MOSFET DRIVER

Some details would help !

Graham

You need real MOSFET driver ICs, of which there are many types. There are also lots of H-bridge driver ICs. You don't mention your voltage requirements, or much else, but this gives me a chance to mention my favorite Intersil HIP4080A and HIP4081A. They're only good to about 80, but they're good performers in the 15 to 80V region: fast, low power, and work to over 1MHz.

On 2 Jul 2006 17:05:00 -0700, Winfield Hill wrote in Msg.

For more voltage (600V), check out International Rectifier's IR2110 and relatives.

robert

which are very intolerant of stray inductance. but do work.

Cheers Terry

The IR2100 and 2200 family of 600V-capable MOSFET drivers are somewhat slower than the 80-volt 4080 family, but the important thing to realize is that most of them are capable of delivering an amp or so of gate current with a few tens of ns risetimes. That's good because that's what high-gate-capacitance MOSFETs and IGBTs need for fast switching, but high dI/dt risetimes also means they're vulnerable to dangerously-high V = L dI/dt voltage spikes developing across any excess MOSFET wiring inductances. It goes with the territory, I'm afraid. I position the drivers right next to the MOSFETs, and maintain a small enclosed-area for the gate-drive and return wiring and have had good results. With a home, pay attention to where the water is going. With MOSFET drivers, pay attention to where the currents are going. E.g., for dI/dt > 2A/10ns, keep the total wiring-loop L under 25nH for Vspike < 5V, otherwise, watch out! 25nH isn't much.

indeed.

my first experience with IRs 1200V half-bridge drivers was a demo board they gave us. which blew up. three times. oddly enough, we didnt use that part. turns out a novice laid out the demo board, and badly. funny really, but we could have bought perhaps 100,000 chips. if they hadnt expired so violently..

the great thing about this gate drive stuff is that by the time you know enough to use these gate drivers, you know enough to roll your own. its hard to beat some of these for size and convenience though.

Cheers Terry

On 4 Jul 2006 19:28:22 -0700, Winfield Hill wrote in Msg.

...or if you don't need ultra-fast switching, add a resistor into the gate lead. 10 Ohms will already work wonders.

robert

Hello Terry,

There appears to be a disturbing trend to delegate app note writing and eval gear to younger folks. And it shows. App notes are often bland and shallow, eval board, well, you said it.

A year after my degree I met a friend from university at an airport. "Hey, where did you end up?" ... "At xyz company but they stuck me in app engineering for the first year, still there".

In the good old days app notes were written by guys like Robert Widlar and Bob Pease. Back then these were even better reading than Alfred Hitchcock. Not anymore.

:)

In this particular case I doubt it would have helped IR if the eval board did work, as I've never had a sensible price from them in NZ. even for non-trivial volumes.

On the whole, I agree with you. I have seen some execrable demo boards. they usually go horribly wrong with things like power supplies, oscillators, ESD measures, terminations, stuff like that.

John Larkin made an interesting point though, about finding gotchas by looking at eval boards & app notes.

Cheers Terry

Hello Terry,

I have seen eval boards from very reputable and big manufacturers where I had the impression that the designers must have thought ESD was some kind of rock band.

That's a good one. But I moved to building protos from scratch now. Bought a batch of extra large Veroboard with ground plane. Saves a lot of time, no need to find eval board bugs first. If I find a bug I know that I dunnit. Except in cases where I found that some folks had added a wee "stretch" on the datasheet.