Author Topic: Switching mosfet in leaded package faster than 1000 A/µs ?  (Read 2264 times)

0 Members and 1 Guest are viewing this topic.

Offline MiyukiTopic starter

  • Frequent Contributor
  • **
  • Posts: 908
  • Country: cz
    • Me on youtube
Switching mosfet in leaded package faster than 1000 A/µs ?
« on: August 23, 2019, 12:33:51 pm »
Hi folks,

I have a question
Is any reliable way of switching mosfet with Vgs rating of +-20V in leaded package (TO-220, 247...) faster than 1000 A/µs

If possible I want to get somewhere about 3-5 000 A/µs and packages with kelvin lead is not available
« Last Edit: August 23, 2019, 12:57:56 pm by Miyuki »
 

Offline SiliconWizard

  • Super Contributor
  • ***
  • Posts: 15797
  • Country: fr
Re: Switching mosfet in leaded package faster than 1000 A/µs ?
« Reply #1 on: August 23, 2019, 02:27:46 pm »
1000A/µs? :o
 

Offline rhodges

  • Frequent Contributor
  • **
  • Posts: 354
  • Country: us
  • Available for embedded projects.
    • My public libraries, code samples, and projects for STM8.
Re: Switching mosfet in leaded package faster than 1000 A/µs ?
« Reply #2 on: August 23, 2019, 04:29:41 pm »
Are you building an atomic bomb?
Currently developing embedded RISC-V. Recently STM32 and STM8. All are excellent choices. Past includes 6809, Z80, 8086, PIC, MIPS, PNX1302, and some 8748 and 6805. Check out my public code on github. https://github.com/unfrozen
 

Offline MiyukiTopic starter

  • Frequent Contributor
  • **
  • Posts: 908
  • Country: cz
    • Me on youtube
Re: Switching mosfet in leaded package faster than 1000 A/µs ?
« Reply #3 on: August 23, 2019, 04:51:49 pm »
No just experimental high current fast synchronous buck
Fsw 250 - 300 kHz
Iout 80-100 A

Today mosfets have so small gate charge and so low Rdson they can easily withstand this kind of current in single chip/device and at moderate gate drive current 3-4A have switch time in tens of ns

But then is here problem of 80A on 80ns  > 1000A/µs
And gate drive itself can go much faster

Yes I know it can be simply avoided by just using bunch of small ones in parallel
But solution with single TO-220 will be nice (it shouldn't have even big power dissipation just lower tens of watts, so thermally easy manageable)
 

Offline duak

  • Super Contributor
  • ***
  • Posts: 1048
  • Country: ca
Re: Switching mosfet in leaded package faster than 1000 A/µs ?
« Reply #4 on: August 23, 2019, 06:21:26 pm »
With a dI/dt of 1 A/ns you ought to consider the loop areas of the switched circuits and their interactions.

According to this loop area inductance calculator https://www.eeweb.com/tools/rectangle-loop-inductance a loop 10 mm * 2.5 mm with 1 mm dia wire (estimate of TO-220 package leads and wirebonds) has an inductance of 6.68 nH.  The counter-EMF generated by this inductance is = L * dI/dt = 6.68 V.  This voltage is in series with whatever is being switched in the source-drain circuit.  The gate-source drive circuit will also have some or all of this voltage in series with it because of mutual inductance.  Connecting to the drain tab should reduce the loop area somewhat but won't be as effective as a kelvin connected gate drive circuit.

If you can model the circuit well enough you might be able to pre-distort the gate drive waveform to compensate for the intrinsic series inductances and deliver a better waveform to the gate. One way is to discharge a capacitor charged with a higher voltage to get things going and count on the various reactances to limit the gate to source voltage to a safe value.

If the circuit voltages are high enough, the counter-EMFs will not play as large a role in increasing current rise & fall times.  What voltages are you planning to work with?
« Last Edit: August 23, 2019, 07:13:54 pm by duak »
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 7043
  • Country: nl
Re: Switching mosfet in leaded package faster than 1000 A/µs ?
« Reply #5 on: August 24, 2019, 07:32:53 am »
I've seen mentions of using an inductive kick to quickly turn a MOSFET on.

I don't know how normal MOSFET gates protected, but at least the automotive one should have gate protection ... so that should give you some safety margin.
 

Online David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: Switching mosfet in leaded package faster than 1000 A/µs ?
« Reply #6 on: August 24, 2019, 05:53:37 pm »
The first thing I would try to push switching speed with standard parts if two packages are acceptable is to use a two transistor cascode switch so the gate of the large high voltage power device is decoupled directly to ground and switching occurs at the source using a much smaller low voltage high current part.

Distorting the gate drive waveform would be fun.  It might be a good application for pulse shaping using a lumped element transmission line.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Switching mosfet in leaded package faster than 1000 A/µs ?
« Reply #7 on: August 24, 2019, 08:49:04 pm »
Right, 1A/ns * 7nH = 7V, so good luck with that.

Such is the problem with leaded parts.

The higher drive voltage of SiC helps reduce the impact of lead inductance.

SMT components perform better (D2PAK has about half the stray inductance of TO-220), but leadless components are required for absolute maximum performance.  This is why GaN devices are available almost exclusively in CSP, BGA and DFN packages.

A perspective you should consider: switching impedance.  If you're switching 320VDC at 80A peak, that's an impedance of 330/80 = 4 ohms.  That's quite low.  Most structures (device pins, PCB traces) are in the 50-100 ohms range.

4 ohms at 80ns (one full cycle), is 3.2nF and 50nH.  We really only need 1/4 of a cycle, allowing up to 12.8nF and 200nH!  So this doesn't sound too terrible, if you can just find devices with Kelvin gate terminals to overcome the source inductance problem.  There are some D2PAK, TO-247 and 264 style devices that offer this, and also SOT-227 modules if you don't mind the cost.

Otherwise, consider increasing the inverter impedance by wiring multiple channels in parallel.  This raises the switching impedance of a given channel, putting it closer to the characteristic (transmission line) impedance of the connections, minimizing their inductive or capacitive impact.  You still want a lower impedance (i.e., less L_s, more C_oss) by some factor, to keep the peak voltage down (I_pk * Zsw = Vpk).  So, say, Z_inv ~ 100 ohms and Z_sw ~ 25 ohms.

You can intentionally add capacitors to increase C_oss and decrease Z_sw in that way, but it's probably preferable to trap that energy rather than dissipate it, i.e., use a dV/dt or peak clamp snubber, the energy of which can be recycled immediately (quasi-resonant snubber) or dumped into a common rail and another converter used to put it back into the system (heh, the electronic equivalent of a sewage pump I suppose?).

Also, when you have multiple channels, you can consider driving them independently, yielding perks like phase interleave which reduces input and output ripple current.

Note that a phase-interleave converter (buck/boost/whatever) must have independent current control, otherwise one channel can pull slightly more PWM% (due to timing differences between channels) and blow itself up.  Approach it as a series of transconductance amplifiers, outputs wired in parallel.  Command their output currents based on a common current setpoint, from a single voltage error amp.  This architecture is fully scalable, and gives full voltage and current control, making it particularly suitable for, say, bench supply use.

If you're doing something with a single huge AC output, like an induction heater or RF amplifier, you can't really afford phase interleave, but you should still have 0° power combiners between inverters, to account for slight differences in propagation delay.  (Or for the RF case, other angle combiners are reasonable, allowing you to cancel out some distortion; but, this doesn't really mean anything in a class-D context that's already "100%" distortion.)

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 7043
  • Country: nl
Re: Switching mosfet in leaded package faster than 1000 A/µs ?
« Reply #8 on: August 24, 2019, 09:55:39 pm »
The very simple method from "A new gate drive technique for superjunction MOSFETs to compensate the effects of common source inductance" seems to work well for them (ie. using a very low Rg and adding gate inductance). Might be worth a try given the simplicity.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Switching mosfet in leaded package faster than 1000 A/µs ?
« Reply #9 on: August 25, 2019, 12:03:06 am »
I wonder a bit about adding a small source inductor (on board), and coupling it to ~2 turns in series with the gate.  It's positive feedback, but only enough to compensate for the uncontrolled lead inductance.

Downside is it adds more series inductance in the power path.  Which if that's tolerable, then that's cool.

The transformer would be air cored, and could be planar (made of PCB traces), or, basically jumper wires formed properly.

Example:



Load current comes in from the capacitors on wires (top/left) and flows through the metal-ribbon shunt resistor.  Which obviously has significant inductance due to its shape (~10nH?).  The blue wire is the Kelvin sense wire going to the oscilloscope; it makes two turns beside the shunt, opposing its inductance and correcting the waveform.  Two turns are necessary because the coupling factor is less than 1; spacing is adjusted to get the waveform correct.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline duak

  • Super Contributor
  • ***
  • Posts: 1048
  • Country: ca
Re: Switching mosfet in leaded package faster than 1000 A/µs ?
« Reply #10 on: August 25, 2019, 08:03:48 pm »
Marco, I couldn't access the link you were probably referring to.  Is it the IEEE paper?

I did find this link to an app note on driving MOSFETs showing the effects of the intrinsic and parasitic reactances in a run of the mill SMPS supply using leaded parts: https://www.onsemi.com/pub/Collateral/AN-9005.pdf.pdf

I haven't designed with power MOSFETs since the 90's so superjunction and SiC parts are all new to me.  More stuff to not have time to catch up on.
 

Offline MiyukiTopic starter

  • Frequent Contributor
  • **
  • Posts: 908
  • Country: cz
    • Me on youtube
Re: Switching mosfet in leaded package faster than 1000 A/µs ?
« Reply #11 on: August 26, 2019, 02:55:10 pm »
SiC are another beast with their low threshold and sometimes insane internal Gate Resistance
How do you want to drive a gate with RG(int) 26 Ω  :wtf: at any reasonable speed

Also choice of devices with kelvin lead is somewhat limited and sometimes way more expensive than alternative without it
And also just 600+V devices are available with kelvin lead, lower voltage ones have no one and SMT packages mostly have complicated cooling

Something as inductive coupled gate kick would be nice and can work mostly passive as it is self regulating by load current
Sadly I also have no access to IEEE papers, do they have any example solutions or is it just theoretical study ?
 

Offline filssavi

  • Frequent Contributor
  • **
  • Posts: 433
Re: Switching mosfet in leaded package faster than 1000 A/µs ?
« Reply #12 on: August 26, 2019, 03:44:05 pm »
For general experimentation you can try using inductive kick or some sort of resonant network, however if you are serious about it (as in you want to integrate it in a product at some point in the future) I would avoid such hacks as any number of problems could lead to anything  from loud bangs with plastic shrapnels to fireballs, as to push such currents for any length of time will require a very low impedance source.

Also take into account that as the mosfet get faster the gate gets weaker and weaker (especially for state of the art GaN devices) and your driving must be on point as even a mild over voltage leads to sudden death

I say this not to scare you, you must have a realistic expectation that doing that with to220 will be very hard
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 7043
  • Country: nl
Re: Switching mosfet in leaded package faster than 1000 A/µs ?
« Reply #13 on: August 26, 2019, 03:53:36 pm »
Sadly I also have no access to IEEE papers

If you don't have strong moral convictions on the sanctity of copyrights most of their papers are available from our friendly neighbourhood Kazakhstani ...
 

Offline langwadt

  • Super Contributor
  • ***
  • Posts: 4857
  • Country: dk
Re: Switching mosfet in leaded package faster than 1000 A/µs ?
« Reply #14 on: August 26, 2019, 04:19:57 pm »
SMT components perform better (D2PAK has about half the stray inductance of TO-220)


how is that, just from mounting alone? isn't D2PAK just a TO-220 with a leg and tap chopped off?
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Switching mosfet in leaded package faster than 1000 A/µs ?
« Reply #15 on: August 26, 2019, 10:51:44 pm »
SMT components perform better (D2PAK has about half the stray inductance of TO-220)


how is that, just from mounting alone? isn't D2PAK just a TO-220 with a leg and tap chopped off?

Precisely. Inductance is length.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline langwadt

  • Super Contributor
  • ***
  • Posts: 4857
  • Country: dk
Re: Switching mosfet in leaded package faster than 1000 A/µs ?
« Reply #16 on: August 26, 2019, 11:14:59 pm »
SMT components perform better (D2PAK has about half the stray inductance of TO-220)


how is that, just from mounting alone? isn't D2PAK just a TO-220 with a leg and tap chopped off?

Precisely. Inductance is length.

Tim

but you don't have to have a to-220 flop around in the breeze, you can mount is pretty much like a D2PAK

 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Switching mosfet in leaded package faster than 1000 A/µs ?
« Reply #17 on: August 27, 2019, 12:37:55 am »
You can, but when do you get the chance to do that?...



This by the way is an ST part, nickel plated tab, so it tinned very poorly. :(

That's one obstacle already; not hard to find tin-plated tabs, but then you need to make sure they're always purchased as such, and find another in case the manufacturer changes it.

But now you can't heatsink it!  At least, not as easily.

You're not required to need some "flopping in the breeze" for heatsink mounting, but it is more likely.  A lot of vertical heatsinks are designed to hold the body at full lead length, these for example: https://www.digikey.com/product-detail/en/ohmite/C220-075-3VE/C220-075-3VE-ND/2764652
Custom ordered, or done by hand, you can specify exactly the right board-to-mounting-hole distance, but your off-the-shelf selection is more limited.

Or you might mount the devices underneath the board (legs poking up), in which case lead length is dictated by lead form and board standoff height.  Which actually, if all the devices are the same thickness, you can even use the board itself as the tension clamp, and avoid needing mounting screws at all (and the ills that come with them: uneven clamping force).  Or if the board itself is dissipating power (say, snubber resistors, filter chokes/caps), just keep going and make a clamped sandwich of two heatsinks, a big soft thermal pad, PCB, transistors and a thin thermal pad.  (I want to do that some day, I just haven't had an application for it yet. :) )

Or you could mount the board parallel with the leads, and hook the leads into holes right near the board edge.  Or make full-body cutouts.  Or surface mount them -- I've actually seen that in a car amp, bodies clamped to heatsink, leads surface mounted, just sticking straight out over the board, lap soldered.  Not that it needed it, but apparently it just worked out that way.  That would be really good.

So yeah, absolutely; but if you need closer to an order-of-magnitude reduction, you're still looking at many devices in parallel, or flatter devices.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf