another LDO

Ahem Joerg,

where's your low cost analog dedicated super brain?

That's a real no brainer...

A 5V opamp attached to the positive rail and tie your ref to the output rail so that you can float all that above ground.

Sorry the AAcircuit ascii art program refuses to fire up right now...

Where are the "dedicated super brains" right now? With half my brain tied behind my back, solution to too-low ESR caps is right there in front of all eyes... better ripple and better regulation as well... stored away in my secret bag of tricks. ...Jim Thompson

There are little power N-FETs available that can handle that service and only

Yup. 71° of phase margin, 65dB of gain margin, with only 10mA of pre-load; improves towards 90° as you load it.

John "Frickin' Amateur" Larkin can Tourette all he wants, but he sure as hell can't analyze feedback loops ;-) ...Jim Thompson

Then why do all my hand-rolled ones work reliably while many LDO chips don't?

That's where thorough design methodolgy comes into play. If you don't think through start-up and shut-down you have no viable design yet. It has to be done, and it ain't rocket science.

This is in no way meant to paint me as Mr.Superman here. It's all in the thoroughness of work. It is why when someone wants me to design a complicated regulator and says "We've budgeted four hours for this" I'll turn them down.

ROFL! You don't honestly believe that, do you? I have personally experience the contrary numerous times. Up to the point where IC designers exclaimed "I can't believe this is happening!".

Maybe you need to get out there some more, out from behind the big Cadence screen, to where the stuff is really used :-)

My opinion is this: If an LDO becomes instable because there's too many low-ESR ceramics on the power rail for bypassing, then the LDO design is junk. I can't use it, and I won't.

Careful. For example they are guaranteed 9ohms at 0.9V alright, but only for 10mA:

John needs 50mA.

Careful with the LND250. I've heard the clanging of the last order bell for those.

Not impossible, of course, but can get iffy on start-up. Also, RR amps cost more than a quarter of an LM324 (so we are talking 2.5 American cents here ...). Add in 8 cents for a BSS139 and that's it. No extra parts, no extra stuffing costs. It'll have to be enhanced a few hundred mV to do the 50mA but that's no sweat here. Best of all: You can have current limiting for free, gratuitement.

=20

Aren't you talking about a capacitor bypass in the voltage divider of the o= utput into the error amplifier:

Please view in a fixed-width font such as Courier.

. . . . Vo . OUT -----+----+---------+-->

. | | | . | | C | . [Ru] --- | . | --- | . Vf | | +| . error

output into the error amplifier:

Just exactly how do you know how some chip will start up? Start up is quite tricky and never well defined. It may not be repeatable in the future since chips are moved to new processes. You are simply deluding yourself or being ignorant on purpose.

LDOs are complete systems where all the parts are designed to work in unison, including start up.

I am amazed at how system designers think as compared to chip designers. When you sell something by the millions, you worry about everything.

It's NRND, or "deprecated" as people say lately, but the LND150 is essentially the same part.

And there seem to be Infineon versions.

Uh, that would be sez you.

Sorry, I never had to say that regarding performance. ESD, latch-up, that is something else.

If your parts have 0.001% supply tolerance, I deem them junk. What else can I say. Real life parts have what I consider a reasonable operating range that can easily be met by COTS LDOs.

Now if you need better tolerance, then like I said, you need to be thinking buffered reference, not LDO.

That all makes so little sense I wouldn't know where to start to respond.

Nope, my clients.

You wrote, quote, "tested as a system". To me that means a whole lot more than just ESD and latch-up. Most LDO chips fall woefully short there. Give me an example of a low cost LDO chip (meaning

Tolerance is almost never the problem with LDO. Stability is.

If I need tight tolerances I know what to do.

output into the error amplifier:

It means IC's can't be "designed" by throwing s*it at the wall, building whatever sticks, then "tweaking" it... your method. ...Jim Thompson

Jim Thompson a écrit :

output into the error amplifier:

I doubt the LM324 is up to the ppm level task.

So much for the 2.5c opamp :-)