RomRaider

Nice Merch! Thanks for the in-depth CBS Evening News-style reporting!

No problem.

I checked out the A4TF520F rom (usdm 04 wrx). The delay table is also split along transmission lines (even though the rom is tranny specific). But instead of atmospheric pressure, coolant temp is used to determine which of the 4 grouping of values to use. However, because of the comparison value, it will always use one table only. For ATs, this is the first delay table. For MTs, this is the second delay table.

i have an 05 so i dont have the ATM table that i can see

Right, yours is the same as the 04 wrx explained above

ok so let me clearify what i understand and pls correct me if im wrong

there are 16 columns in the delay table
First 8 (1-8) are for AT
second 8 (9-16) are for MT

column 9-12 are < CLCT (closed loop coolant temp)
and column 13-16 is for > CLCT

currently CLCT is set at 10.4 deg F.
that means 99% of the time ill be using column 13-16
I dont have that cold a temp

No, there are 4 tables total. The first 4 values are for AT, the 2nd grouping of 4 is for the MT. The 'Closed Loop Coolant Temp' table displayed in RomRaider has nothing to do with it. It is a separate comparison with coolant temp but this is irrelevant as it is effectively disabled (so the last two groups of four values are never used).

Some more information:

So why do the usdm 04-05 WRXs have issues with the closed loop to open loop transition compared the 02-03 wrxs? That is, why do these cars tend to stay in closed loop longer at the stock values?

See the screenshot below (02 wrx on left, 04 wrx on right). The problem is not the closed loop delay value. In fact, when warmed up (depending on atmospheric pressure for the 02), the 04 wrx has a lower delay or the same as the 02 for the MT.

The problem lies with the fact that the CL threshold for throttle and calculated load on the 04/05s is a single value, whereas on the 02/03s, it is referenced by rpm. On the 02/03s, the threshold is a 2d table by rpm and the values are zero'd out at 4000 rpm to redline. Also note that the CL rpm for the 04/05s (blobeye) is 5200 rpm whereas the 02/03s (bugeye), it is 4300 rpm. So, on the bugeye, when you hit 3600 rpm, the thresholds for load and tps begin to drop until it is zero at 4000 rpm. However, with the blobeye, the threshold will still be the same from 0-5200 rpm (69% tps, 5632 load).

So, to summarize, on the bugeye at 4000+ rpm the threshold to transition to open loop is any load and any amount of throttle (depending on the fuel map) with no delay. In fact, at 3600+ (because the value are interpolated), the thresholds will begin to drop. On the blobeye, however, the calculated load and TPS thresholds remain the same from 0 - 5200 rpm regardless (5632, 69.4%) and these include a delay as well.

So, what is the fix? Zeroing out the delay (especially the last value in each of the 4 groupings), would be the best solution as this would eliminate the CL load and tps thresholds from the equation entirely. Then the transition will be determined by your fuel map. Other alternatives, although maybe not as optimal for higher power setups, would be to lower the cl rpm to somewhere around 3600 rpm and drop the cl cal. load and tps thresholds.

What would you recommend the CL calc load and TPS threshold and other setting to be if the 4th column is not zeroed out

Attach is my 2005 CL/OL map

I would use zero delay personally. Otherwise, what you have looks good. Update your defs - you are using the old version. For load, with the new defs, divide the calculated load value by raw injector flow scaling (6442 if you haven't modified it and are using stock injectors). Then multiply by 2. This will be the engine load (2*g/rev) threshold. If you have an AT then you'll definitely want to lower the load threshold as this is "Calculated Load A". MT is B.

thanks
i updated the def file

what are the numbers on the Closed Loop Delays table represent ?
milliseconds ?

By placing a zero on the last table#2 (column #8) what would i experience ? Quicker transition from CL to OL ?

From my logs, each unit of delay is about 10 milliseconds.

It is important think of a delay value of zero as a completely different type of transition than a delay value of 1 or greater. The advantage of zeroing out the delay would be that the transition would be determined by your fuel map not any load or tps thresholds. The transition should be immediate. The 4th value in each table is the delay after about 8 minutes of the engine running (and beyond).

to add to bill's post and clarify a bit, you still effectively have a load-based open loop crossover since the fuel map itself is load based.

thinking about this last night, something occured to me... we might be able to use these values to STAY in closed loop for a longer time.

ie, put a large value in the delay period. then put very high values in for tps and "load" 2d crossover maps. run a flat fuel map of lambda 1 = no enrichment. finally, use the 3d "load/rpm target lambda bias manual" correction map to set where you want your target afr. i don't know whether or not this would actually result in closed loop enriched fueling or not, but i'm willing to try it out.

i FINALLY have a tuning laptop again after months of withdrawal! gonna try a basic small changes reflash this morning and if all goes well i'll be digging back into some more esoteric reflash testing.

ken

Maybe I am missunderstanding what you are saying but I see no point in this. If we zero the delays, it is quite easy to tell the ECU where you want C/L and O/L crossover to start (fuel mapping, like you stated). I had zeroed my delay values on my 06 WRX and it worked great. Crisp response, and the car retained good gas mileage as all of the load/rpm cells were at a high enough AFR to remain in closed loop at idle and low loads. I logged the transition a few times on that map and it was quick, lightning turbo spool, no jerkiness like the stock map had or anything.

FWIW, the 07 WRX STI is plagued with problems (hesitation, lag, jerkiness) as it stays in closed loop way too long. You can hit 16 psi in closed loop easy. Thats not good. I actually am at 1200 miles now so I have been getting on it on a regular basis and needless to say a stock 06 WRX is easier to drive fast (thank god the STI has sticky tires). I wouldn't take this car to a road race track even (on the stock mapping), let alone drive fast on the hill country roads we have here as the car will get squirly when its at full boost and the C/L - O/L transition occurs (huge TQ dip, then spike).

I have been and am getting bad gas mileage as well on the stock 07 STI mapping. The ECU just doesn't work well at all in C/L under boost (pulls timing, TQ spikes occur, etc), especially at part throttle.

It seems to me in my testing that the only reason one would want to stay in C/L longer would be for emissions. Still you can accompilsh this with just the fuel map.

-Gabe

Right, load or rpm based. The difference would be no delay. And the fact that the other closed loop tables would not have to be tuned as they would have no impact (unless you left a delay while the engine was warming up < 8 minutes).

you're assuming that my target for ALL closed loop operation would be stoich, or close to it (i actually target 15.5:1 for closed loop).

in actuality, what i am curious about is whether or not the ecu will target richer than stoich in closed loop. the only question at this point is whether or not the table i mentioned has enough authority to offset 2.5 afr points so that i can run say 13:1. i also don't know whether or not the ecu will remain in closed loop if the table is "in effect" (although i am fairly sure it will, since the other factors mentioned by bill that trigger open loop fueling would not be satisfied). finally, i do not know whether or not the offset error introduced by greater levels of exhaust gas backpressure will skew the front o2 sensor's readings too much to be of use for closed loop operation at higher engine loads.

however, IF it works, there are some substantial benefits to this, especially if one runs a utec on top. if the afrs at the point of handoff are actually closed loop controlled they will be extremely repeatable, and all one has to do is align the utec's fuel map to those afrs.

the utec in SD mode on top of an oem reflash is absolutely killer. an added bonus is that when in SD mode the utec does NOT offset the MAFV in any way, so you see the true mass air flow and engine loads. i am shocked at how stable the SD afrs are despite having NO IAT corrections programmed in the utec at this point. i've been laptopless for 3 months now and the temperatures have changed drastically, yet i am only running about 0.7afr richer. but i digress....