How to quickly turn on/off N-FET switching high side of 55V?

Or overcomplicated ones.

John

You can use the topology of a voltage-doubling charge pump, only instead of doubling a single input voltage, you take two input voltages and add them. In the following circuit Vout = HV + 12, minus two diode drops.

12v HV | | | | | 1N5818 V ,----------, - |555 osc. | | 1N5818 | 200 kHz | | | pin3|---||------+--->|---+-----Vout '----------' | 0.1uF = 10uF | gnd 1N5818 schottkys for speed and low voltage drop.

If you have to work with 5 volts, double it and add it to the high voltage source. Here, Vout = HV + 5x2 = HV + 10 minus three diode drops.

5v HV | | | | | 1N5818 V ,----------, - |555 osc. | | 1N5818 MUR160 | 200 kHz | | | pin3|--+---||------+--->|---+------>|---+-----+----Vout '----------' | | | | | 0.1uF = 10uF | = 10uF | | | | | gnd | gnd | 0.1uF | '-----||-------------------------'

If that's not enough you can ad another stage to the cascade, tripling the 5 volt source and giving you Vout = HV + 15.

SNIP

For a dollar apiece, no problem.

SNIP

I have some high side switching to do too. I'm going to skim through some of the 100 postings in this thread for some gleanings...

Aha, looks like Fred Bloggs beat me to it. He calls the voltage adding circuit SCVM. Looking at his diagram, I also see I got it wrong in using the HV as one of the inputs to the SCVM voltage adder. You should connect the adder's input to the source of the mosfet, as he has it, and not the drain/HV as in my diagram.

Just because it's not an NIF type of end use does not mean we trivialize and get sloppy with the electronics.

Here's John's Gate_Driver.jpg sketch as an ASCII drawing.

Silly Gate Driver JL Dec '07 V+ | D1 R1 K1 | +12 --|>|--/\\/\\/--+---------o R2 |--' ,--->| ,o---/\\/\\---|| / _|_ C1 ,---o' |-->

opt + ---' --- | : | PV - ---, | | : | \\ | | : | '--->'-----+---------------------+ : | ---uuuuu--- | LOAD

Well, John's idea is certainly more simple than Fred's.

The idea of using a relay to control a 30A MOSFET means the relay can be small; a 30A relay would be a beast.

My favorite small relay type is Panasonic's TN series (formerly under Mitsubishi's NAiS Aromat brand name),

These measure a tiny 0.22 x 0.55" and are 0.39" high. I often use the TN2-4.5V part, rated for use at 4.5V, but turns on at 3.38V max. Its 145-ohm coil takes 31mA at 4.5V, so it's easy to drive.

True.

Well- okay, this does simplify things for the OP and it does seem relatively bullet proof as far as protecting his low voltage stuff. But I still think he should fuse each battery. The automotive aftermarket is supplying 42V blade type fuses these days. The Littlefuse MAXI 42V line, ( and maybe Bussman has similar), is a 58VDC fuse with

1000A interrupt capability and available in high current. The OP could use a 30 or 40A job. These things are rugged.

Or 50A. I hadn't seen your fuse suggestion before, but I certainly agree, with these serious batteries, that's a mandatory addition.

We use these a lot,

fairly similar. The latching versions are handy, and have zero thermal emf's. They switch in under a millisecond, and we've had zero failures so far.

I still like relays.

John

The dc/dc converter and relay isn't a bit sloppy. It's ideal for someone who doesn't want to do a lot of design and analysis and electronic assembly. If I had this switching problem myself, that's the way I'd do it.

Playing with various driver circuit ideas is fun, but the OP is probably better off with a simple, bulletproof circuit he can wire up in a few minutes from a few common parts.

John

Know of anything small like that capable of switching ~750 mA inductive at ~18V 60 Hz? Using a G2R-14 at present.

Ed

The Fujitsu should be OK for infrequent switching. But if it's going to switch a lot, and you need small, an SSR would probably be more reliable.

John

--
Bounce on make won\'t matter, but bounce on break very well might.

--
The OP stated that the device would be used in a high-G environment,
so without knowing what his environment looks like and what the
relay can take, touting its use as a panacea _is_ sloppy.

"touting its use as a panacea"?

What, are you studying journalism in night school?

John

How so? A mosfet gate will stay charged for hours or days. If the contact bounces on break, nothing happens to the gate voltage until the wiper hits the opposite contact.

The contact bounce will *not* be visible on the gate voltage.

John

--- Don't know much about contact bounce, do you?

It could easily look something like this on ON make and OFF make (when I said "break" I meant "OFF make") View in Courier _____ . . .___ ___/ \\ / \\_ __/ \\ ___/ \\____

Looking at your silly gate driver, since there's no way to determine what the hash on the contact will look like when the common contact bounces against the +12V contact, there's no way to tell how long the contacts will stay mated until the first, (and subsequent) bounces. Ergo, since the gate capacitance has to charge through the series resistor the RC to Vth may be longer than the first make time.

The same will be true when the relay common contact bounces against the ground contact.

Also, depedending on how long it takes the MOSFET to transition from full OFF to full ON, or full ON to full OFF, the power dissipated in the channel may be more than trivial.

-- JF

--
And that has _what_ to do with the subject, Mr.
let\'s-avoid-the-subject-by-changing-it?

I know that machanical contacts can generate clean sub-nanosecond edges, at tens to thousands of volts. And I've recently measured the bounce behavior of the cute little Fujitsu relays.

Consider that first transition:

When the wiper lifts off the "ground" side, the gate voltage is frozen low. If the wiper bounces off the low side a few times before it finally lets go, the gate voltage stays put.

While the wiper is "in the air" between contacts, the gate stays low.

When the wiper finally hits the +12 (or whatever) contact, the initial connection, before the first bounce, will last many microseconds, plenty of time to charge the gate to +12. Subsequent bounces, which might last a few hundred microseconds in a small relay, don't matter, because the gate is already charged.

Try it.

Well, make the gate resistor the right value. That's what engineering is about.

That's true for any gate drive. Fact is, in the 10's of volts and up range, the rising edge of a mechanical contact closure rivals the speed of the fastest semiconductors; subnanosecond edges are easy.

Try it.

John

OK, can you cite me touting any panaceas lately?

John