on/off times for power GaN MOSFET not in spec sheet!

[dbctronic]

I'm designing a multi-kHz solenoid driver amp (class D). I found a transistor (EPC2012C) that looks OK - except the datasheet does not list on/off times!!
Are they expecting me to diddle around with device capacitances and derive these times somehow? Or just what? I'm not finding thousands of suitable devices on DigiKey, just a handful, so I may not be able to blow this device and its complement off so easily.
I need to switch 200V at up to 5A, probably about 5 kHz tops. It will just be putting out fixed sine waves, and a few percent distortion is probably fine, so I don't need insanely short on/off times- 10 to 20 nS seems to be available in similar devices and would probably do fine.
This hobby o' mine... so much to learn...every freakin' time I turn around.

[uer166]

Is that a... serious question? If yes, on/off times are defined by much, much more than the FET alone.
"I need to switch 200V at up to 5A, probably about 5 kHz tops."-> then why are you using GaN?

[wraper]

I need to switch 200V at up to 5A, probably about 5 kHz tops

Finds some bare die part intended for switching at multiple MHz  .

I'm not finding thousands of suitable devices on DigiKey

According to your description of application, just about any MOSFET with suitable voltage/current rating should work.

[Hiemal]

Yeah, echoing what everyone else said; for what you're doing, genuinely pretty much any mosfet or transistor would work. Turn off and on times at such low switching speeds really don't matter, and using an exotic FET is so far beyond overkill I don't really even have words for it.

[dbctronic]

No, the capacitance question was facetious.

Maybe my math is off. Let's see, at 5 kHz, one half cycle is 100 usec. Switching that for class D amplification at a chunky 30 chops per half cycle is 3 usec per chop. With many MOSFETs showing on/off times around 30-40 nsec, the transistors spend 30 nsec turning on and 30 turning off, so they're in linear mode 60 nsec every 3 usec, or 2% duty cycle. 200V at 5A is 1 kW, so a 2% duty cycle implies around 20W loss. A higher sample rate for cleaner waveform would raise that figure. Other GaN devices I looked at have times around 10 nsec, but some of them are looking scarily like they're going off market or only available on a nonstock basis.

The solenoid I'm driving probably won't draw much more than 20W. It just didn't sound right to me to make a class D amp with less than 50% efficiency, and I didn't know what MOSFETs are in common use for class D. This particular transistor said it was well suited for class D amp use. And it only costs a couple bucks, which doesn't seem too 'exotic' to me. I'll gladly spend that for a one off project rather than spend 60 cents on a plain old job and wonder how it will work.

Is my math off? And, back to my original facetious question, why would a transistor datasheet, especially for a power transistor, leave those times out? If any of you professional engineers had your boss come to you and say 'use this one--make it work', what would you do? Buy a few and measure?

[Miyuki]

1) Your math is way off

2) with Qg = 1nC and internal gate resistance of 0.6Ohm is the only limiting factor how sharp is your gate drive signal, the transistor will follow it perfectly

[T3sl4co1l]

Where are you getting 200V 5A from, if you're only drawing "20W"?

You don't want GaN transistors.  The problem is not about efficiency, it's speed.  Single nanoseconds are typical.  Speed kills: without very careful layout, they'll probably just blow up.  You need specialized drivers, or serious mitigation techniques, to interface with that kind of speed.  I suppose you could get away with picking up a dev board, or one of those integrated inverter ICs (the kind with transistors, drivers and bypass integrated); but you still have very little leeway to develop with it, as a single microsecond of short circuit and *poof* they're blown up.  They're fast to switch, and fast to blow.

Tim

[Alex Eisenhut]

Reading OP's post history, I think he wants to build some sort of very fast hammer. so instead of fighting physics, work *with it*!

https://www.eevblog.com/forum/chat/electromagnetic-hammer/msg3097350/#msg3097350

[dbctronic]

OK, I get it. Looks like I'll go with standard MOSFETs. Many thanks for the wall of text. I will study it closely.

As for the 20W draw from the solenoid, remember, a 200V 5A source provides 1kW of apparent power. Actual draw is the resistive, reactive, and core loss components. The resistive will be quite small, as this solenoid can only be at most 3 mH to keep required drive amplitude down to 200V. Core loss may be five or ten watts. Reactive will be mostly what's delivered to the sample. I'm guesstimating a few tens of watts.

That's why I want a class D amp (and of course SMPS) in the first place!!! The solenoid impedance will be too high for commercial audio power amps, hence it will require too high a drive voltage or I'd have gladly bought a used one.

What I'm making is a device to apply an oscillating pressure to rocks, mostly concretions with fossils inside, to crack them open. Hammering destroys most, so in my device the clamps will hold the sample with a prebias pressure that the solenoid augments at the sample's resonant frequency. The clamps don't hit, they squeeze. I'm hoping for an amplitude of several pounds.

Thanks again, all. This project is mostly for the learning experiences, which have been many, especially in the magnetics department. If it never actually works I won't be heartbroken. But I will try hard.

[Alex Eisenhut]

In that case you want a vise. No need for such complexity.

[dbctronic]

I have two vises. Can't say I see how to make them apply an oscillating pressure at several kHz, though. Please enlighten.
Simply tightening a vise down on a concretion will crush it at the contact points until it's crumbs or it decides to split any old way it wants, likely not in the fossil specimen plane as desired. That's assuming the vise can develop enough pressure to finish the job. Takes a big vise, they're tough buggers!

[T3sl4co1l]

What you're after, is probably not a resonant frequency.  Resonance means stress/strain is peak at periodic intervals through the material.  Perhaps a plane wave can be applied, at the right frequency, with the right mounting (because boundary conditions matter for reflection and absorption), and perhaps the material is consistent enough (similar density, modulus, speed of sound) that, you can manage to get a line or plane of maximum strain right across a bedding plane, and *pop* off it goes.

But that's a looong list of assumptions, for a material that is explicitly inhomogeneous -- not to mention, once a crack forms, the properties of the material change (as the crack region turns from solid to broken, suddenly it's a lot more flexible), and most likely more power and a lower frequency are needed to continue spreading it.

This sounds more like something that needs piezo stacks to do?  The displacement is small, probably tens or hundreds of microns, and the force (static and dynamic) need to be quite large.  A magnetic transducer can't really do either one; imagine clamping down on a loudspeaker cone for example.  The frequency range is also much higher -- when you hammer a rock, that "tink" sound is its (and the hammer's) resonant frequencies, mostly low order flexural and longitudinal modes.  This is largely in the kHz, and the stiffness and density (at least of a solid homogeneous sample) are well matched to piezo transducers (being ceramics themselves).

I don't know how you'd mount it.  Fluid, maybe?  You'd have to use liquid metal, if you want density comparable to the material.  Maybe some mismatch would actually be desirable, to get some system Q in the resonances?  Then you could maybe get in-plane resonances of low order and hope to crack it in layers of half, thirds, etc.  Solidified metal might work too, as long as it fits tightly; several low-melting alloys expand on cooling, which might serve to clamp it in place, I don't know.  Or maybe it needs to be epoxied in place, so there's some holding force.  Or perhaps the samples are usually oversized, and can be polished down to flat planes, then clamped between metal anvils and vibrated.

Yeah, as mentioned, clamping or vibration on a concretion will simply crumble from the contact points (or worse) -- but this is still true even of very hard rocks, the contact points are stress raisers and will crumble under sufficient load.  We're talking enough load to try and cause internal rupture, and it's a good bet that at least as much force will be required on the rock surface to get there.  You need a wide contact area to have any hope of distributing that load.

Tim

[dbctronic]

T3sl4col1l:

Thanks much for the thoughtful reply. There are some valuable ideas in there. Most siderite is fairly homogeneous in bulk, tending to fracture in a characteristic pattern somewhat like glass when hammered rather than in any obvious pattern of local weak spots, so a variety of internal propagating or standing waves are probably possible.

The main inhomogeniety in a concretion is the presence of a thin film of degraded biomaterial in the central plane of the concretion - that's the fossil. When concretions sit out in the open, moisture gets in and freezes and thaws, which is the main way to open them up - leave them outside (for years) in buckets. It does a pretty good job of exploiting the specimen plane, just takes forever. I figured it might make a resonance forming plane. As the concretion splits, my device would constantly seek the new resonance frequency via a transducer opposite the driver, forming a closed loop oscillator with the concretion as the feedback element.

As for piezos, I very much worry that they would vibrate themselves to pieces while trying to destroy the siderite--it's tough, as anyone who tries to open concretions with a hammer can testify. A specially designed solenoid with a gap of a few mils will be able to generate multipound forces in the kHz range. I'm hoping that operating at the concretion's resonant frequency will be able to focus stress on the specimen plane, exploiting its weak points piecemeal. The resulting reductions in overall resonant frequency may mean that continuing to seek resonance will continue to focus stress on the remaining weak points in the specimen plane. Might work, might not.

Everyone else:

Looks like I've finally reached the point in my electronics design projects where I need a circuit sim. Just downloaded LTspice and have started diddling with it. Looks like fun! I won't grumble about the UI or documentation--it's freeware. But I'm still not exactly sure how to 'sprite' a component, or if that's the same as 'selecting' it. New term on me, not defined in docs as far as I can tell.

[Alex Eisenhut]

Fair enough, from what I thought I knew about materials science, a vise would work with a homogeneous substance.

This isn't it. 

How about magnetostriction? Can that be used to generate enough force?

[dbctronic]

Large transformers carrying saturation level currents seem to generate only a gentle hum. I don't think magnetostriction moves over a great enough distance to steal much power from the core and deliver it to an external load. And I've never heard of it being used for anything, only mentioned as an interesting physical phenomenon, not even as a significant inductor loss mechanism. My proposed design of a solenoid with a gap of a few mils, on the other hand, will generate multipound forces and be able to couple considerable power to an external load.
And from what I have read, it seems that materials with the highest levels of magnetostriction (read: hard to acquire specialty materials) don't display much greater effect than plain old transformer steel. I had thought about it several times, but couldn't see a design that would get it to perform well. And the fact that the phenomenon is one of those that only seems to exist over a very narrow dynamic range is discouraging.
I press on. LTSpice is definitely helping with the driver circuitry design!! Learning a lot about transistor switching behavior!!