Highly Intermittant Production and Tantalum Caps

Thanks, that would seem to be useful, but that last bit about "Rs is the value of series resistance specified in the circuit for surge current testing as defined in the applicable procurement specification" seems a bit dodgy. What "procurement specification" do you think they are referring to. This would seem to be an important parameter.

Not at all. Digikey and Mouser have millions in stock. Not so good on polymers.

You are cherry picking to try to win arguments. You are telling me that I have a problem that I don't have. That's not engineering.

We're paying 8 cents for some tants, and they aren't failing. They stop LM1117s from oscillating. Nice small parts that pick-and-place well.

Not according to Octopart.

Not according to Octopart. You never look at the links I provide. Check the prices I quoted and see if you can find any better.

You have some incredible purchasing agent. Most of us mortals have to buy from digikey or newark. They don't offer those prices. Polymers are just as small as tants and pick-and-place fine.

[...]

apply the same purchasing magic to polymers and you could get them for peanuts.

Then, instead of depending on uncontrolled and unknown esr of tantalums, you can apply a known and controlled resistor in series with a low esr polymer. You can adjust the damping as needed, instead of having to put up with whatever you get with tantalums.

Using the unknown esr of tantalums is not engineering. Resistors are well known and controlled.

Thanks for the discussion. I learned a lot about tantalums that I never knew before.

I will never use them. Polymers are the way to go.

Oh, I had a JFET amplifier where I had to enforce the source current O/P by a silent CCS and at the same time, for AC the source had to be GND. I decided to use some Oscon to accomplish this. That was one complete catastrophe. The lower noise corner was in the KHz range and it was much, much worse than 1/f. Completely unusable for anything. Every electron that defected through the electrolytic was missing on the output side, and they defected in large groups, Afghan style, so to say.

High voltage AL electrolytics were much better than the organics, and tantalums were much quieter than AL.

A wet slug tantalum pushed that noise corner down into the milliHz region. BTW AVX wet slug tantalums did cost less than half than for Vishay at the same size. Still too much for that job.

In a space project, solid tantalums were allowed, electrolytics were a complete NO. I had to derate the tantalums only by half and not 1/3. But then they were declared switchproof in their data sheet and had about 6 times the volume of normal ones.

Gerhard

Thanks. Good info. You are a genius. I enjoy your posts.

By organics do you mean polymers? I wonder how they would do now. The performance has increased dramatically over time. Polymers are now available in ratings up to 450V. They are basically tantalums with a different insulating layer. I don't know how the leakage compares to tantalums. That probably has a direct bearing on the noise.

What voltage derating is required with the polymer caps?

The part number being used is Kemet T491X157K020AT in a 7343 package, height is virtually unrestricted on a 12V circuit.

Am 25.08.21 um 16:18 schrieb Steve Wilson:

DigiKey P16300-nd as on the bag. Panasonic 16sepf1000m

The SMD version seems to be svpf, at least they are in the same drawer here.

leakage is clearly worse than tan.

Gerhard

Here are some measurements.

Maybe someone can do some tantalums.

Wet-slugs tend to corrode. They are full of acid!

I'd imagine that small temperature fluctuations would affect high-tempco charged capacitors and make 1/f noise. Just putting a plastic cover over an xo or an opamp can radically reduce lf noise.

I wonder if a charged cap has a different specific heat as compared to a discharged one. That might be measurable.

Polymer/organic solid electrolyte caps are not just aluminum and tantalum but now niobium too.

Has anyone here got any experience with niobium caps?

piglet

On a sunny day (Wed, 25 Aug 2021 08:32:29 -0700) it happened snipped-for-privacy@highlandsniptechnology.com wrote in snipped-for-privacy@4ax.com:

Very long time ago I build an audio preamp with OC13 transistors those were transparent Ge with black paint over it:

the table one day I heard noise coming and going, black paint was clearly damaged, sunshine noise. Made nice photo detectors too if you removed the paint. I still have some.

Or room/bench lights.

The early plastic GE silicon transistors were potted in amber epoxy. That had interesting effects. I bragged to the boss how good my discrete opamp was, and he came over to see and blocked the light, and the offset went to hell.

As a kid I had some OC71s I scavenged out of a scrapped Philips telecom system. They were glass painted black and filled with some kind of grease. Still very limited in power (50mW or something like that) and a bit leaky.

Two useful things about alpos are that for a given size and value, you can get them with widely differing ESRs, and that the ESR is pretty stable.

Cheers

Phil Hobbs

Sounds like thermal drift. Any electrolytic cap has a lot in common with a very dead battery, after all.

Random drift has a 1/f**2 spectrum and unidirectional drift goes like

1/f**4.

Interesting. That's one for the bag of tricks. Do you have any plots handy?

Cheers

Phil Hobbs

Stabilizing a regulator isn't a high-precision task. If it isn't OK over a 2:1 range in both directions, it's too close to the edge.

I do sometimes use resistors plus alpos. In fact, the usual zero-ohm jumper that all sensible engineers (*) put at the regulator output often has about the right resistance for the job (0.05 ohms or so).

The issue is that the resistor has to survive some fairly extreme abuse. If you short the supply to ground, or connect it to a very stiff rail such as a car battery, the total energy in the fully-charged capacitor gets dumped into that poor resistor in way under a millisecond.

You have to use pulse-rated resistors with adequate I^2t that they won't blow up. Those don't grow on trees, and they take up precious board space, not to mention the amount of time spent trawling through datasheets to find the right one, and then hoping they stay in stock.

Cheers

Phil Hobbs

(*) A slight troll, but not much of one. Jumpers really are Good Medicine.

Cheers

Phil Hobbs