Open

Open

COBB Protuner Support Page: http://www.cobbtuning.com/protuner-support/

• This page is maintained by COBB's Director of Protuning and includes web links for our Accesstuner Pro software builds, information about COBB-provide training options, links to various logos to use within your marketing materials, etc.

Factory EMS and Hardware Overview

Spark and ignition control is straightforward on modern turbocharged Subarus; crank and camshaft position are measured via reluctor and Hall-effect sensors, depending on model/year, and are then synchronized within the ECU, which operates four coil-on-pack ignitorless coils in sequential firing order. Ignition system upgrades are rarely needed as the factory hardware has been proven shown itself to not only be durable but also capable of supporting roughly triple the factory power levels with no meaningful degradation to spark energy or combustion quality.

All turbocharged Subarus for USDM markets to-date (MY2002-MY2014) feature port injection, controlled a traditional 3 bar bypass fuel pressure regulator with manifold reference rate of 1:1. The factory fuel rails are in a series configuration. A traditional in-tank hanger, located in the passenger side of the saddle-style fuel tank, handles fuel delivery from the tank. Like other Japanese vehicles, this assembly also contains a "siphon" system operated by the returning fuel flow to sump the driver's side fuel tank reserves. Fuel starvation during right-hand turns is a known limitation of the design; many track vehicles and fast street cars use an auxiliary surge tank to alleviate this issue.

A PWM controller that is signaled by the ECU regulates fuel pump speed. The ECU can command three unique low-middle-high states, which are 33.3%, 66.6% and 100.0% from the factory but can be edited as necessary during fuel system calibration. Switching between states is primarily controlled via IPW or IDC thresholds and sometimes RPM breakpoints, model dependent. The pump is triggered to "high" for a period after startup and falls to "low" (assuming state thresholds have not been wildly altered) after the "startup" or "hot-start" periods end. The elaborate speed control is presumably used to promote pump life, decrease fuel heating and lessen FPR overrun (the factory FPR has a short spring and small orifice) during periods of low fuel demand.

Factory specified injectors vary, mostly along vehicle generation lines, though all are provided by Denso. The 2002-2005 WRX utilized ~440cc top-feed injectors (light blue). The 2004-2006 STI, 2005-2006 LGT, and 2004-2005 FXT featured ~550cc side-feed injectors (yellow). The 2007-2014 STI, 2006-2014 WRX, 2006-2013 FXT and 2007-2012 LGT use ~565cc top-feed injectors (dark blue), though the part numbers and body coloring may have changed across a few years. Uprated side-feed injectors have proven to be particularly problematic; even the factory side-feed injectors are known to have a high failure rate. Top feed conversions are common practice, even for otherwise-stock vehicles.

Mass Air Flow is measured by a traditional Denso-provided integrated intake air temperature and hot-wire mass airflow meter. This is located in an "airbox" MAF housing from the factory or traditional "filter-on-a-stick" intake housing in the aftermarket. No post-compressor or post-IC temperature pickups are used unless the vehicle has been converted to use our custom Speed Density code.  Speed Density is when an alternate IAT sensor is installed into the post-IC charge piping or intake manifold. Accurate Mass AirFlow measurement, meaning correct MAF recalibration following intake hardware changes, is exceptionally critical to accurate fuel calculations and delivery.

Our custom Speed Density code allows for the computation of mass airflow and load (which are functions of one another in the Subaru ECU) based on the traditional PV = nRT and volumetric efficiency models. When using the Speed Density method for tuning, Mass AirFlow (and thus load) is calculated using a Volumetric Efficiency model that includes adjustable engine displacement, air temperature compensations, coolant temperature compensations and others. Assuming proper end-user input of critical items of said variables, the model is highly effective for very precise mass airflow estimation. This style of tuning is traditionally used by those whose power demands exceed that which a MAF housing of ~80mm or smaller internal diameter can provide or those who require greatly customized turbocharger and intake tract configurations that do not allow for proper draw-through MAF or recirculating BPV positioning.

Open and Closed Loop operation is determined by various thresholds; open loop traditionally is enabled under higher load and "boost"; closed loop covers idle and cruising conditions. The closed loop system is inherently limited by the primary O2 sensor's location in the high-pressure pre-turbocharger exhaust piping without correlating EGBP pickup and compensation, thus its accuracy is largely limited to when intake manifold is under vacuum. Lambda target is determined in either mode by look-up tables and compensations that lead into a final fuel multiplier (Fuel-Air Equivalence Ratio) coefficient.

Boost control is also a simplistic yet effective system. The factory "2-port" boost control solenoid, which is plumbed via bleed style operation and is powered by the ECU via 12v PWM signal, operates on a 0-100% duty cycle scale using a closed loop boost targeting system. When energized, the solenoid is a low-pressure vent between the compressor nipple (boost reference) and the wastegate actuator, causing effective pressure at the actuator to decrease and boost to rise. A variety of restrictor pills are used within the boost control vacuum lines by the factory to fine-tune the system's responsiveness. A base wastegate duty cycle value is referenced, actual vs. target boost is evaluated, and corrections or "dynamics" are applied to correct for the boost error. This system is commonly upgraded to a plug-and-play "3-port" boost control solenoid with vacuum lines replumbed to operate the actuator in interrupt mode, where 100% wastegate duty cycle represents a complete blockage of boost reference to the wastegate actuator. The factory control system is highly scalable; it has proven effective for controlling small turbos making less than 15psi and 200whp all the way up to full motorsports vehicles running in excess of 30psi and 600whp. More detail about the factory system is available a thorough document entitled "How Subaru's Factory Boost Control System Works" by Christian Krahenbuhl.

Ignition Timing Control

Ignition advance, and total applied timing, is calculated and controlled by a somewhat complex strategy within the factory ECUs logic. With some basic knowledge and a clear understanding of the various tables and monitors involved, however, the system is not only very accurate, but it is also exceptionally easy to tune. Here are the basic formulas for how total ignition advance is calculated:

• Total Ignition Timing (all 2.5L): Primary Ignition + (Dynamic Advance * DAM) + Feedback Knock Correction + Fine Knock Learning + Compensations

• Total Ignition Timing (only 2.0L): Primary Ignition + (Dynamic Advance * (DAM/16)) + Feedback Knock Correction + Fine Knock Learning + Compensations

Terminology Definitions

• Total Ignition Timing: The total ignition timing value, in crank angle degrees relative to top dead center before or after the beginning of the power cycle. Positive values represent ignition "advance" or BTDC ignition events; negative values represent ignition "retard" or ATDC ignition events. Negative values are typically not used on the Subaru EJ series engine. This monitor for this value on-vehicle is "Ignition Timing" within Accesstuner and Accessport logs; it is important to note the logged value is for Cylinder #1 specifically and actual timing for other cylinders may vary, depending on per-cylinder compensation tables and thresholds.
  
  

• Primary Ignition: This value is defined by a traditional 3D table with Load as the x-axis, RPM as the y-axis and degrees of ignition timing as the z-data. This is a "Real-time" tunable table for all ECUs and also appears under the Advance (Primary) submenu. Depending on the vehicle, alternate Primary Ignition tables based on the TGVs Open/Closed status will be referenced. It is generally advised to only adjust the TGVs Open ignition tables as these will be the ones used primarily as far as performance or aftermarket tuning needs; in the absence of additional data, the TGVs Closed tables can be assumed to have been well-tuned from the factory.

Assuming the Dynamic Advance (see below) tables have been tuned to a pre-determined level before beginning on-vehicle tuning, most ignition timing adjustment will be confined to the Primary Ignition table during the calibration process.

• Dynamic Advance (DA): This value represents a global addition to the base/primary ignition value and is the maximum DA value allowed. Unique logic within all current years of STi can cause potentially unstable ignition timing or undesired increases to overall current DA value being used. Depending on the year, the STi has one or more Dynamic Advance Max. Adder tables as well as one or more Dynamic Advance Max. Primary tables.

Because the DA value applied is multiplied against the DAM value, the general shape of the primary DA tables is by and large determined by the relative probability of inducing detonation under given conditions. When the engine load is low, or RPM is high, the engine is more resistant to detonation. When the engine load is high or RPM is low, knock is more likely. With this, the factory generally uses the most DA where torque is highest, in order to achieve the largest dependence and response to DAM changes. In general, the Dynamic Advance tables are "pre-tuned" based on the calibrator's preference, then the Primary Ignition table is shaped around it as such to generate the total final Ignition Timing value. Spreadsheet programs such as Excel can be very useful for overlaying the Primary Ignition and Dynamic Advance tables to help visualize the total timing value requested as well as look for any anomalous or insane values.
See "STi-Specific Considerations" below for additional information specific to the STi models.

• Dynamic Advance Multiplier (DAM): This value represents a global adjustment to the Dynamic Advance component ignition timing. In general, this value is determined by historical detonation/knock as well as the default mapping value. For the 2.5L ECU, DAM is stated as a decimal ranging from 0 to 1. For the 2.0L ECU, DAM is stated as an integer between 0 and 16. DAM will tend to vary the most immediately following a reflash or ECU Reset procedure. The starting value for DAM is an adjustable value within the Ignition Tables -> Advance (Dynamic) submenu. Various thresholds, including RPM and load, and determine when DAM can or should be evaluated for potential increases or decreases.

One of the more critical DAM evaluation thresholds to observe is based on the current requested Dynamic Advance value. When Dynamic Advance requested is below the threshold value as determined by the Course Knock Learning (DAM) Modify (Min. Din. Adv. Map Value) table, DAM will not be evaluated. This can cause DAM to be "stuck" at too low or high of a value and a global timing error results.

• Feedback Knock Correction (FKC): This value represents a current real-time minor timing correction made by the ECU in response to a perceived noise. Feedback Knock Correction is the default correction used by the ECU. The ECU will "respond" with this monitor by immediately removing timing and then slowly decaying the value back to zero assuming no further noise is detected. The amount of timing removed and re-added per incremental change are both adjustable values within the Knock Control submenu. Various thresholds, including RPM and load, determine when Feedback Knock Correction can or should be evaluated for potential increases or decreases.
  
  

• Fine Knock Learning (FKL): This value represents minor learned corrections currently being applied by the ECU as a product of historical noise. These values are reset if the DAM value is changed; once DAM has stabilized, any learned corrections will be intermittently evaluated for sanity when certain thresholds/conditions are met. Fine Knock Learning is learned (and thus applied) for specific ranges defined by load and RPM; once a correction value has been learned into a particular range, it will be applied whenever the engine is operated within that range. The amount of timing removed and added per incremental change are both adjustable values within the Knock Control submenu. Various thresholds, including RPM and load, determine when Fine Knock Learning can or should be evaluated for potential increases or decreases. Fine Knock learning values can only be negative when DAM is at its maximum value; both negative and positive values can be observed when DAM is below 16 (2L) or 1 (2.5L).

The Fine Knock Learning table is used by the ECU to store learned Dynamic Advance information. The ECU populates it with adjustments to total timing in degrees. This table uses engine RPM for the y-axis and calculated load for the x-axis. The ECU can make positive or negative adjustments to ignition advance in this table. If the ECU determines the motor can be further optimized by running additional ignition advance for a given RPM and calculated load, then it will populate this table with positive values, but only when DAM is below 16 (2L) or 1 (2.5L). If the ECU determines correctable detonation has occurred for a given RPM and calculated load, then it will populate this table with negative values. This fine adjustment allows for memorized corrections to prevent the need for a reactive knock correction response each time the engine is subjected to similar operating load/RPM conditions. This table, and any newly-learned corrections, can be refreshed while you are connected for live tuning with the by pressing the F5 key. The stored values within this table are helpful for fine-tuning of the Primary Ignition table based on the detonation threshold observed over time.
The learned values within this table are cleared whenever the ECU is reset as well as any time that Dynamic Advance Multiplier is adjusted.

• Knock Sum (where applicable): This is a somewhat arbitrary monitor and logging value that should only be analyzed under the specific conditions you wish to evaluate, such as during wide-open throttle (WOT) operation over a set RPM range. Some later ECUs can report Knock Sum on a per-cylinder level, some only on a global basis. It is useful to determine if a particular cylinder is especially prone to detonation in relation to the others; however, it must be noted that Cylinders #1-3 lack knock detection accuracy due to their relative distance from the knock sensor. Cylinder #4 can be heard most reliably thanks to being located directly below the knock sensor, which has a difficult time perceiving real engine noise over the noisy boxer engine on the more distant cylinders. If noise is perceived, this value will increment at all times, regardless of if the knock detection system is deemed to be accurate or not. It is not uncommon to see these values increment under even the most mundane conditions, such as idling or while operating the vehicle at slow speeds in a parking lot. It must be emphasized that this is simply an indicator that a noise of unknown source has been detected, which must still be evaluated by the ECU for source, sanity and plausibility (IE, is it likely that the engine is actually detonating and on which cylinders).
  
  

• Compensations: Within the framework above, this is a summation of various ignition timing compensations. Depending on the year and model, these exist on as per-cylinder corrections, per-gear adjustments, intake air temperature based adjustments, etc. Depending on how the compensation tables are calibrated and the conditions under which the vehicle is operated, these compensations can wildly alter total timing or conversely have a very small effect. We tend to use minute or zero compensation for "normal" operating conditions – such as mild temperatures at sea level elevation on a fully warm but not overheating engine – and then add compensations for when those types of conditions are at more extreme values, such as relatively cool or hot air temperatures, high elevations or when the engine is overheating.

STi-Specific Considerations

All current years of STi ECUs contain additional logic can cause potentially unstable ignition timing or undesired increases to overall current timing being applied. Depending on the year, the STi models have one or more "Dynamic Advance Max. Adder" tables as well as one or more "Dynamic Advance Max. Primary" tables.

The logic used for determining when the "Dynamic Advance Max. Adder" tables are active is not straightforward, which means that values from these tables can be arbitrarily added to total current timing at undesirable times. As such, it is recommended that all "Dynamic Advance Max. Adder" tables should NOT be populated with any timing values in order ensure consistent total ignition advance. This can be easily done by selecting the entire table(s), then pressing "E" and entering the number "0".

The "Dynamic Advance Max. Primary" tables will always be applied, but the conditions to determine which table is active are not the same across years. The 2007-2011 STi vary with the TGVs Open/Closed status, which is a known set of thresholds such as Requested Torque and RPM; the 2004-2006 STi vary with a more complex and ambiguous set of thresholds based on historical knock. As such, setting all "Dynamic Advance Max. Primary" tables to the same values will ensure a consistent Dynamic Advance value is applied under all conditions.

Tuning Guidelines – Ignition Timing

During the calibration process, with specific regards to timing, it is recommended practice to always log (at least) these monitors:

1. Dynamic Advance Multiplier

2. Feedback Knock Correction

3. Fine Knock Learning

4. Ignition Timing

5. Engine Speed

6. Calculated Load

Additionally, to expedite the calibration process and reduce variability, it is recommended to set the default DAM value to 1 or 16, for 2.5L and 2.0L, respectively, for your base starting value and then performing an ECU reflash. Once this has been done, you can be certain that any changes/reductions to DAM, FKC or FKL have been caused by a recently detected event.

Assuming DAM stabilizes at its maximum value following the reflash/reset, pay very close attention to FKC and FKL during the calibration process. Depending on the model and year, the minimum timing increment varies between roughly -1 and -2 degrees. The decrement values can be viewed, or adjusted in unique circumstances, under the Knock Control submenus. Single increments of corrections to FKC and FKL will be observed from time to time during the calibration process; please note that FKL is truly a "learned" value and will be applied under those Load/RPM conditions until it is reevaluated and potentially removed. Intermittent single increments, such as -1.40 for a 2008-2014 STI, are not uncommon, but if proven repeatable under WOT conditions, an adjustment to the Primary Ignition table should be considered. Any value larger than the single increment is nearly certainly an indication of prominent noise and calibration adjustments should be made.

Open

Ex: Datalog demonstrating a single negative Feedback Knock Correction increment while performing a dyno calibration on a Stage2 2012 STi. This knock event was repeatable and only observed after raising ignition timing from a previous knock-free level while. Restoring the lower values eliminated the detonation. Note that the safety and power of the dyno tuning process allowed this operator to quickly abort a run once they observed anomalous operation.

If learned values persist within the FKL table, or DAM remains below its maximum value, fine-tuning can be difficult. The values within the Fine Knock Learning Table can be cleared using the "ECU Reset" function, which also returns DAM to its default value. Do note this will also clear all other learned trims such as A/F Learning #1 and DBW/TPS learning.

In general, the Subaru engine does not suffer from any extreme efficiency irregularities under full power, so a smooth timing curve free of unreasonable jumps or dips is usually desirable. Pay attention to the Ignition Timing monitor for large abrupt changes to ignition timing and smooth the Primary Ignition table accordingly. Closely monitor torque changes.

Pay careful attention to the various ignition timing compensation tables, in particular, the Ignition Timing Compensation (Intake Temperature) table. This can play an extremely large role in the timing values observed on the dyno, especially when dyno-room airflow causes an ambient vs. IAT temperature discrepancy. Conversely, it is important that the values in this table be appropriately set up to help reduce the potential for detonation as air temperatures rise. Other compensations, such as those for Per-Gear and Per-Cylinder adjustments, can be used advantageously to account for the increased mechanical load to the vehicle or airflow imbalances within the engine. However, you MUST account for how the compensation will adjust timing as those variables reach more extreme states, such as a very hot afternoon or while accelerating at WOT in 6^th gear.

Advanced Tuning Strategies – Ignition Timing

Now that we have outlined the various tables, monitors and logic that encompass the ignition timing and knock control strategies, we must actually adjust the timing curve and various thresholds in order to maximize efficiency and power (at the minimum cylinder pressure levels necessary for each) while minimizing the likelihood for inducing detonation. In practice, the objective of ignition timing calibration is to find MBT (or MTBT), which is Minimum timing for Best Torque. Finding MBT can only be safely accomplished with a good loading chassis dyno that has the ability to load the vehicle and measure the torque output at the same time. Chassis dynos such as Mustang Dynamometer, Mainline DynoLog, DynoCom, Bosch, and Dyno Dynamics have this ability, as do some traditionally inertia-based dynos such as Dynojets that are now being upgraded to include Eddy Current retarders. You can start in the higher gears (lower engine RPM) and have the chassis dyno to hold the vehicle and give you the torque output of the vehicle at an RPM breakpoint on your ECU calibration. You can start out at very light (low TPS) loads holding the vehicle in one specific cell in the Primary Ignition table and slowly add ignition advance until the vehicle does not make any more torque or gets close to the knock threshold for the motor. One suggestion is that you increase ignition advance for each particular cell until you see that torque no longer increases with the additional ignition advance. Now back off 2-4 degrees of ignition advance to keep the calibration on the safe side. Once you find where the motor produces the maximum amount of torque with the least amount of ignition advance, this is MBT. Tuning for cruising and partial throttle MBT will take a very, very long time and is not suggested unless you are very experienced with the particular chassis dyno you are using and the engine you are tuning. OEM calibration teams have vast resources at their disposal for calibrating the engine under these conditions, so their work is to be trusted in lieu of more accurate data. It is critical, however, to ensure that the ignition timing curve is very well calibrated over the entire "boosted" range of the engine, which can be accomplished by beginning WOT tuning at the minimum mechanical boost pressure, methodically adjusting each load range for MBT as boost pressures are increased. Please take into account that once you exceed MBT (Minimum spark advance for Best Torque output); it is possible to make less power with more ignition advance. Due to the various properties of the Subaru engine, it is common to encounter the detonation threshold before MBT when advancing ignition timing under boosted conditions. It is important to re-scale the load axis as necessary on the Primary Ignition and Dynamic Advance tables to ensure that you have suitable resolution for ignition timing adjustment when power levels greatly exceed that of the factory vehicle and load axis scaling.

Open

Ex: Here are the Primary Dynamic Advance TGVs Open and Primary Ignition TGVs Open table from a typical Stage2 2013 STi running on 93 octane gasoline. Note that the load axis has been rescaled to a maximum value of 3.60; peak load values of ~3.35 were observed during dyno tuning (stock load scaling ends at 2.95 g/rev). This extra resolution allows for an appropriate change to ignition timing should post-tune load values exceed those observed during the tuning session.