Drivers with more source current than LM5112?

The LM5112 diagram for source current looks a bit dismal when operated at 5V. Figure at bottom right of page 6:

formatting link

Is there a chip like that with more oomph? Doesn't have to be a Schmitt (would be nice though) but as usual must be cheap (

Reply to
Joerg
Loading thread data ...

Reply to
John Popelish

Reply to
D from BC

Exactly. If there isn't a spec I can't use it in low voltage designs. Datasheets aren't what they used to be in Robert Widlar's and Bob Pease's time.

--
Regards, Joerg

http://www.analogconsultants.com/
Reply to
Joerg

Except that they show NPN follower as output. Wonder if that's true or just "representative"?

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

--
Regards, Joerg

http://www.analogconsultants.com/
Reply to
Joerg

Reply to
John Larkin

Rds(on) is of limited usefulness, because the output currents saturate. They have plots showing 7A, 4A sink, source currents. We usually want to know what the driver can deliver into a Miller-clamped MOSFET gate, etc., and they specify into a load at 5 volts. From the falltime, risetime curves into 10nF, we can also deduce the output currents must be near to the 7A, 4A stated levels. Not bad.

One very nice features of NSC's LM5112 driver IC is its isolated Vee pin. Very useful.

As far as Joerg's need for a higher output-current capability is concerned, most folks in the industry, so far as I've observed, use cheaper lower-current driver ICs with a pair of P-channel, N-channel output MOSFETs in soic-8 cases. These can easily deliver up to 30A or so, if necessary, at an attractive low price. And as a bonus you get better control of the current paths and the high inductive voltage drops, with discrete high-output-current-delivery transistors.

Reply to
Winfield Hill

About 250mA. Forward converter, the usual, but I've got only 5V this time.

Yep, almost ready to roll my own. Again. For npn/pnp there are usually no resistors needed (no quiescent current).

Even the FAN3224 is IMHO poorly spec'd. And Digikey doesn't carry it. Says OUT is at VDD/2 for 2.8A sourcing. Most of the graphs just talk about thresholds, no sink/source graphs at all. Who is writing those datasheets these days?

--
Regards, Joerg

http://www.analogconsultants.com/
Reply to
Joerg

It's a MOS/bipolar combo output but precious little info about it.

--
Regards, Joerg

http://www.analogconsultants.com/
Reply to
Joerg

Pull a "Larkin" and characterize it yourself ?:-)

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

That's what I think, too. However, no spec -> no design in :-(

Pretty soon western world datasheets might look like many Asian ones. Two pages or so, most of it being ordering information.

Yes, but then you have to provide lots of extra parts to handle cross currents, brown-out situations and all that. Often I use npn/pnp followers but that eats almost 1.5V off the available swing.

--
Regards, Joerg

http://www.analogconsultants.com/
Reply to
Joerg

Sure, and I bet it would work. But when designing stuff for regulated markets that doesn't fly.

--
Regards, Joerg

http://www.analogconsultants.com/
Reply to
Joerg

You're giving us a demonstration of how a weenie dies from a data sheet. It's pretty clear that it's a follower, and, from the curve of current versus supply... guess what?... you know the source impedance. Unless, of course, they didn't teach Thevenin equivalents in Germany ;-)

Example... what's the source current into +5V from a TTL totem-pole output stage ??

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

If you boost a cmos gate with pnp-npn emitter followers, it won't drive the mosfet gate to the rails. Add one resistor from the emitters to the bases, so the big high-current Miller swing will still blast through the bipolars, but the resistor will then pull the fet gate up/down that extra 0.7 volts to both rails. A lot of the commercial cmos+bipolar drivers do this, except that they use big bipolars for the main swing and small cmos fets for that last 0.7 volts. Near the end of a transition, when the Miller capacitance is gone, the fet gate is relatively easy to drive.

Even better would be an inductor in series with the resistor, but that's usually not worth the hassle.

AND gates are slick, to drive the bipolars, because it's easy to add some rc's for non-overlap delay, to a push-pull forward converter. If you use a pure flipflop (complementary square waves) to drive a pair of logic-level fets, the overlap shoot-through can get grim.

John

Reply to
John Larkin

Understood but I want to drive a small transformer directly, no FETs.

Ah, ye olde peaker inductor.

Either that or a concoction of two resistors and a diode for each. But in high-cost countries SMT placement costs can spoil the fun.

--
Regards, Joerg

http://www.analogconsultants.com/
Reply to
Joerg

Yeah, that could be very handy.

I'm using an LM5112 in my little +12 to -12 converter. The fet is dinky, an NDT2955 with 22 ohms in the gate. I'm seeing 4 ns rise and fall times at the LM5112 output, 12 volt swing, using a 100 MHz scope and probe!

At the fet drain, swing is 25 volts p-p, rise of 5 ns, fall 18.

John

Reply to
John Larkin

Jim, I know it's going to work. And maybe we can even use it. But in the med biz the old rule "no spec -> no design in" can be pretty stiff. I'll be on the horn with National tomorrow morning, find out if they have more hard data.

And they do teach Thevenin there even though Leon Charles was French ;-)

BTW I once had a discussion with a V.P. who wanted to know whether the Kirchhoff rule could be fudged a little. Seriously.

--
Regards, Joerg

http://www.analogconsultants.com/
Reply to
Joerg

Zero!

John

Reply to
John Larkin

Well, yes, but what mankind is really interested in is how far from either rail will it be when sourcing/sinking 100mA, 500mA, 1A etc. That used to always be spec'd with drivers, with guaranteed max values. It's kind of important to gauge stuff such as dissipation. Not spec'd for the LM5112. Harumpf. Grumble.

--
Regards, Joerg

http://www.analogconsultants.com/
Reply to
Joerg

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.