A'PEXi Super AFC Tech Notes

A'PEXi S-AFC FAQ:

As a result of several discussions at the i-Club, I have decided to compile a brief FAQ regarding tuning with the A'PEXi Super AFC (S-AFC). It is hoped that this will save people, myself included, the trouble of digging through the i-Club archives (like I had to do) for information on this topic.

Q: What is open-loop (or closed-loop) mode?
A: In closed-loop mode, the ECU uses readings from the O2 sensor to determine how to adjust the fuel injector output. It is a self-adjusting mode to help the engine run at a specific air/fuel ratio. It is suggested that the ECU uses this feedback, likely along with other sensors, to adjust other parameters such as ignition timing as well. In open-loop mode, the ECU ignores the O2 sensor readings, and instead, reverts to fixed fuel injector output settings stored in its memory. Since the manufacturer has to use the same fuel map for the stackup of all manufacturing tolerances and operating conditions, these open-loop settings are typically poorly optimized for performance driving.

Q: When does the ECU go into open-loop mode for fuel mapping?
A: As far as I can tell, this happens at about 20% and higher of throttle position, which corresponds to about 1.0V from the throttle position sensor (TPS). I have not seen any evidence to show that the ECU looks at any other parameter in determining when to operate at open-loop conditions.

Q: What should the Lo-Thrtl and Hi-Thrtl points be set to? Why are they there?
A: Based on the above information, set the Lo setting to 20%. The Hi setting can be set to anywhere from 65% to 80%. This will give your TPS a usable range for use between closed-loop mode and WOT.
They are used by the S-AFC to calculate fuel corrections at various TPS readings (throttle positions) at each RPM. At TPS settings below the Lo-Thrtl point, fuel corrections are constant - this is good for letting the ECU operate properly in closed-loop mode. At TPS settings above the Hi-Thrtl point, fuel corrections are also fixed - good for compensating for a throttle pedal that doesn't go as high as 100% (such as caused by thick carpeting or floormats). At TPS settings in-between the two, the S-AFC always calculates a correction proportional to the Lo-Thrtl and Hi-Thrtl numbers. Close to the Lo-Thrtl point, the correction will be more like the Lo setting at that RPM; while close to the Hi-Thrtl point, the correction will be more like the Hi setting at that RPM.

Q: How does an Air/Fuel Gauge connected to the O2 sensor tell you if you are running rich or lean?
A: The O2 sensor produces an analog voltage depending on how much oxygen it detects. 14.7 A/F ratio corresponds to about 0.5V; rich will result in higher voltage (less O2); lean will result in lower voltage (more O2). O2 gets consumed when more fuel is available to consume it.

Q: How does the S-AFC adjust fuel? Why does it need the TPS and RPM signals?
A: By intercepting the MAF or MAP sensor voltage, and modifying it, the S-AFC tricks the ECU into thinking that there is either more or less air going into the engine than there really is. In theory, the ECU will blindly add or subtract the appropriate amount of fuel to compensate. The S-AFC uses the TPS and RPM signals to allow the tuner to set specific operating regions where fuel is to be added or subtracted. Unfortunately, this is a rather oversimplified method of doing things, since ideally one would also be interested in knowing about engine load and/or boost (for forced induction applications).

Q: Are all O2 sensors the same? Can any generic Air/Fuel Gauge work with them?
A: No. While most are the generic type with a 0V-1.0V output range, there are certain so-called "wide-band" sensors that produce a much larger voltage output. Note: the "wide-band" term is a misnomer, as it should really be called "high dynamic range" instead ("wide-band" is a communications term that refers to the amount of information that can be transmitted over a particular signalling channel). Anyhow, the S-AFC will only work with the generic 0V-1.0V type.

Q: What gauges are necessary to perform tuning? Air/Fuel (A/F) Gauge? Exhaust Gas Temperature (EGT) Gauge?
A: This topic is open to debate. However, the consensus is that an EGT offers a better indication of the engine's condition, and is less susceptible to errors caused by environmental and temperature variations than the O2 sensor.
Also, certain ECU's may be tuned to match specific O2 sensors, and there are bound to be differences among various brands. The EGT probe provides a more consistent and universal way of measuring combustion conditions, as long as it's located at the same spot on various cars (about 3" from the exhaust header flange). Some tuners elect to mount the EGT probe at other locations, such as at the turbo, and this will affect tuning parameters.

Q: What's this about using the second input channel to read the O2 sensor voltage?
A: The S-AFC has two input channels in order to support cars that have two O2 sensors (such as the awesome Nissan Skyline GT-R). Since we only use the primary input for the O2 sensor, the S-AFC ignores the second channel in the monitor screens. However, we can still view the voltage readings at the Sensor Chk screen, although it is awkward, and there is no logging facility for it like for the other primary sensor inputs.

Q: What should the In and Out settings be? Any other settings?
A: For a Subaru EJ22 or EJ25 with a MAF (Hot Wire), Use Type 4. For a MAP (pressure) sensor, Rallispec uses the Type 5 setting (no information is available to me to suggest otherwise). See the next section below for an analysis of the various sensors and how the S-AFC interprets the sensor voltages. Number of cylinders, of course, is 4. The Throttle is the increasing type, with the arrow pointing up and to the right.

Q: What are the Ne-Point settings for? Should I change them?
A: They allow you to change the fuel curve's calibration points to fine-tune certain RPM ranges. The S-AFC default of 1000 rpm increments is not exactly optimal, since most cars have a redline much lower than 8000 rpm. Therefore, I recommend using the following values for the Ne-Point table:

These settings will allow finer control of the correction values where one typically needs them most - at midrange and near redline. We assume from above that the ECU operates in closed-loop mode at below 20% throttle. It's a happy coincidence that with 4 cylinder engines, we're used to downshifting to develop the necessary torque for performance driving anyways. You don't need to follow these settings; however I do recommend using something similar so that you can take advantage of all the tuning resolution possible from the S-AFC.

In the turbo column, the 1000-2500 range is used to tune for off-boost conditions, while the 3000+ range is set for full boost conditions. This assumes a boost threshold of about 3000 rpm. You should adjust this transition threshold to suit your particular turbo setup.

Q: What correction numbers should I use?
A: This is a difficult question to answer, since every car is different. But I can provide some simple instructions, assuming that you have an EGT to tune with, and an A/F ratio gauge to verify which side of the stoichiometric curve you are on.
1. Leave the Lo-Thrtl settings at 0%. This will allow the ECU to learn the injectors the way the factory engineers designed the fuel system. It's a different story if you go with large injectors (more than 25% larger than stock), but I'm still trying to figure that out.
2. Set the Hi-Thrtl settings to get an EGT reading of about 1450°F. Start at the low rpm Ne points, and work up the rpm range. The best way to do this (short of making your own chassis brake) is to short-shift and tune while going up a stretch of highway that allows you to maintain a particular RPM long enough to take a good EGT reading. A long uphill with a constant grade is perfect for this purpose.
3. Start at 0%, do a run, and wait for the EGT gauge to settle. If it reads low, take away a few %; if high, add a few. Repeat the run. Use trial-and-error to find a setting that will give you 1450±50°F, while staying on the rich side of stoich. For an Impreza, this should correspond to a 12:1-13:1 A/F ratio.

Q: What EGT readings should I expect?
A: At idle, you may see anywhere from 800-1000°F. In city driving, it may range from 1000-1300°F. On the highway, you may get anywhere from 1200°F to slightly over 1500°F. Don't be alarmed if you do see this, as the engine is under light load and will happily run this way forever. We already know what readings to expect under power, since we will have tuned for 1400-1500°F, right?

Q: Will the S-AFC lose its settings? (Does the S-AFC have a CMOS-type memory backup battery?)
A: All settings are stored in non-volatile programmable memory (EEPROM), so there is no risk of losing your settings if your car battery dies or if you disconnect the battery. (There are no batteries in the S-AFC.)

S-AFC Sensor Types

Hot Wire

I have attempted to characterize S-AFC airflow readings for various Hot Wire sensor types. Hopefully, this will clear up some of the confusion regarding sensor types for the Subaru Impreza. My S-AFC did not come with a list for the EJ22 or EJ25 engines sold into the North American market. Types 1, 4, and 10 are listed in the Instruction Manual for various Subaru EJ20 engines, while Type 6 is listed for Nissan SR20DET engines used in the S13 cars (Silvia, 180SX).

However, A'PEXi does have a supplemental manual for many North American cars. This guide still only lists the EJ25 '96-'98 engine (as Type 4), so EJ22 owners like me are still left scratching their heads. I have attempted to compare the various settings to see exactly how different each of them are. Note that this is a comparison of the voltage map internal to the S-AFC, which will show what percentage airflow the S-AFC will display depending on the input voltage.

From these results, we find that:

• The responses for the selected Hot Wire sensor types are very similar. The main differences are in the lower voltage readings. Otherwise, there is at most a 4% variation (of full scale) between these four types. At the higher ranges, they are similar enough that I wouldn't be too concerned about it.
• Threshold responses differ between the various sensor types. This is the point where a MAF's output corresponds to measurable airflow. Some sensors will report an airflow from as low as 0.1V (Type 10), while others require at least 0.3V (Types 1 and 4). I do not see how this variation will affect S-AFC operation, since most S-AFC users will be tuning at much higer MAF readings.
• The Airflow % will never read 100%, since a full 5.0V input corresponds to between 92% and 94% airflow. I do not know why this is so, and don't be concerned if you can't extract those last few ounces of air into your engine.
• The % reading you see on the S-AFC display is really meaningless. It provides a qualitative view into how the airflow is changing, but you really don't know if it's the correct amount. That's why other sensors such as the O2 and EGT sensors are needed for tuning purposes.
• Under certain conditions, it's possible to max out the correction factor, which may leave you wondering why you're not getting more/less fuel than you had hoped. For example, say the MAF is reading 4.75V, which corresponds to about 80% airflow, and your correction at that setting is 50%. In this case, the higest voltage the ECU will see is still 5V, and not 6V. Therefore, the engine will only see 95% airflow instead of 120%.

The next step is to perform airflow measurements to characterize the MAF sensor itself. Unfortunately, I do not have easy access to equipment for taking such measurements. Therefore, it's safe to say that Types 1 and 4 are similar enough that you will not notice any difference between the two. However, you may want to try out Type 10 and see if that has any effect on low-speed drivability.

Pressure

Along the same lines as above, I characterized the Type 5 setting, as this is the sensor to be used with 2000 and newer Impreza's. In order to not confuse Excel, I had to convert the boost numbers in kg/cm² to the equivalent in millimeters of mercury (mm Hg). I also added data for the Type 6 sensor, which is widely used in Honda models.

What do we learn here?

• The Type 5 sensor maxes out at 0.26 kg/cm², which corresponds to about 3.7 psi of boost. Therefore, MAP corrections probably won't work at any higher boost levels.
• The transfer function is very linear, all the way from full 760 mm Hg of vacuum to 0.26 kg/cm² boost. This is good. I should characterize a few more sensor types for comparison.
• There is quite a difference between Types 5 and 6 sensors. Although both snesors have linear responses, the slopes are quite different. Perhaps one can take advantage of the In/Out feature to replace, say, the Type 5 sensor with a Type 6 sensor. In this case, the S-AFC setting would be In 6 and Out 5. In theory, the ECU would not know the difference from a stock pressure sensor.