I got a PM from Chas (NOFX0617) asking for some information about tuning. I ended up writing a long reply and thought I would post it here incase anyone wanted to take a look at it.
Alright, let me see what I can type up here for you before my fingers start to hurt.
The ECU basically has 3 operating modes, closed loop, open loop and limp mode. Limp mode is reserved for potentially damaging proplems the ECU may encounter. The main problem of which, is a nonfunctional MAS. Without feedback from the MAS the ECU has no idea the mass of air passing to the motor. At this point it enters limp mode and runs extremely rich to protect against what could be disaster. This allows you to get your car to the shop or home so it can be repaired immediately.
Closed loop is where you spend most of your time. It is a loop that involves the O2 sensor and the ECU. As the ECU receives a voltage from the O2 sensor it determines if the previous combustion cycle was rich or lean. It then uses the Short Term Fuel Trim (STFT) and Long Term Fuel Trim (LTFT) to adjust the fuel delivery for the next combustion cycle. The ECU is shooting for an AFR of 14.7:1 which is the stoichometric point where all of the fuel is exhausted during the combustion cycle. This allows for maximum use of the fuel injected and gives you the maximum mpg possible. An AFR of 14.7:1 corresponds to an O2 voltage of 0.50v.
If it determines that the previous combustion cycle was lean, i.e O2 voltage <0.50v and AFR >14.7:1, it increases the STFT which corresponds to an increase in IPW (Injector Pulse Width) in an attempt to bring the O2 voltage back to 0.50v. If the previous combustion cycle was rich, i.e O2 voltage >0.50v and AFR <14.7:1, it decreases the STFT which corresponds to a decrease in the the IPW.
STFT is the ECU's display of what is happening right now, whereas, LTFT is pretty much a running average. If the ECU sees a consistent rich or lean condition, it will transfer the STFT value to the LTFT and return the STFT to zero. When running a SAFC, you need to adjust the airflow so the STFT and LTFT add up to zero. For example, if the STFT is +5% and the LTFT is -5%, sooner or later they will cancel each other out and both the LTFT & STFT will be 0.
The other "mode" the ECU operates in is called open loop, this is normally considered WOT. In this "mode" the ECU uses things like barometric pressure, Intake Air Temperature (IAT), coolant temperature, etc. to calculate the mass of air entering the combustion chamber. Based on this air mass calculation, the ECU attempts to inject an appropriate amount of fuel to reach the desired AFR. On a 2g, the factory programmed maps are designed to hit an AFR of 9.5:1. This is rich compared to the closed loop stoichometric value of 14.7:1, but the increase in fuel delivery is used to prevent knock by using the extra fuel to cool the intake charge. When running at WOT, the ECU uses feedback from the knock sensor to monitor engine operation and protect the motor. Keep in mind, that the O2 sensor feedback is completly disregarded during open loop operation.
See, the ECU has what can basically described as 3-dimensional maps for fuel delivery and timing values based on the mass of air being delivered, rpm, etc. For example, at an air mass of 2.1 g/rev the timing is capped by the ECU at 16 degrees of advance. So, if everything is perfect, and within the predefined specs, you will see 16 degrees of timing when running over 2.1g/rev. This isn't always the case. Based on feedback from the other sensors listed above, the ECU will modify the timing value from their preset target. For example, if the IAT is >84 degrees, you will loose 1 degree of timing. So now, at 2.1g/rev, you will only see a maximum advance of 15 degrees.
Timing values are also heavily influenced by feedback from the knock sensor. The ECU uses a linear relationship to pull timing based on the knock count. So the more knock the ECU "hears", the lower your timing value will be. This is how we can use a log to determine knock on a 2g. The 1g guys have it easy, because knock values are output to the OBD port. On a 2g, the ECU does not transmit this value to the OBD-II port. This forces us to "estimate" knock based on the timing curve. For example, you make a WOT pull and you see a steady timing increase from 3k up till you hit 5k. At this point the timing begins to drop, from let's say 15 degrees, down to 13 degrees. Most likely this is caused by knock, but it could also be caused by your coolant temp going over 210 degrees and the IAT going above 84 degrees simultaneously, each of which will cost you 1 degree of advance. What you do in that case, is make another pull and log coolant temp and IAT. If you don't see these values exceeding the limit imposed by the ECU, then you know the problem is caused by knock. At this point, we need to determine what is causing the knock.
Normally with the SAFC, this involves adding fuel in the affected rpm range. But you have to be careful, because there is also rich knock. So, you make the appropriate adjustments on the SAFC and make another pull. If the knock goes away, great! If not, we need to keep making adjustments.
Using the SAFC is like ice skating uphill. As you use the SAFC to molest the mass of air the ECU "sees", the ECU increases timing. Remember, the lower the mass of air seen by the ECU, the higher the ECU will allow the timing to increase. This causes a problem, because we really aren't running a lower value of air, we are just telling the ECU that we are. So as you increase the size of your injectors, you need to remove more and more fuel. See, when you add larger injectors, the ECU doesn't know, and frankly, doesn't care. It treats them as stock and continues to use the programmed IPW. But since the injectors are larger than stock, the same IPW results in additional fuel that the ECU interprets as a rich condition. Seeing a rich condition, it decreases the STFT and LTFT until it runs out of adjustment range. I believe this occurs at -17%. When it hits it's adjustment limit, you get the good old P0170 CEL, Fuel Trim Malfunction (Bank 1).
This is where the SAFC comes in.....By lowering the airflow value before it reaches the ECU, we can manipulate the IPW. Now that the ECU sees less air, it drops the IPW and basically compensates for the larger injectors. This is also what causes problems at WOT. You want larger injectors to support additional flow generated by a turbo upgrade, right? So now we have a turbo that flows significantly more that stock and larger injectors to support the additional flow. So what do we do??? Add a SAFC and manipulate the airflow to the ECU. This works, but causes a big problem. The ECU is now seeing significantly less air than is actually being "ingested" and allows for additional timing over the 2.1g/rev cap of 16 degrees. High timing + high airflow = knock.
This is the problem with running the SAFC, over something like DSMLink. It is "doable" but is definately not the best way to approach the situation. I am running 650cc injectors and a SAFC and have a pretty good tune, so it can be done.
My fingers hurt now, so I'm going to stop typing ....If you want some additional information, let me know, but that should give you something to chew on for now.