Fred wrote:Delta wrote:while it may look like the gate is set at a fixed voltage thats not the case. A comparator cannot remain at a fixed voltage (which is whats driving the gate). it drives towards either rail. The voltage at the base of the solenoid indicates that the mosfet is infact turning on and off quickly.
I didn't pick up on that either
<snip>The low resistance required to empty the capacitor which is supporting your freewheeling in a reasonable time to stop the injector overrunning is far more wasteful heat wise than the MOSFET being part open and will probably require several 5/10W resistors (very large) in parrallel.
Any chance of a rerun of that in more plain terms for me please?
The LM19?? sorry can't remember it off the top of my head doesn't support MOSFET's as the low side driver directly, only BJT/Darlington Pair. Driving it through a two resistor network to allow switching still doesn't allow driving mosfets.......not sure why this part was chosen as a candidate if were using mosfets????
This is my fault, I (wrongly apparently) assumed that it would just work. I'm sorry for causing confusion there.
Well done! Now, what sort of dissipation are we talking in this circuit and how large are the snubber components? Can you run through some more details of this setup? I think I like it, looking forward to seeing Jean's response!
Fred.
The snubber resistors should be 1W for safety, and the Capacitors should be non-polarised and rated to 100V - once again for safety. The diode in the DRC network should be able to support a flow of 10A peak and say 2A continuous (just incase someone wants 8A/2A).
To explain what happens when you do on-pwm is quite difficult - I wasn't very clear, so I'll try again hahha.
Basically at the beginning of the cycle the FET is turned on, which causes current to rise proportional to the inductance and resistance in the injector. When you get to 4A (in this case) it regulates it to 4A by switching the fet on and off, and then after some period (or as soon as you reach the 4A in some systems) it switches on and off at a rate that lowers the limit to 1A. Now - when the FET is switched off without any other components two things happen - The current stops flowing through the inductor - which causes the second thing - the voltage at the bottom of the inductor (which wants current to continue flowing) rises sharply (1-1.5kV). This spike is very destructive (although in the case of the autofet it won't spike high due to a reverse biased shottky diode being part of the bulk - but this diode has a capacitance to ground which is undesired). so we need to reduce it (preferably to below the autofets shottky barrier). Most people do this by using a reasonable sized capacitor. Each time switching occurs (which is frequent if your doing PWM to limit current) the capacitor absorbs current so that the voltage doesn't rise to high. (this means a very large capacitor is required - but we'll come to that later) You then need a way to discharge the capacitor. This is done by connecting a resistor in parallel to the capacitor. If the capacitor is very large, the resistor has to be very small to discharge it quickly meaning massive power. (once again more on this later) The resistor capacitor network also supresses ringing - ie it damps the oscillations that occur when you switch it off completely as its trying to basically make the voltage on both sides of the network equal. So that takes care of the ringing and the spikes. During the off time of the PWM the inductor still needs to have 1A flowing through it - or the injector pintle will close and we will need 4A again to open it. To achieve this you need a flyback diode. If you just connect a diode, then the capacitance built up at the junction of the diode and its freewheeling ability, means all thats removing power from the system is the resistance of the inductor - as when freewheeling most current flow through the diode rather than through the RC network. This means turn off time is greatly increased. To counter this, we have another Resistor limiting the freewheel current - this however means that the current lowers too quickly, so we take advantage of the fact that the voltage spikes when the inductor is turned turned off to store a large amount of current current in the diode side also. The size of the capacitor to alow this current flow is quite large. The over all capacitance has to be large (but its mostly on the diode side and used to keep the inductor open) so the RC snubber can have a lower capacitance and hence higher shunt resistor value - this means small power loss as we supress ringing, and fast response to spike voltage. The DRC network has a much larger capacitor but a much smaller resistor, however because the diode forces this side to shunt to the voltage rail, at turn off time after the transient spike has been eliminated it can dissipate to BOTH the inductor AND the RC network. This decreases final turn off time.
To allow all this to happen without having huge current draw through the DRC side the mosfet must be held close to its on position and switched in and out of the on position very quickly - infact as fast as possible while keeping current at the desired limit. The only was to achieve this (as I now know) is to use a fast op-amp or comparator to quickly drive it to the on position, then back down to near the threshold. At lower currents its possible to do it many ways, but when this much current is flowing around the loop you either have to switch very fast - or have HUGE power ratings on the snubber and RCD snubber networks so you can switch slowly and still keep the inductor current flowing.
davebmw wrote:Also the supply rail for U1 should be on the clean supply, but thats just me being bloody picky!
Like I said, if your driving a normal MOSFET then you leave it attached to Vbat - I don't have an autofet in the LTspice simulator so I sim like that. If we're driving an AutoFET then yeah it will be powered by 5V.
I hope that explains this a little better.....if not..then ummm...I'm stuck hahaha.
I can't improve on what you have done there it has excellent noise rejection and does what it should.
