. I'll have to experiment with it some time, though.

Interesting - I just tested zeroing this table (since I implemented the engine load B table hack to the multiplier, I would like to clean up my scaling) - and unlike the last time I did it, it did not go into open loop - however - it switched AF learning #1 off completely, only leaving AF correction #1 functioning.

What's also interesting is that the range of AF correction #1 is much narrower - normally I get all the way from 25% to -25%. Now it's about 8-10% either which way.

EDIT: had a closed look at the logs - only one had closed loop, and then only for a very short while after starting up the car. Thereafter, it reverted completely to open loop.

I can confirm that changing the comps to zero (not necessarily a lean compensation) forces it out of closed loop, whether it's truly open loop or not I couldn't confirm, read on. I tried this only in the cruise region of the closed loop region of my fueling map http://www.romraider.com/forum/viewtopic.php?f=15&t=7725. Interesting thing is that the AF target freezes on whatever value it was at during oscillation when it enters the zero comp cell, meaning since it is oscillating for cat efficiency it could be as great as 15 or maybe 14.5. There's no control of what AFR will be when you traverse into the cell with zero comp. Check the link I provided and take a look at the plot. Zeroing out the entire table might have a different result because presumably you'd never enter closed loop, therefore never oscillate, I guess it would just target stoich?

I wanted to use this route due to the 'controversy' associated with setting the closed loop compensation to zero (also, there is a whole boatload of tables in it).

I don't understand why I can't change "Minimum Primary Load Enrichment" to a value larger than my OL fueling max. I tried setting it to 14.75 so I didn't have to change my OL AFR, but it didn't work apparently.

I'd be fine running in OL 24/7 actually.

Yeah, the threshold has to be something smaller than 14.7. It would probably be possible to change that by hacking the ECU code, but that kind of thing takes a lot of work.

You can set the threshold to 14.69 and set your fuel table to 14.68 if you want to get closer to the 14.7 ratio.

Ah, I see. Will give it a go. I'm going to make some changes on my map anyhow, then I'll try it out and confirm.

Works like a charm. AFL/AFC show that they are locked at zero as a nice confirmation.

The smallest increment I can make on the open loop fueling tables is down to 14.59. So I set everything at 14.7 down to 14.59 and set the minimum enrichment to 14.65. I'm not overly concerned about running at 14.59 - last I checked the gas has ethanol added anyhow and stoich is closer to 14.2 =P

what is the purpose of this? wouldnt you get better mileage with a leaner mixture? I dont see a real benefit or a tuning purpose behind this. can you please explain?

It's buried in the middle of the first post:



And I just noticed a badly worded sentence in there, which I'm going to fix now...

so how do you properly scale your maf now in the low load low rpm range? you cant use maf scaler in romraider right? is that not accurate enough and more economical when it comes to closed loop operation and fueling?

I have a feeling the the romraider maf scaling tool is absolutely correct, but its REALLY hard to feed it PROPER data. Tip-in should be either spot on, or disabled completely. Running should be done somehere where you wont be having to shift out of 6-5 gears, apply throttle plate changes, and most importantly use 1-4 gear accelerations. If you do, scatter goes all over.

I guess MP load conpensation tables should either be zeroed, or not touched afterwards.
factory avcs maps also ruin your data. I say 0 overlap for proper o2 readings.

+1. I do mine manually anymore.

I just take measurements at steady state data and correct the MAF at that load range. It's difficult to determine what corrections go in the MAF scaling and what corrections go into the MRP vs RPM table though.