Linear 3.3V power supply anywhere?

All my alarms go off when a datasheet uses the word "optimum." It's code for "something is really wrong with this part and you can work around it with luck and difficulty, if at all."

John

used a TL431 in that function and a LM317 as a

Really (LM317)? What bandwidth?

SioL

used a TL431 in that function and a LM317 as a

Full audio. It was for a little AM transmitter.

I was about to rant about the fluffy choice of words,

"small resistive element" "sometimes" "optimum stability"

but decided to download the LT1764 datasheet and see for myself. Text searches for "resistive", "ringing", "sometimes" or "optimum" turned up nothing ("Optimum" appears once in a different context). Instead, the datasheet dedicates half of page 11 to output capacitor choice, which itself can be seen a warning against the part, but at least they provide a fairly thorough treatment of the problem including a discussion of different dielectrica.

In short: in order to do LT justice you need to update your datasheet. But LDOs are still nasty.

Somewhere I have kicking around a pdf about a "zero-drop" linear regulator using a P-MOSFET as pass element. I can email it to interested parties.

--Daniel

That was a LTC datasheet. I didn't make that up:

Ok, how about this text in the version you may have looked at, can undoubtedly be seen on page 11 (upper right): "The LT1764 regulators are designed to be stable with a wide range of output capacitors. The ESR of the output capacitor affects stability, most notably with small capacitors. A minimum output capacitor of 10µF with an ESR in the range of 50m? to 3? is recommended to prevent oscillations."

Sounds nice, but every bit just as scary to me. What exactly is a wide range here? What is "in the range of"? I'd expect hardcore min max datasheet values, not some nice looking graphs and carefully minced words.

BTW, in the other datasheet above they are talking about down to 20mohms ESR for optimum transient response. Hmm...

Thanks but I'll pass. I am cured of LDOs. If I need one I either roll my own or design a switcher around a Schmitt inverter and some discretes. Works every single time :-)

Ah, that's the LT1764A. I looked at the LT1764.

Anyway, I wasn't trying to convince you to use LDOs -- in fact it was reading *your* posts on this NG that, lacking experience of my own, is going make me steer clear of LDOs should I ever feel the urge to use one.

It does sound scary, but those damn buggers seem to be hard to get stable by their very nature. And since the mfg doesn't know the impedance you're going to hang in parallel to the output cap they necessarily have to be vague about the impedance of the cap as well. What it boils down to is that you have to use DS values as starting points in prototyping, and then work your way from there.

And then your gadget goes into production with LDOs from a different batch or a second source and all hell breaks loose...

Just how does your Schmitt inverter switcher work, and over which range of duty cycles? I've built simple PWMs myself, but they always relied on a sawtooth generator/comparator combo.

--Daniel

I wouldn't mind them if there were guaranteed stability limits in the datasheets. I just will not design per "typical" graphs or text that sounds too vague to me.

Been there. It hurts. A lot.

Sawtooth is fine, too. Basically just an oscillator where a BJT or FET tugs on the input, pulling it down as the desired output voltage is being reached. IOW slowly "starving" the oscillator by reducing its duty cycle. You'll have to make sure there is enough oomph to drive FETs and the like so the whole thing reaches the desired enough efficiency. And always mind any RHP zero issue, of course ;-)