Depending on where your mapped AFR values are, those values I posted may still lean you out too much.

 

I reactivated lean spool with revised values with something similar to the tables above. I also reduced the time step tables a few clicks each (dropped the maximum lean side time step from 12 to 7). My intention was to make the AFRs just lean out a bit under spool up then bleed off very quickly to the tuned AFRs. I want to do this so it will basically lean out 1st and 2nd gears a decent amount and then basically leave the higher gears alone.

With the factory values, I use to see the AFR go richer then what it was tuned for very briefly, then go quite a bit leaner and then settle down to a final AFR that was about 0.5-1.0 point leaner then what the main table showed.

With the new settings, I got the same rich-lean swing as boost came on, but it very quickly went to the main fuel map value. The swing definitely happened a lot quicker and with less magnitude the with the factory values. It also does not run ~1.0 leaner any more like it would with the factory values.

I want to get rid of the rich condition and just have it run leaner in the lower gears and during spool up. Is there by chance a second AFR table for the rich side settings like there is for the lean side? Would setting the time step to 0 on the rich side time step get rid of the enrichment without messing up something else in the code?

 

I though the main problem with lean spool was that each time you shifted to the next gear it would richen up and deviate from the fuel map, and that in a single gear pull it would match the tables correctly.

 

From what I have seen with the factory values, the AFR goes quite rich from a very brief period of time on spool up. Then goes quite lean. Then tapers off to more normal AFRs.

I think in third gear with stock mapping, my car would drop to about 9.0:1 for about 0.2 seconds, swing up to 11.5:1 in about 0.1 seconds, then slowly enrich from 11.5:1 to 10.5:1 over a span of about 2 seconds. All the while, the factory main fuel mapping was aiming for like 9.8:1 AFRs the entire time.

 

I gave a bit more thought to the lean spool decay/reset tables, and I think I've come up with the correct scaling for the decay/reset rates as shown below for the 88590015 ROM. If these timer scalings are correct, then "Lean Spool" doesn't really look like a lean spool strategy. The maximum decay/reset rate is 3.92%/sec which is pretty slow. It would take 25 sec to make the AFRs decay from the "Lean Spool" values to the AFRs in the fuel table. Definitely too slow to have a lean AFR during spool and then make it go to normal AFRs at full boost. It seems more like a strategy to create a lean condition for the first several runs through the gears and then go to normal AFRs after that. And the factory reset rates will bring the AFRs back to the lean spool values rapidly during high rpm decel, e.g., perhaps in race track conditions. Suggests that Mitsubishi thinks that high rpm decel will help cool the cylinders.

<scaling name="LeanSpoolChangeRate" units="%/sec" toexpr="10/x*100*(1/255)" frexpr="10/x*100*(1/255)" format="%.2f" min="0" max="6" inc="0.05" storagetype="uint8" endian="big"/>

<table name="Lean Spool Decay Rate" address="338e" category="Fuel Lean Spool" type="2D" level="2" scaling="LeanSpoolChangeRate">
<table name="RPM" type="X Axis" address="6e62" elements="14" scaling="RPM"/>
</table>

<table name="Lean Spool Reset Rate" address="33a2" category="Fuel Lean Spool" type="2D" level="2" scaling="LeanSpoolChangeRate">
<table name="RPM" type="X Axis" address="6e62" elements="14" scaling="RPM"/>
</table>

 

I just remembered that there is actually a way to increase the decay/reset rates. The interpolation step size can be adjusted (although Mitsubishi did not intend it to be adjustable). The stock value is 1, but it can be set to anything from 0 to 255 with 255 causing the AFR to go to to normal AFRs in a single step. The entries for the 88590015 ROM are below. The Lean Spool Decay rate timer scaling in the post above would need to be changed as well. The "1" in the Lean Spool Decay rate timer scaling formula would need to be changed to whatever value used in the Decay/Reset Step Size.

<table name="Lean Spool Decay Step Size" category="Fuel Lean Spool" address="17b19" type="1D" level="2" scaling="uint8"/>

<table name="Lean Spool Reset Step Size" category="Fuel Lean Spool" address="17b3b" type="1D" level="2" scaling="uint8"/>

 

I hope your times units are correct, but I couldn't decipher the timers from disassembly because I can't see where (and more importantly how often) the timers are decremented to zero which then results in one click in the interpolation. I can see where they are compared with zero, reset and how the interpolation works. I planned to log the interpolation variable soon. I did notice that the JDM 8MR goes richer much quicker than my JDM 9.

 

mrfred, those new timer units dont "feel" right to me.
I tend to agree with Mats view in that the factory engineers would want and need something easy to work with.
And we are getting different values for different model evos and bins, so they are changing the recipe "as required".

 

Attached is a log of a short driving including some 3rd gear WOT from 3000-7000 RPM so you can see what the lean spool interpolator does. It starts at 0 and increases to about 60 on a WOT run through 3rd gear, but then on lift off jumps to 255 and decrements. There was too much traffic about to see what happens if you drag through a few gears one after the other, but I'll find out some time this week.

 

I pretty much took a WAG on the timer units (x/10) based on other timer RAM variables I've come across. I'm seeing almost no change in AFRMAP during a single pull for the car I'm currently working on. This is consistent with %/sec decay rates with my scaling formula. I'll have a definitive answer tomorrow though. I've sent the guy a new ROM with the decay timer value and the interpolation value piped into the MUT table for logging.

 

I did find a bit of the log with part throttle (but enough load to have lean spool active) with a few short shifts as attached. This time on gearchanges it seems to start decrementing briefly and then start incrementing again in the next gear.

The x axis units are the sample number, on this log about 7 per second. Based on that, about 15 seconds of 40% midrange short shifting would be enough to get the interpolation all the way to the rich end. Will see more in mixed driving this week as I'll be running this log continually.

 

Looks like I will be unable to get the logs I need to determine the timer scaling for now. The guy has been unsuccessful at installing the modified Data.xml file that I sent to him. Looking forward to what you determine for the timer scaling, or I'll figure it out with my own car when I get back from vacation.

 

After running with LS enabled and disabled using ABA Bit 0. I determined that my car runs substantially better with LS off. I lost power, idle was horrible, the car would stall on occasion, AFR swings were dramatic. I did notice a tad better boost hold durring shifts though. I might attribute this to the weather or me just shifting faster.

I will not enable LS again.

 

I did some further logging today, and it does seem to behave like the graph in post #101 even on full throttle. I don't know why before it would go to 255 on lift off. Will keep watching it. I like the system on my engine.

 

With the new found understanding of how these maps work, I think they would be a very good thing to use.

For drag racing, you could increase the step size so that it will count up quicker to lean out the lower gears. Quicker ETs with safe tuning under heavy load.

For auto-X, it would be similar to a drag setup, but less aggressive so that the car has good power through the tight slow speed stuff, but on more open courses keeps things safe.

For road racing, the harder the car gets beat on (and the more heat that is getting built up) the more conservative the car runs. Might not be best for competition, but for the average track guy, could save a motor and the tuner from explaining why the high power tune the customer asked for melted the motor.

If somebody has time to find the counter step increment (factory set to 1) in 94170015, I'd be interested. Right now, I have the car set to where it only leans out a bit as the counter is too slow to get it to lean out in the low gears from a drag start but keep the higher gears happy.