Thanks for the kind words Dave!!
I've found a few interesting things with the MS45.0 and VF kit recently:
1.
The MAF sensor scaling maxes out at 1024 kg/hr. This equates to 137% VE (5.4 psi on a 3 liter motor) at 6800 rpm under standard conditions. Any airflow beyond this amount is "off the charts" and not read by the ECU, which can lead to a dangerous lean condition.
2. I calculated the maximum boost level using the 3.6" pulley provided with my V3 SCi impeller.
- 6800 RPM redline
- 3.61 step up ratio
- 3.6" supercharger pulley
- 5.5" crank pulley
I found the impeller speed to be 37400 rpm. Using the compressor map available on Vortech's website (link), 37400 rpm corresponds to a pressure ratio of 1.55 (8 psi of boost).
So, my VF kit in standard form put out 8 psi of boost at redline.
3. I pulled up my dyno from years ago on the original tune and pulley (picture attached). As you can see, the car was pig rich (10.0 AFR) until ~6300 rpm, where it started to lean out. At 6300 rpm, the impeller speed is 34650. Using the compressor map, that equates to a pressure ratio of 1.46 (6.8 psi). 1024 kg/hr (max reading of MAF sensor) at 6300 rpm equates to a VE of 121% (about 7 psi of boost).
Conclusion – my original dyno backs up the theory that the MAF sensor is maxed out, even with a basic VF kit.
4.
Axes values for ignition timing, fueling, and VANOS all max out at 700 mg/stk. While the MAF sensor's scaling is in units of kg/hr, the tables with a load axis (timing, fuel, VANOS, etc.) are all in mg/stk. A conversion is required, but it's not difficult, and it's dependent on rpm:
Mass Airflow (g/cyl) = #Cylinders * Mass Airflow (g/s) * 60 / (RPM*2)
When the MAF maxes out at 6300 rpm, the tables are already well beyond maxed out at 903 mg/stk. I'm working on rescaling the axes, which increases range at the expense of reduced resolution between cells. Max values for the axes are easy, but only significant data logging will determine the optimal sites (steps between max and min).
5.
There are injector correction factors for each injector, and a global injection correction factor. I have seen some tunes that have these mixed up, which risks a single cylinder not receiving the fuel it needs and running extremely lean / detonating, and a wideband after the collector wouldn't necessarily pick it up because it averages the header bank's overall AFR. A 15% lean mixture in a single cylinder would only show as being 5% lean by a wideband installed after the primaries merge (2.5% lean after a y-pipe). A wideband installed on the other bank wouldn't pick up anything.
6.
The MS45.0 uses narrowband primary O2 sensors, whereas the MS45.1 uses wideband primary O2's. This is a limitation for the MS45.0. It's possible to make big power on a dyno queen, but drivability and reliability will never be as good as a car with wideband O2 sensors. A narrowband sensor can only read AFR across a narrow band, right around stoichiometric (14.7). If the AFR deviates a few tenths above or below stoichiometric, the car only knows that the mixture is lean or rich, but not by how much.
The ECU's tables are in units of lambda. To move from lambda to AFR, multiply by the stoichiometric ratio of the fuel you're using (lambda 1.0 = 14.7 AFR on pump fuel).
There are four primary fueling tables in the MS45.0: closed loop, open loop, transition, and correction factors. Other fuel tables exist, like cold start and cranking, but they don't necessitate discussion here.
The closed loop tables are all lambda 1.0. This is what the ECU targets at cruise, and the ECU will use O2 sensor feedback to try and keep the mixture from cycling slightly lean to slightly rich. I could go into a long discussion about how this helps emissions (NOT efficiency) via outputting different types of toxic emissions when lean vs rich, that combine and produce CO2 and water in the catalytic converter… but I'll save it
. The point is that with narrowband O2 sensors, lambda 1.0 is the only mixture you can target in closed loop mode.
In open loop mode, the ECU makes a calculation based on load and rpm, and it pulses the injectors to provide the appropriate amount of fuel without any feedback. So, you better be sure your injector scaling is correct, because the ECU is blindly guessing as to how much fuel it needs to dump. This is where an aftermarket wideband sensor and AFR monitoring becomes so important. The ECU has no way to know how close it is to its AFR target. This table is 2D, not 3D, meaning that it only considers RPM (and not load) with a 1x16 matrix (really a vector).
Another table is the transition or "full load threshold" table, which determines when the ECU moves from closed loop to open loop mode. This is also a 2D table, and it's based on throttle position (not accelerator pedal position).
The last table I'll talk about is a correction factor table, with load and rpm axes. This helps the ECU be more accurate with its fueling, but doesn't necessarily change its target AFR. I have seen it used in a roundabout way of increasing fueling for full load when the MAF sensor is maxed out. The table axes also peak at 700 mg/stk, so a basic VF kit is off the chart at high rpm.
Again, there are many other tables in the MS45.0, such as fueling tables for when VANOS is active vs. not, and so on. There are separate 91 RON and 98 RON tables.
What does this mean? The MS45.0 ECU either targets lambda 1.0, or a WOT mixture (based on rpm), and nothing in between. This transition isn't very noticeable on a stock car, but as boost increases, it can become more abrupt. Even more critically, ignition timing is based on load and rpm only, and doesn't take into account whether the ECU is in closed or open loop mode. It's possible to be in boost at part throttle. Advancing ignition timing when the ECU is targeting an AFR of 14.7 on a boosted engine is dangerous, especially in a high compression engine that was originally naturally aspirated (and thus more prone to detonation). From a reliability standpoint, the only things to do are reduce ignition timing to a safe value for both closed and open loop (limiting power), or set the WOT transition point to occur at a lower throttle input (affecting economy and power at part throttle).
This is how it's possible to tune a car to make great power on the dyno but sacrifice reliability on the street.
7. Fortunately, the knock sensors seem to be pretty good at picking up early signs of detonation and retarding ignition timing.
I'm working with Dave to rescale the appropriate axes, and I'm working on a hardware solution to the MAF issue. By the way, I had the green VF fuel injectors tested, and they are 30 lb/hr at 100% duty cycle with 43.5 psi fuel pressure (BMW runs higher fuel pressure than this, around 50 psi IIRC). Variance was 4.3% at 20% duty cycle after cleaning, and 2.8% during static flow. They will be too small for my power goals, so I'll be working on an injector solution too.