VP44 EXPLORER

My project replaces the control board that sits on top of the pump with a remote board. The project uses an STM32f303 Nucleo dev. Board and I used Arduino IDE. A custom PCB was made in two parts. One is the main logic and sensor interface. The second part is the solenoid drivers and power supply. The current driver board uses a high-speed switching buck converter to operate the fuel solenoid. A linear drive board is under development. The linear method is the OEM style. It generates a lot of heat and is cooled by the fuel flowing under it in the pump housing. The switching method results in significantly less heat. The downside is the switching signal can mask the current monitoring of the solenoid. As the armature plunger moves it causes a change in inductance which causes a small current blip. The blip is useful to determine the solenoid dead time.

The VP44 shaft has a toothed wheel, the tooth spacing is 3 degrees of crankshaft rotation. Every 60 degrees (120 at crankshaft) is a pair of staggered teeth. These I refer to as the index tooth. A dual magneto-resistive sensor with a back bias magnet senses the teeth as they pass by and creates two sinewaves. The one that leads I refer to as phase A. The second follows by approximately 90 degrees, I refer to it as phase B. Both signals are adaptive zero crossing detected by internal opAmps and the DACs in the STM32 and then fed to timer captures. The software detects the index tooth region and resets a tooth edge counter to zero. On the tooth edge 19 of the B phase the radial injection plungers are at their maximum extension. As they follow the cam ring into compression it's at this position the injection solenoid valve must be fully transitioned to the closed position. This is the start of injection SOI. After a variable amount of degrees, the valve de-energizes this is the end of injection EOI.

Since the solenoid has deadtime on both turn on and turn off the switching must occur before edge 19 or EOI actually happens. This is done by computing shaft position, velocity, and acceleration to predict the relative time from a current tooth edge to edge 19. Obviously, the closer we are to the edge 19 event the better the prediction will be, but at higher RPMs the deadtime of the solenoid results in the required prediction to occur many degrees in advance else the valve will be closed too late or early. That results in late injection, reduced fuel delivery and hazy smoke. The valve must be de-energized before the moment of desired EOI because of mechanical delay I call this release time, which is about 850 microseconds. The dead time is mostly dependent on the battery voltage and must be calculated often. A full current is applied to get the valve moving then the current is reduced to prevent the valve from slamming on its seat and bouncing. Finally, the current is reduced to a safe level for the majority of the injection event. The switch-off induction spike at EOI must be clamped at a higher voltage and low current for a fast release. The magnetic field is determined largely by A x N turns, so keeping current low speeds the armature release. Simple clamping / recirculating methods are too slow. A zener diode RC circuit is used. Some manual adjustments to times and current values must be determined with an oscilloscope and are stored.