# Automotive MOSFET active clamping circuit - LC ringing between coil load and MOSFET gate?

I'm trying to design a circuit equivavlent to fully clamped automotive MOSFETs (like the STP62NS04Z) using discreet, cheap and readily available IRFZxx MOSFETS. The goal is to drive a high impedance fuel injector which is equivalent to a resistor and a inductor in series. The inductance is about 8.5mH and the resistance is about 12 Ohms.

I designed this circuit:

The automotive fuel injector is supplied with a 14.4V constant supply and is switched to ground. When the load switches off, the zener clamps the output voltage to ~Vds - 10V and the MOSFET turns on, bleeding this potential energy to ground.

What I'm experiencing is ringing between some components of the circuit:

The red line is the microcontroller output, the blue line is the injector (load) current and the green line is the MOSFET gate voltage supplied via the zener. The voltage at the gate is oscillating for ~0.5ms after the inductor is bled of energy.

Is this ringing caused by a LC ringing circtuit between the MOSFET gate capacitance and the inductance of the injector? Will this cause any issues with MOSFET gate drivers like the generic TC4424 (COS4427 and similar)?

There is a small AC current on the circuit input:

Can this damage the MOSFET drivers?

Try an R+C from drain to source. Exact values depend on cable length, solenoid impedance, etc., but typical values would be $$\C \ge 3 C_\textrm{oss}\$$ and $$\R = \sqrt{L \,/\, C_\textrm{oss}}\$$, or about 3k and 1nF here.

Larger snubber C can be used to slow the free rise / ringdown time, which may help with EMI. Different resistor values will apply for different rates/capacitors, or for different purposes.

There can even be reason to use multiple R+Cs. A more dramatic illustration of this comes from switching converters, where the load is inductive (e.g. flyback transformer), and the clamp diode is secondary-referred; thus the primary experiences an initial ringdown on top of the turn-off edge (Coss ringing with transformer leakage), followed by a slower free-ringdown after the flyback pulse terminates (Coss + diode CJ ringing with magnetizing inductance).

The nearest application of this, for automotive use, would probably be if the wiring is long enough that its transmission line impedance is relevant (i.e., R ~ 100Ω, C several times the line equivalent capacitance), and then a different value (larger R and C?) to snub the load itself. But mostly, in this context, switching times can be made slow enough that transmission line effects don't matter.

In any case, the gate driver is more than capable of handling mere ~mA of backflow. A fancy gate driver is hardly required at all; 12V gate drive into 1k resistor is only 12mA, even a CD4000 series gate can deliver on that. The main annoyance is that, there's no TTL-input CD4000 gate, and you'll probably end up with some discrete transistors to do the level shifting instead, and it's a big pain in the [component count], especially if you need a lot of output channels.

Also mind that there's not really any good substitute for the integrated thermal protection, or current limiting or fault detection, that the protected MOS devices can provide; they can be constructed from discrete components even, if you like, but the component count isn't going to be at all competitive, at least anywhere most commodity ICs are available. There are some eFuse and hot-plug / load-switch controllers available that provide, or approximate, this functionality (e.g., sensing and limiting current, and calculating dissipated power, as a stand-in for on-die temperature sensing), that would be most beneficial here.

Or, to put it another way -- what you've done so far is provide load clamping, and potentially transient protection (I don't like the idea of ESD or surge going directly into the gate, but the MOSFET may well act fast enough to protect itself here; I'd want to test it extensively before committing to production!), but no short-circuit or cross-wiring fault protection or current limiting or anything. (Maybe those are next on the list or something, I don't know. This, more just to make it clear to any readers looking for such alternatives.)

• Thanks! Adding a RC snubber almost completely removed the ringing issue, now limiting it to 2-3 cycles. Amazing! Thanks. What does the Coss stand for? Capacitance of oscillation? How did you calculate the 3k/1nF values? A fancy gate driver was chosen to be able to switch from 5V logic (AVR) to 3.3V logic (ARM) during the prototyling/testing phase, to keep the main driver board the same (the uC is on a daughterboard on the prototypes). The ST part i mentioned is a semi-smart MOSFET, it doesn't offer any overload/temp protection but merely clamping/ESD protection only, that was the goal :) Commented Dec 5, 2023 at 20:47
• Coss is the MOSFET parameter, drain/output capacitance. Typically ~400pF for IRFZ46N as example, but it depends on Vds as well. | You can use 5V logic directly with logic-level parts e.g. IRLZ44NPBF, or boost with e.g. 74HCT2G17 to use 3.3V CMOS or TTL-level inputs. | Ah, only clamping, that covers it then. Commented Dec 5, 2023 at 21:24