stumbled across this... simple but interesting read on injector driver circuit design and this exact problem, specifically the slowing of the injector close speed due to draining of the flyback voltge through the other injectors :
http://www.bgsoflex.com/v22/msv22.html
the exerpt:
"If you look on the "output" circuit page, you will notice a modification to the method of which flyback voltages transients generated by the injectors are "clamped". On the V1.01 boards, each driver had a diode which was back-biased directly across the injector. This diode position in the circuit worked well in clamping the kickback produced when the injector current is released, which protects the driver FET from overvoltage, as well as operating very cool during PWM current-limit mode, since the voltage drop is only 0.7 volts. However, this circuit configuration recycled the kickback current through the injector(s), which tends to lengthen the closing time - the amount of time depends on the number of injectors and their impedance, but the delay is in the tenths of milliseconds region. This was not a problem in most applications, because the injector close time was a constant value each time the injector cycled, and showed up as an offset which was simply tuned out when setting up the VE table.
However, there were a few installations where large injectors were employed, with the resulting idle pulsewidths down in the 1.0 millisecond region. The problem here is with such a short pulsewidth, it is difficult to tune the idle due to both the resolution of the MS system (0.1 ms time) and the fact that short open times are dominated by fuel delivered during the opening time of the injector. So, for large injector setups with short idle pulsewidths, the slower injector closing time imposed by the simple recirculation diode increased the magnitude of this effect.
So, a more conventional flyback diode placement was tried in the field, where a zener is placed backwards around the driver FET. What occurs is when the injector closes, the flyback voltage builds until it exceeds the zener diode breakdown voltage (around 36 volts or so) and then the excess voltage is conducted around the FET. This allows the injector to close faster than in the recirculating setup on the V1.01 board. And, the resulting faster closing time helps out those with short idle pulsewidths.
Now, the wrinkle of this is that the MS utilizes a Pulse-Width Modulation (PWM) mode for current-limiting of low-impedance injector setups. When the PWM mode is activated, the resulting PWM waveform causes a flyback spike for each on-to-off current transistion, which has to be clamped. So, the power dissipation requirements of the zener diode is significant when the PWM mode is active. Trials indicated that even 5 watt zener diodes do not survive long with certain injector setups.
To help with the power-dissipation issue, a transistor curcuit was employed to sink the heat generated during the PWM current limit. The transistor Q1 (PNP) acts as a large zener diode - the resistor from emitter to base keeps the transistor off until the voltage exceeds the breakdown rating of the zener diode between the base and collector. This transistor (TO-220 package) is heat-sinked to the ground plane of the board, which provides plenty of power dissipation for any injector setup. So, the result is fast injector closing time and good power dissipation for flyback currents."
[%sig%]
Post Edited (05-31-03 13:58)
