[arghx7]

Originally Posted by jedibow

Another interesting fact was the stock enrichment at high RPM's, if it was not there more power could be made up top, the AFR where in the low 11's, could this be to protect the cats? or do you think it is only a safeguard built into the factory tune for reliability?

I'm sure the biggest reason for that level of enrichment is to protect the cat(s). Remember how the max estimated cat temperature was about 850C near redline? If you lean the mixture out, power goes up to some extent and so does cat temp. The estimated cat temp calculated in the ECU may not reflect that (it depends how the model was designed), but it will definitely get hotter. There are a few different places you can measure cat temp. You would want to put your K-type thermocouple in there and use a high speed data acquisition system.

Another aspect that I failed to bring up to this point, during the torque dip the actual AFR was 13.3 (kind of explains the advance timing in this area however), which correlates to the command AFR,

To further clarify the issue it would be nice to see a chart of commanded AFR, measured AFR coming from the stock Denso limiting-current type wideband o2, measured AFR on a Bosch pump-cell type wideband o2 in the exhaust stream, and measured AFR from a Bosh sensor at the tailpipe.

if what John (visconti) said was true, what is the advantage of running a lean AFR during open loop, especially whe 12.5 produces the best power, if anything it is creating a hot spot in the rev range.

The data to me is still a little murky as far as closed loop at lambda 1, closed loop at lambda 1 with a delay timer incrementing (like on the current WRX/STi), closed loop in a partially enriched state, open loop at an enriched state, always being in closed loop at an enriched state... there are so many ways this could be handled and it may be shifting into different modes under different conditions. I'd like to see more data to interpret for myself, either directly from ECU RAM addresses or from the Universal CAN diagnostic protocol. Specifically, I'd like to see the Fuel System Status PID over the bus or specific flags/bits from the ECU RAM addresses.

The issue here is CO emissions on the US06 test cycle. A WOT pull on a chassis dyno is not the same as a driving a US06 test cycle, but the strategies used to meet CO emissions tend to affect WOT control as well. The US06 test cycle doesn't have WOT, but it does have heavy acceleration that can be difficult for low torque small engines that run at high loads. The current standard for this vehicle should be around 8.0 grams per mile, but in a few years when LEV III emissions comes out that may be cut in half. The vehicle still has to meet certain in-use testing requirements 1 year and 4 years from now. So you can't tune the thing on the ragged edge of the emissions standard, or there will be fines and recalls in a few years if enough of the cars fail.

That's why you go as lean as you can during transients without exceeding targets for cat temperature and knock. As I mentioned above, there are a lot of ways you can do this depending on how accurate your sensors are and how much manpower you want to put into developing software and engine calibration. I would be surprised if the FA20 just goes from closed loop lambda=1 to open loop enriched based on a couple look up tables while still meeting emissions. These days some kind of intermediate step usually occurs--delay timer, closed loop enriched, etc. I don't want to make some kind of judgment based on insufficient data.

Maybe I will take the EGT out of the evo to get some readings. I cannot log it as it is a cheap autometer gauge, but I could atleast observe the difference between the torque dip, and once the AFR get down to 12.5 like it should be.

FJO makes a two-channel converter box that allows you to convert a K type thermocouple to a 0-5v analog signal. I've had pretty good luck with this system on my own projects; but you would be spending time and money to merely satisfy curiosity. The Autometer EGT gauges read too slow. They are useless for this, don't even waste your time.