ping Win LND150 avalanche

In general MOSFETs are perfectly happy in avalanche; in fact most power parts have avalanche ratings. You just have to stay within the power dissipation limit. (If you analyze a datasheet "avalanche rating" you'll see it's simply derived from junction-temp limit and thermal mass from the transient thermal impedance plots, using the inductor value, voltage and currents in question, so there's nothing special about avalanche.)

That part has a sensible power rating. If you want it's easy to use MOSFETs in series, while avoiding avalanche, with a stack of resistors. See AoE III, Figure 9.110 on page 696. The inset shows how you use one LKND150 to set the current, with additional series MOSFETs, even ordinary enhancement-mode types. The resistor stack divides the voltage for the gates. You'll need five resistors, if you limit them to 300V.

You could run the LND150 at say 300V, the 2SK4117 can have the remaining 1100V, a relaxed value for both of them.

OK, thanks. I guess I'll use three LNDs.

If I let two of them avalanche, they will hog the power dissipation, but that's OK.

Again, the problem with using an enhancement fet is that it needs gate voltage from somewhere guaranteed to be available always. Deriving that from +1400 volts is a nuisance. A battery would be silly.

Here's my HV regulator. It has almost zero quiescent dissipation and fast programmable rise/fall times, about 1KV/ms up or down.

Why are there two opto-couplers in parallel? What's the output current spec? Lots of load capacitance?

Gawd-awful schematic color choices :-( ...Jim Thompson

It's a dual coupler, used elsewhere on the board, so why not? If the CTR is too high, I can always change the 100 ohm drive resistor. But I want a bunch of drive current.

The load is 22 nF, which I hope is enough C to bypass the output pulser stage. The customer wants to be able to change the voltage fast, up or down. The average DC load current is tiny.

My layout guy likes those colors. I don't.

I was testing some opto-fets last week for this - there was nothing in the datasheet. But an application note says they must not be subjected to transient voltages outside their rating.

Customers might connect them to a wire with inductance, or a relay coil perhaps.

I started with 100uH but they seemed to be quite happy switching a couple of Henrys at 0.5A all day. (It was supposed to be a test to destruction but I felt sorry for them so stopped there!).

I will be using a TVS anyway now but was curious how much there is to worry about, really.

Then there is ESD.

JPG

pg

Hi John,

How could you master the current through the opto ACPL's when the TLP191 dr ives the MOSFET VGS ? How could master the power dissipation on MOSFET and opto ACPL ?

Don't understand how it works ... may be some math would be more explicit ?

Habib.

The TLP191B acts like a battery. It has a stack of diodes and runs continuously, providing about 7V (at 10uA) of gate bias.

The optocouplers drive the FET source, in common-gate fashion.

I know how the TLP191 works, thanks.

My questions was in relation with the current through the MOSFET and power dissipations (hey 1400 V !) ; TLP191 gives 7V --> the MOSFET flows current

--> Vgs becomes smaller when current is increasing through 100 Ohm resistor --> then vgs becomes smaller ... There it may have needs some math or Spic e simulation isn't it ?

Habib.

The LND150s have gate protection zeners, and body diodes, and can presumably handle some avalanche energy, so they should be pretty tough.

But I may as well use a resistor stack into the gates - one more part

- so I don't have to argue the situation in the design review.

This is likely intentional, because it effectively creates a current limit, which is desirable. Of course the FETs have to be able to handle this. Ultimately the 1.5kV supply will sag and shut down. The current limit would act until the HV byass capacitors are drained.

You could go for a more robust depletion-mode MOSFET, see Table 3.6 in AoE III, page 210, for possibilities.

I Spice-modeled the TLP191B... it's on the Device Models & Subcircuits page of my website.

The schematic pin-out doesn't match the datasheet I have. ...Jim Thompson

We have the LND150s in stock, and they behave very well. We also use DN2530, sweet little SOT89, but it's only 300 volts.

I checked out AOE3 as you suggested, and then put the book down on my desk, opened to page 210. Now the book is making very strange crackling noises.

Of course as you say "1.5kV supply will sag and shut down". E(capacitors) = 1/2 * C * V *V = 0.5J P max admissible for th MOSFET = 80W

Let's say the discharge time is about Td = 10ms

E = P.t then P = 50W --> Ok for the MOSFET

But if Td = 1ms the P = 500W --> The MOSFET should blow

May be i missed something ...

Habib.

If there's no photo current in either input loop, the 2SK4177's are OFF... though I'm not sure how well-behaved they are with an essentially floating source.

Needs some kind of shut-down means to ensure no photo-current. ...Jim Thompson

We number the four pins 1-2-3-4. Toshiba uses 1-3-4-6.

The 100 ohm resistors and the opto CTR limit the mosfet currents. That's not so much to protect the fets as to limit power supply dip.

One 2SK4177 can dissipate about a kilowatt for one millisecond, and around 3KW for 100 us. It could easily discharge all the energy that I have available.