power opamp?

Doing that right now, but my stamp is round and I have gigaohm impedances and all that goes 125°C. Fun...

It's much easier to get real components to burn.

RL

Make it small enough, and even microwatt applications will overheat.

RL

I'm going to modify the drawing a little:

The R1/R2 is made just big enough that the OUTPUT swing is rail to rail for the op-amp selected.

The parallel combination of R1 and R2 sets the "gm" of the "OUTPUT stage".

To ensure that it is stable, you may need a capacitor from the output of the op-amp back to its inverting input. A heavy and capacitive load can add a pole in the feed back loop.

All of the Linear models can't de trusted for the supply currents.

The LT1498's model can be used to power up your circuit. Each one makes about 0.2V IIRC.

A surprising number of circuits are done discrete because of assembly in Asia. The prohibitive placement costs of SMT parts are long gone. So the crossover point where a chip design makes since can be in the tens of thousands of units per year because you must amortize the desigs costs and NRE over a reasonable (small) number of years.

About ten years ago I did a chip design including proto fab for under $200k and this was on an automotive HV process. I don't think that's possible anymore.

This catches many engineers by surprise. 50mW might not sound much but for a resistor the size of a grain of salt that's toasty.

That's what I keep saying, the good old days are now. When I was a kid we didn't have Digikey and a 74LS90 or an RF transistor would cost a week's allowance. You wouldn't dare to burn one up back then.

Unfortunately many of the new grads have never built stuff with real parts.

I regularly do chip designs that come in at around $100K total. You just have to shop carefully for a foundry.

If you think TI, Motorola or Zarlink are good places to go, I have some seaside property in Arizona to sell you.

...Jim Thompson

But I bet that does not include your Syrah and Merlot :-)

The 200k were including my time and that of the layouter.

Oh, we did shop around. Problem is, 45V or 60V processes aren't found seaside in Arizona ;-)

One place I was always quite happy with is AMI, for non-HV stuff. Just not much fun to travel up there because it's so freaking cold in the winter.

X-Fab DMOS?

I'll say. Of course I have an old project, resurrecting for an upgrade, that uses AMI.

Pocatello is so-o-o-o much fun in the Winter.

Last time I was there I made the car rental guy walk me to my car... for fear I'd get lost in the parking lot due to the blizzard.

And those sideways landings in little two-engine prop planes are so thrilling too ;-)

BTW, OnSemi just bought AMIS. I don't whether that will be good or bad.

...Jim Thompson

We used Delco. ABB-Hafo was another contender but they didn't really want to, AFAIR because we had to seriously "bend" a lot of design rules. It was for a disposable and didn't have to live more than a few hours. That tends to give old school semiconductor folks goose bumps.

I still remember my wife's words: "Let's get outta Idaho!"

Yes, ON Semi.

But IMHO ON Semi is a pretty good company. Do you know if Christine King will stay on?

Don't know who is staying. I sent an E-mail to my last device model contact... no response. So I'll call the rep and find out what's going on.

...Jim Thompson

The no response thing is unfortunately becoming somewhat normal. I am still waiting for some app EE answers. More than a week now. They don't even send an acknowledge email anymore and when I called I was told that my case is "in progress". Things are not looking up for some US semiconductor places unless they fix that, and pronto. Seems like mnany of those companies aren't run by engineers anymore and that would be bad news. For them, because us guys just move on. My clients don't want to wait weeks.

I complained too soon. My contact now has another job, and forwarded my E-mail to the present keeper of IP, who replied just in the last few minutes.

...Jim Thompson

Did you make an assumption about gender or age?

John, you'll get better rail performance when you connect a PNP above and an NPN below and pull their respecive base from the supply through a resistor with the power connections of the Amp itself. The current not provided by the Amp is delivered by the Transistors. What ist the requeired bandwidth?

V+ V+ | | R e +---b PNP \\ | c \\ | \\ | --+----- / | / | / | c +---b NPN R e | | V- V-

Rene

If you really want it to do rail to rail:

v+ | v+ R | | e \\VCC----------------------b pnp \\ c \\ R1 | \\-------+--/\\/\\------+-------OUTPUT / ! | / -/\\/\\--GND ! / R2 c VEE---------------------b npn | e R | | v- v-

Rene, You're way behind. I posted that configuration six days ago...

From: Jim Thompson Newsgroups: sci.electronics.design Subject: Re: power opamp? Date: Sun, 02 Mar 2008 14:50:36 -0700 Message-ID:

...Jim Thompson