gate driver

Hi,

I am guessing the erratic behaviour you are noticing is related to the boost cap charge state. That IC needs to have PWM to charge up the boost cap for feeding the high side fet gate, if you leave the top fet on too long with a gate draining resistor that might discharge the cap before you turn the top fet off.

I did a project with a related halfbridge fet driver IR2184(4) and noticed I had to use 99% max PWM otherwise 100% PWM wouldn't work for this bootstrap capacitor reason.

cheers, Jamie

HIP4081 perhaps? It's a full-bridge driver, but has independent control of both half-bridges. It'll make your high-side bias, control dead-time and prevent shoot-through.

Nice part, but it needs 8 volts, and I only have 5.

Thanks

From the data sheet...

GND (Pin 4, Exposed Pad Pin 9): Chip Ground. The exposed pad must be soldered to the PCB ground for electrical contact and for rated thermal performance.

Maybe devise a spring clip to contact the back-side pad in your breadboard? ...Jim Thompson

I usually put a blob of solder on the pad, and another one on the board. Then I melt the board side with the iron lying almost flat, set the chip on top of the iron till both sides are melted, and pull the iron out. Usually works pretty well--it's how I heat-sink LM78xx regulators in breadboards. Not exactly a production-worthy technique, but I've never had one die yet.

Cheers

Phil Hobbs

With only 5 volts in, can't you roll your own? Small boost or charge capacitor supply to make a 15V rail and a capacitor coupled gate drive. (You need to be sure you don't have nasty transients on the 5V input

Regards

Klaus

Been trying to find a picture, without success, of an over-top-dead-center clamp that I used in my Motorola days to press a flat pack down on a PCB to test it without soldering. ...Jim Thompson

Are the FETs extra large? This chip performance peaks at 3nF, and that's with everything laid out perfectly with low to nonexistent series inductance. The back GND pad floating could easily interfere with switching performance at its advertized speed.

Take a look at Carr-Lane website, toggle clamps.

-bill

There you go...

We had a "well" that the package body fell into to align the leads easily. ...Jim Thompson

Like this, but with a N channel in the high side, and no diodes (this was for a 2 switch forward converter)

Regards

Klaus

Too late!

It's on the SO adapter, upper left. All wired in.

It seems to be working at 5 volts, and I got some SOT223 nfets that are happy with that much gate drive. It rings a bit on the edges, but a proper PCB layout may fix that.

The driver's not getting hot at 1.5 MHz, so I'm OK on the breadboard.

I need some SO adapters with the power paddle connection, vias to the bottom maybe.

Grounding the power pad can only help. I've got it working OK now, so it should be better on a proper PC board.

I'm switching at 1.5 MHz, 4 or 5 amps, so things are inherently radical.

The LTC4442 is cute in that the input pin has different thresholds for turning on the upper and lower fets. So if you drive it with an RC in series with the input, you can control the both-fets-off deadtime. Combine that with gate resistors, and you get some knobs to turn to trade off noise/efficiency/whatever, to get out of trouble maybe.

FDT439N mosfets seem about right. 1 volt threshold and then 17 (17!) S transconductance.

You call me a "mechanic", but this stuff is fun. Some people, architects and civil engineers and even some artists, never get really involved in their own medium... don't handle the stuff themselves. Rocket Scientists don't get to ride on rockets. Dremeling and soldering and scoping is cool.