-110dB Low Noise Ripple Filter

This does not show why capacitor leakage caused your problem.

Kemet Polymer Capacitors

KEMET is the market leader in polymer capacitor technology. Our organic capacitors are solid electrolytic devices constructed with a conductive polymer cathode capable of delivering optimized performance in a broad range of applications. Combining very low ESR and improved capacitance retention at high frequencies with a broad dimensional offering, KEMET's portfolio supports designers through innovation and technology.

Type : Tantalum Dielectric : Polymer Tantalum

Polymer Datasheet:

Polymer Tantalum Capacitors Continue to Toughen Up to Meet Automotive Demands

New materials and manufacturing processes enable polymer tantalum capacitors to pass automotive AEC-Q200 stress test requirements, as well specific electrical transient tests. Although the market supply chain challenges with high CV MLCC's availability and new automotive requirements with electrification, digitalization and autonomous driving are bringing new opportunities.

Tantalum Polymer SMD construction is shown in Figure 1. The anode body is a sintered tantalum powder, the dielectric is a thin film of Ta2O5 generated by electrochemical oxidation, and the cathode is a highly conductive polymer layer deposited in the oxide layer. The contact layers consisting of carbon and silver require prior assembly and molding protection. Tantalum Polymer SMD parts are polar capacitors; therefore, is important to pay attention to the polarity marking in the component. Reverse polarity is permitted only up to values indicated in this paper. Although, several application guidelines should be consider by the designers.

Polymer capacitors

The KEMET Organic Capacitor (KO-CAP) is a solid electrolytic capacitor with a conductive polymer cathode capable of delivering ultra low ESR and improved capacitance retention at high frequencies. KO-CAP combines the low ESR of multilayer ceramic, the high capacitance of aluminum electrolytic, and the volumetric efficiency of tantalum into a single surface mount package. Unlike liquid electrolyte-based capacitors, KO-CAP has a very long operational life and high ripple current capabilities.

'Bad practice?'

Why? You do have to worry about startup issues and so on, but once the circuit is running, it's a very good thing to use the regulator's supply rejection to keep the input's junk out of your voltage reference.

Cheers

Phil Hobbs

What? Me? Worry?

Alfred Meumann

RL

I see, you're MAD. Makes sense. ;)

Cheers

Phil Hobbs

.-"-. _/_-.-_\_ / __ } {__ \ / // " \\ \ / / \ '---' / \ \ \ \_/`"""`\_/ /

Yes!!!!

Fingernails down blackboard.........

Yeah, if the output is 0V, and this is the power for the regulator, its zero, which is a solution. Thus it never power ups.

Its a disaster for any competent production design. Anyone doing this, should be fired.

Sure its an excellent technique, *if* one has a formal start-up circuit.

A somewhat novel one is here: :-)

-- Kevin Aylward

I've had that sensation reading a fair few of your posts. ;)

There are all sorts of possible lockup states that have to be avoided. What's specially heinous about that one?

In an IC, sure. At board level there are lots of possible ways of eliminating lockup states, e.g. a low-capacitance TVS from the input power rail to the regulator output.

Nice. I don't usually need to have my startup circuits power themselves down like that.

So the net seems to be that it's another one of those cargo-cult engineering prejudices, like avoiding beta-dependent circuits even when they work over the full datasheet range of betas plus a safety factor.

Cheers

Phil Hobbs

That is very elegant! Thanks.

piglet

You skipped RL's following post:

Bandgaps are too noisy. They are not used in this application.

You're replying to Rob, not to me.

Cheers

Phil Hobbs

I am replying to you.

I did not need to reply to Rob. He found the problem himself.

Well, Kevin then. I didn't comment on your circuit, and anyway, if one keeps alert to lockup states, there's nothing whatsoever wrong about running the reference off the regulator's output, whether you did it or not.

Doing that is sometimes a bit subtle, because you have to consider all the relevant threat models--cold start, warm start, output short, input short, overcurrent, and so on.

Bob Pease wrote about a boss of his from long ago who tested Pease's regulator by running a bastard file across its output terminals, back and forth, to see if it could be induced to blow up.

One does have to be careful, but it's often very fruitful to disregard shibboleths such as "don't run the reference off the regulator output or the (unspecified) boss will fire your ass."

Cheers

Phil Hobbs