High Current Inductive Load Switching Issues

[T3sl4co1l]

Yeah the beefy resistor is quite out of place, but it's also a tiny fraction of the solenoid's power consumption, so, it's really quite meaningless in terms of overall efficiency.  Fun, huh?

Speaking of totem poles -- I would go even further, and recommend a 2-switch configuration:
https://e2e.ti.com/cfs-file/__key/communityserver-blogs-components-weblogfiles/00-00-00-07-88/2450.Capture-6.JPG
I would... but it is more complicated, and it's not necessary here.  So, this is just FYI.

Anyway, say you were to use this -- the advantage is, you have two transistors, with respective clamp diodes, all right there on the one board.  The transistors switch each end of the solenoid, so when they turn off, one end flybacks up (pushing into +V) and the other end flybacks down (pulling from GND).  You need a bypass cap to account for the sudden reversal (going from a +40A load, to a -40A return), but the turn-off is as fast as turn-on (faster, actually) because it's applying the full 26V, in reverse -- rather than shorting it out with a clamp diode (at ~1V).  And, all the energy stored in the solenoid is returned to the power supply -- not that that's a problem with a slow cycle like here, but it can be valuable for high speed driver applications (like print heads and injectors).

Tim

[Marco]

I think you want a diode with fast forward recovery time.

Diodes don't have forward recovery time. There's resistance modulation, which will start high but the higher the dI/dt the faster the resistance drops, the forward voltage is self limiting (although it can be still quite high).

[T3sl4co1l]

I think you want a diode with fast forward recovery time.

Diodes don't have forward recovery time. There's resistance modulation, which will start high but the higher the dI/dt the faster the resistance drops, the forward voltage is self limiting.

Well whatever it is, it looks the same -- I've (semi-accidentally) applied almost 60V to a diode before -- in about 20ns.  It looks like it has series inductance, but what would've been stored reactive energy is lost as heat in the diode instead, it doesn't come back out.

Tim

[Pizzashape23]

Just curious, but assuming I go the TVS route, could I put a fairly low value resistor in series with it across the mosfet to try and collapse the field faster?

[TimNJ]

Yeah the beefy resistor is quite out of place, but it's also a tiny fraction of the solenoid's power consumption, so, it's really quite meaningless in terms of overall efficiency.  Fun, huh?

Speaking of totem poles -- I would go even further, and recommend a 2-switch configuration:
https://e2e.ti.com/cfs-file/__key/communityserver-blogs-components-weblogfiles/00-00-00-07-88/2450.Capture-6.JPG
I would... but it is more complicated, and it's not necessary here.  So, this is just FYI.

Anyway, say you were to use this -- the advantage is, you have two transistors, with respective clamp diodes, all right there on the one board.  The transistors switch each end of the solenoid, so when they turn off, one end flybacks up (pushing into +V) and the other end flybacks down (pulling from GND).  You need a bypass cap to account for the sudden reversal (going from a +40A load, to a -40A return), but the turn-off is as fast as turn-on (faster, actually) because it's applying the full 26V, in reverse -- rather than shorting it out with a clamp diode (at ~1V).  And, all the energy stored in the solenoid is returned to the power supply -- not that that's a problem with a slow cycle like here, but it can be valuable for high speed driver applications (like print heads and injectors).

Tim

True. I was mostly talking about heating up the MOSFET PCB more than necessary. If this board is jammed in a little, sealed enclosure, 1.5W can make the ambient temperature pretty high within the box, since we were talking about thermal runaway and semiconductor stress, etc.

[TimNJ]

I think you want a diode with fast forward recovery time.

Diodes don't have forward recovery time. There's resistance modulation, which will start high but the higher the dI/dt the faster the resistance drops, the forward voltage is self limiting (although it can be still quite high).

Poor terminology on my end, I suppose. I have seen the term "tfr" stated in some datasheets, just not commonly.

After some additional reading, it seems that trr and "tfr" may in fact be in inversely proportional to each other. That is, lower trr may come at the expense of higher "tfr". I need to do some reading on that...and maybe pull out my semiconductor physics textbook for once.

I wonder what metrics we can use to gauge "tfr", when it's not stated.

[Marco]

The problem with using the term forward recovery time is that it gives the impression that it will be like reverse recovery time, just with an open instead of a short.

The forward recovery time is just the time till the peak forward voltage, that time is mostly independent of dI/dt ... but it's also not really valuable to know, because the diode is not an open at all. For a given dI/dt there will be peak forward recovery voltage and that's what you're generally going to be interested in.

[TimNJ]

Just curious, but assuming I go the TVS route, could I put a fairly low value resistor in series with it across the mosfet to try and collapse the field faster?

If you put resistor in series with the TVS, the peak drain-source voltage will be higher, which may defeat the purpose of the TVS a little. On the other hand, if your TVS cannot handle the energy, it may be necessary. Higher series resistance = lower average power dissipated in TVS while clamping = higher peak voltage on the drain. A trade off.

[T3sl4co1l]

A 36V TVS will discharge the inductance faster than a clamp diode.  Mind, I'm recommending using both -- you'll need a HUGE TVS to handle the charge in the solenoid, better to let it discharge through its internal resistance (which takes some seconds!), using the clamp diode.

Tim

[Pizzashape23]

The clamp diode was one of the issues I had in the first place. It just didn't seem to work. How would I go about correctly choosing the type of diode and then choosing the right specs? Clearly my judgments not that good from the two dead mosfets sitting on my desk.

This would then mean I am increasing the gate discharge resistor, placing a zener across 12V and ground, placing the TVS between the Drain and source and using a clamping diode across the solenoid.

[T3sl4co1l]

The clamp diode is fine, as long as it's heatsinked (really, just some thermal mass will do I think) and both sections are in parallel.

The problem is it's not beside the transistor with a capacitor, so you need the TVS to deal with the inbetween.

This would then mean I am increasing the gate discharge resistor, placing a zener across 12V and ground, placing the TVS between the Drain and source and using a clamping diode across the solenoid.

Yes; for "across/between X and Y" meaning, cathode and anode respectively.  They're still regular diodes in the forward direction (anode to cathode), don't want to short out the supplies.

Tim

[Pizzashape23]

Would this be a sufficient TVS Diode?
https://au.mouser.com/ProductDetail/Littelfuse/SMDJ36A?qs=zL8PTTxXkF3y4QOmGeQP1A%3D%3D

[T3sl4co1l]

Yes.  Even an SMAJ would do, but you aren't going to hurt anything with a bigger one.

Tim