another LDO

OK, everybody who follows SED should send me a dollar.

The check is in the mail. :)

I promise to match every dollar and donate it to an organization that does real good for the poorest people in the world.

OOooh! I LIKE it! Great potential to be the ONLY LDO that does NOT oscillate!

Damn! You should _charge_ for things like this..

BUT the COTS LDOs just LOVE to oscillate..

Who, obama?

..

It's been a well known phrase for as long as I can remember.

-- Bill Sloman, Nijmegen

Neither was explicitly specified in your original post ... and one of the symptoms of "not invented here" is specification creep.

-- Bill Sloman, Nijmegen

We used to call it a complementary Darlington. They can have tendency to oscillate. Putting a resistor in parallel with the output base- emitter junction made the whole thing a bit more designable.

-- Bill Sloman, Nijmegen

He said "poorest", not "richest".

Idiot. Read the post that started this thread.

down

-90...

You would be the idiot. I wouldn't have made the comment without re- reading your original post - in fact I copied what I was writing onto the clipboard and discarded my first attempt at a response so that I could re-read your original post.

Point out the specific requirements that show up in your original post.

"It needs to work down to maybe 1 volt drop, and needs to be very accurate, especially longterm drift and TC" wouldn't qualify as detailed specification anywhere I've worked.

-- Bill Sloman, Nijmegen

I quote what you snipped:

"It might be fun to power the opamp from the +10 rail, if I could prove that's safe at startup. That does keep from enhancing the fet and preserves its Idss current limit. I only need about 50 mA out, with a nearly constant load. I could even add a source resistor."

Worked?

I suppose you don't think the part you snipped, about needing 50 mA, is specific enough to keep you from recommending that I use a part good for 20 mA.

I noted the inadequate-current problem in an objective manner. You escalated with the fatheaded "NIH" and "feature creep" and were entirely wrong.

LM10 is unsuitable for other reasons, too. Read the data sheet.

I said "poorest", not "dumbest."

Sure, I do that a lot with switchers but have never needed it in a linear regulator of this kind. The opamp is typically very able to drive this with gusto. But on large FETs it does occasionally need a small gate resistor and a 1000pF or so from output to IN-.

They are quite similar except that their bias is shifted.

I think the main reason is that younger engineers have all but forgotten about depletion mode. Even experienced ones often think that's the domain of JFETs which just isn't true. Younger ones have "heard about it" at the university but never calculated or let alone designed anything with them. In the BSP series you can get a nice selection of depletion mode FETs in SOT223 for around 30c, almost buffet-style. That's a nice price considering who that can often avoid the cost for a charge pump or the necessity to keep a bootstrap cap charges.

They are getting better, finally. The ones I like are stable down to zero ESR, namely with a bunch of ceramics on their outputs. I assume they don't have much in the way of internal rolloff... just a bunch of wideband gain.

NIH syndrome refers to engineers or organizations who think that anything they did not invent themselves is crap.

There is also IH syndrome, where they think that anything that *they* invented is crap.

I can potentially slow down the servo loop really hard. The LM1117 is the prime regulator, and all I have to do is slowly tune out the DC error and TC. A 10 second time constant would work.

I'll probably go for 100 millisec or something safe like that, so I don't have to simulate the loop dynamics with the LM1117 model involved. Actually, I have an 1117 on a brassboard, so I can fake the load and drive its ADJ pin from a function generator and get a good idea of what its frequency response is, and just make the servo a lot slower. Soldering is more fun than mouse clicking anyhow.

If I use a slow integrator for the servo, it will limit the turnon dV/dT and protect the tantalum cap from exploding.