another LDO

Ha, interesting.. I'll have to remember that one :)

Jamie

But, But, that takes the fun out of it!

Years ago I had a product in my hands and was examining the board and noticed the Tan caps had a brushed on coating that was applied after board assembly. Apparently, they didn't trust them and this coating was a flame retarding that emitted a gasses compound to deprive the area of oxygen.

My guess would be FEP or halar.

Jamie

A dry tantalum doesn't need oxygen. The MNO2 is an oxidizer, and tantalum is the fuel. Once it detonates, the reaction is exothermic.

Detonation is caused by high peak current, namely dV/dT.

I don't think detonation is the right term here. Ignition maybe? But everything else including the pc board requires O2 to burn. Even a small series R can be a big help in preventing the tantalum caps from firing off.

Have a good Memorial day.

Nope, I'm pleased to see you inserted a zero in the loop :-)

...Jim Thompson

Just add a bleed path for the ADJ current... then the output comes up and the OpAmp takes over... maybe, it may overshoot... or it may not start at all... you'll just have to try it, it'll be OpAmp architecture dependent. ...Jim Thompson

ork down

ly

amp-90...

ly

It could still drive a pass transistor/FET/MOSFET just as you did ...

-- Bill Sloman, Nijmegen

down

It's an expensive sole-source near-obsolete part, and the TC would be terrible.

I think I'll go with the servoed LM1117. It's designed to be a regulator, with all the goodies, and a slow servo is all I need to improve the accuracy.

With big ones you may be better off building a wee charge pump because big depletion mode devices are expensive. Unless this project is for low qty gear which isn't very price-sensitive.

Yep, In our I/C world, a charge-pump is cheap. The problem is that an MOS, operated in its resistive region, presents a pole that moves with load... and will, thus, inherently "hunt" (not that is a big deal, it just annoys Larkin that I found it :-) This moving pole also presents transient anomalies dependent on load. I'm investigating a possible way to cure this in a custom CMOS chip... I don't see a way to do it with discrete's, except at great expense. ...Jim Thompson

I won't. It doesn't need it. I can easily have the integrator hit unity gain at 4 Hz.

Your original post, with added zero and feed-forward... almost text-book transient behavior...

Toche. ;-)

But are they (Wall Street, Soros, Buffet, et.al.) the dumbest? They know exactly which side their bread is buttered on. OTOH, they own both sides.

Not really. *Many* years ago, we had an issue with radial ("tombstone") tantalums "igniting" (or whatever you want to call a sudden flash-bang). When they went, they tended to shoot a jet straight down into the circuit board. The bang didn't really harm anything but the flame charred the circuit board, quite often causing a small short between planes. Since there were usually upwards of 1000A supplies behind the boards, often loaded to 200A (or less), there was plenty of margin for the area to continue to heat almost unbounded. This process continued until a lot of magic smoke came out.

Ditto!

Cool, Rds(on) vs. temp has a much lower coefficient than normal.

It's too bad it's "logic level", i.e., Vgs(th) is only 2-4V below the rated Rds(on) spec. Push the gate another 5V and it enhances real nice. If they pushed Vgs(th) down it would have the same effect. I wonder if the increased capacitance and shallower C vs. V profile is also a consequence of the construction.

Tim

Possibly, but I never use them in high-speed apps. Plus they'd be too expensive for that.

Careful. The datasheet is silent about it but I could imagine that the LM1117 suffers if the opamp tries to yank the ADJ pin to the hilt on start-up. That reverse-biases some BE junctions by more than 5V in your case and that is usually unhealthy.

IOW, personally I wouldn't do it.

If you're worried about it, if the op-amp has its own supply, then a diode to Vout.. no big deal. Otherwise, it obviously can't exceed Vout, so no problem.

But if you read the data sheet, it says quite clearly:

"The adjust pin can take a transient signal of ±25V with respect to the output voltage without damaging the device".

o work down

ially

and

op-amp-90...

nally

it

n't

of

I've yet to be terrified by the drift on a reference. Since the LM10 data sheet doesn't bother quoting worst case drifts it's not going to be great. A typical drift of 0.003%/C isn't impressive - 30ppm/C is about what you get in a voltage regulator - but it isn't all that bad. Special purpose voltage references do better, but they do cost more.

It may be expensive and sole-sourced, but it's still one of Bob Widlar's more amazing contributions.

"Some of Widlar's designs mentioned by Swanson and Hodgson, like the LM12 power amplifier and the LM10 ultra-low-voltage amplifier introduced in 1978, remain in production in the 21st century.[78] LM10 is capable of operating with a 1.1 volt power supply, thus instead of a conventional bandgap reference it employs Widlar's sub-bandgap circuit locked at 200 mV[79] and the low-voltage Widlar output stage. [3] For 10 years, no one else in the industry was able to produce a circuit that matched the LM10.[80]"

I suppose anything introduced in 1978 - only 7 years after the 555 - has to be near-obsolete now, but it was a decade ahead of the game when it was introduced, and the 45V to 1.1V supply range is still hard to match.

-- Bill Sloman, Nijmegen