Weird ways to use MOSFETs in series: good or bad ideas?

[PointyOintment]

I'm designing an LED stroboscope, and I have some dual MOSFETs from hard drives (where they drove the platter motors) that I'd like to use for switching the LED rapidly. Because there are two MOSFETs in each package, and just one is sufficient for the current and voltage, I thought I might as well come up with a way to use both. In most of the ones I have, the two MOSFETs are unconnected to each other internally, so I can connect them in any way or use them for totally separate purposes. I came up with a couple of ideas that I haven't seen used before, so I'd like some input on how well they might work.

Circuit 1 is just the single-MOSFET base circuit, where there's one MOSFET doing regular hard switching on the low side of the LED, driven by a gate driver. This is presented only for context for the other circuits.

Circuit 2 has two MOSFETs in series. One of them still does hard switching, while the other operates in its linear region, acting as a variable resistor to regulate current through the LED. The current is measured across a shunt resistor (which, despite being in the same place, has a lower value than the resistor in circuit 1, because it's no longer ballasting the LED on its own). I have a few questions about this circuit. I put the current source MOSFET lower, against ground, so that it would have a constant ground reference even when the hard switching one is turned off, because I thought that might help avoid oscillation. Is that reasoning sound and well-founded? Would having them swapped be better for some reason? Should one of them even be on the high side of the LED in this circuit? (It's a 12 V LED, so that'll need a high-side driver or me to pick one of the devices with a P-channel MOSFET.)

Circuit 3 also has two MOSFETs in series, but this time, they're both doing hard switching. One of them turns on to provide the rising edge of the current pulse, and the other one turns off to provide the falling edge. This is intended to achieve shorter pulse lengths than one MOSFET could (because of the time it takes to move the gate charge). Is this a good idea? Which way around should the MOSFETs go? I tried to optimize their arrangement for stability when one is off, but I don't have much MOSFET design experience, so I'm not sure this is actually the best arrangement.

Circuit 2 is the one I'm more interested in using in this project.

[Seekonk]

You might be over thinking this.  I strobe with a 10W 9V LED and am pumping 36V into it and will likely go much higher without a resistor.  Only a NANO driving a FET directly.

[PointyOintment]

I'm absolutely overthinking it. These questions are way more for the sake of curiosity than building a better stroboscope.

[David Hess]

The cascode arrangement evb149 describes can be used with the current limiting resistor between the MOSFETs.  I have used this very configuration to drive LEDs with an NPN transistor for the cascode transistor with its base tied to the +5 or +3.3 volt supply and a logic gate pulling its emitter down to ground through the current limiting resistor.  Operation down to 10s of nanoseconds and faster is feasible like this.

The switched current sink shown in the second schematic is used in some very fast circuits but not quite in that form.  The sense resistor is in the source line of the current sink and there is another diode or transistor used to keep the current sink from saturating so it is always operating.  With some diode isolation on the output, rise and fall times below 1ns are feasible.

[PointyOintment]

Thanks for all of the input.

I had read about the cascode concept before, and recognized the resemblance, but I had forgotten how it works. That was a good explanation, evb149. I had only known it was good for switching higher voltages. That's interesting that it can achieve higher speeds too. The links, papers and related concepts look interesting too.

Regarding circuit 3, I had compared it in my mind to how some robots use opposed motors for fast twitch motion capability, but I guess the math doesn't work quite the same way and the advantage doesn't transfer to this problem.

Another thing I had found out, but forgot to mention earlier, is that current sense MOSFETs exist. They effectively have two MOSFETs in one package, but sharing the gate (and source, I guess). That way they are always on to the same amount, and a sensing current can be run through one (which is smaller) while the other (which is bigger) is used for the current it's actually being used to control. I don't have any of those, though, so it's a moot point.

Anyway, I'll definitely start with something like circuit 1 for my stroboscope, and only consider anything more complicated if it is unsatisfactory.

[T3sl4co1l]

Can you give us a rough idea of how fast you think this needs to be?

Tim

[PointyOintment]

For what I currently need the stroboscope for (a household fan), <70 Hz, <10% duty cycle.

But I intend it to be useful as a general purpose stroboscope, for people other than me to use for projects I'm unaware of (hackerspace), so if it ends up not wildly exceeding my above specs on the first try (which it almost certainly will), I'll probably look into improving it someday.