Highly Intermittant Production and Tantalum Caps

What's the rate-of-rise of voltage on the test fixture like? If it's much higher than the actual application you may be killing the capacitors by testing.

In our experience, a tant can fail on any powerup, not just the first one. After all, they were powered up at the factory. Peak current, namely dV/dT, can ignite them. It's very batch-dependent.

Tants shouldn't be on power rails, unless the supply is somehow very current limited. Derating about 3:1 on voltage seems to make them reliable.

Linear regs can sometimes be tricked into soft starting, which limits their output dV/dT and peak current.

[...]

Additional info (undated):

Evaluation of Polymer Counter-Electrode Tantalum Capacitors for High Reliability Airborne Applications

One of the few disadvantages of a traditional MnO2 counter-electrode tantalum capacitor is that when it fails it can produce momentary combustion and create substantial residual damage to a PWB if significant current is available.

This damage can often result in the failure of other adjacent components due to the intense heat generated during failure causing parts at adjacent locations to fail.

Polymer tantalum capacitors have the great advantage of a benign failure mode which is desirable for critical or manned applications.

It was because of this benign failure mode that it was decided to evaluate polymer tantalum capacitors to replace legacy devices in several applications although the new components were heavier and more costly than the heritage parts. This evaluation faced several challenges. For example, can you obtain a polymer capacitor with a known failure rate? As has been published elsewhere, polymer tantalum capacitors should share common failure modes with traditional MnO2 products but suppliers insist that they cannot obtain failures using standard Weibull testing. Should polymer capacitor undergo surge current test, if so what should the test conditions be? Is voltage conditioning necessary? What are the equivalent screens and qualification test methods for polymers to those documented in M55365? How do we answer these questions? This paper will describe one approach to this problem and the results of our evaluation.

Introduction

It has been more than 10 years since polymer tantalum capacitors were introduced. Initially they were only available in low voltage ratings and therefore, were not attractive to designers of higher voltage systems. However, this limitation has been overcome and users are now being attracted to polymers because of their benign failure mode and low ESR. In part, it has been reported that these improvements are related to the usage of a new generation of pre-polymerized polymer materials. The most intriguing of these improvements is the benign failure mode. Figure 1 illustrates what can happen when tantalum capacitors fail and there is significant current available.

[...]

We generally prefer polymers, but sometimes a tant has just the right ESR to stabilize a regulator. I use tons of

UNITED CHEMICON APXG250ARA560MF61G 56u 25v polymer

which is happy to at least -5 volts too.

In mid-1990s I recall a datasheet recommending series resistance of one ohm per applied volt. In other words peak capacitor current of one amp.

A few years ago someone here posted a magazine reprint, I think the advice was to slowly ramp up first voltage application after reflow.

piglet

Here are Elna's recommendations for Tantalum caps from 20+ years ago.

I don't see how anyone would be very keen on using them after reading that.

On a sunny day (Tue, 24 Aug 2021 04:09:05 -0400) it happened Spehro Pefhany snipped-for-privacy@interlogDOTyou.knowwhat wrote in snipped-for-privacy@4ax.com:

Well you can say the same thing about electrolytics from all the wall-warts I have, a high percentage had those fail, and from TV sets I have repaired: the same,

Have seen only one tantalum fail in a 19 inch rack parallel to a supply rail in some industrial setup though. Used many in many projects. Small an reliable.

Hype

lots snipped

Reading this thread I was thinking about a design one of my clients builds in 10ks or more, no problems at all with Tantalum caps (100uF

0603 6.3V).

Then I realised it's because we have four in parallel with 100R in series with the power source.

In that application tants are pretty good, we are having trouble getting them but found that ceramic caps of the same notional value were nothing like as good. The caps supply power for a series of pulses into an LED and the effective capacity is pretty much what is claimed. Ceramics used at anywhere near the rated voltage have much less effective capacity (than rated). (By effective capacity I mean Q/dV)

MK

Yes. The voltage limit for small low-voltage high capacitance MLCCs is more set by the reduction of capacitance to some embarassing amount like 20% of nominal than by breakdown. The simsurfing app/web is good for getting those characteristics for Murata parts.

I've found ordinary aluminum electrolytics to be extremely reliable in benign environments, though they do eventually die, there are few to no infantile failures. I've seen MLCCs fail short a few times. If a few failed open usually it wouldn't make any difference. Tantalums tend to be a bit more dramatic.

Electrolytics are like relays, very reliable but limited lifetime, especially in a hot environment. Thyristors, on the other hand, like to fail short from overvoltage/current events that relays easily shrug off again and again.

[...]

Kemet polymers have identical ESR to tantalum but will not explode.

What is the voltage derating factor on these parts?

On a sunny day (Tue, 24 Aug 2021 05:45:44 -0400) it happened Spehro Pefhany snipped-for-privacy@interlogDOTyou.knowwhat wrote in snipped-for-privacy@4ax.com:

Relays, once I had to go round the film editing tables in Bussum (Netherlands) replacing relay contacts... wear out.

I designed high power thyristor supplies as my first job, think 400V 150A etc. Pretty reliable stuff. Big transformers and chokes... Was not really my thing, went into TV / broadcasting next.

Do not they use thyristors to make those HVDC long distance power converters?

Hello Rick, Maybe this will help? For derating components, NASA refers to the jointly-developed standard ECSS-Q-ST-30-11C; it has superceded the U.S. MIL-STD-975:

Solid tants are covered on page 23. It states:

"Surge current shall be derated to 75 % of the Isurge max. Isurge max is defined as Vrated/(ESR + Rs). Vrated is the maximum rated voltage, ESR is the maximum specified value and Rs is the value of series resistance specified in the circuit for surge current testing as defined in the applicable procurement specification."

I didn't see your part specs in the thread. Hopefully your part has these specs provided by its manufacturer. Cheers, RS

That UCC part is about 15 mohms, which is too low sometimes. A comparable tantalum is 250 mohms.

An LM317 or 1117 often drives a power plane with a lot of ceramic bypass caps, which would be unstable. A tantalum tends to damp that system nicely.

I like the UCC part. It's rated for 25 volts, and leakage increases gently to 74 uA at 116 volts, as high as I tried. Some other polymers die hard shorted at around 2x rated voltage.

The UCC is happy for months at -10 too.

1. ESR depends on manufacturer.
2. Why bother with tantalum risk when polymer has such low esr?

1. a polymer with series resistor would provide the same damping without risk of explosion.

There is no need for tantalums when polymers perform so well.

Too low ESR makes many regulators unstable. I explained that.

Our tantalums don't explode. We understand them.

We have lots of nice small tants in stock. No problems.

Comparable polymers cost 5x or more than tants.

Add a series resistor to a polymer. Now you can have whatever controlled damping you need, instead of depending on the unpredictable esr of different manufacturers and batches.

Why bother with tantalums? Polymers are superior and have no risk of explosion.

Which ones? We won't buy them.

No problems as long as you keep them in stock and don't stuff them:)

I'd check the Kemit prices. You might be pleasantly surprised.

I don't see why you need tants anyway. Your boards are multilayer that supply the high frequency bypassing, along with ceramics as needed. For stabilizing linear regulators, a resistor in series with a polymer supplies damping that is much better controlled than relying on esr.

About the only place you need a physical capacitor is in ripple fltering and switching supplies. There, you might even get away with aluminum.

When you really need low esr, polymers are better than tants. That won't occur in many places, so the effect on the bom won't be great. The price difference may not be so great with Kemit caps, especially when you have to derate them, and tants can fail unpredictably and catastrophically in service due to crystallization.

Quote:

Failure avoidance

Solid tantalum capacitors with crystallization are most likely to fail at power-on.[67] It is believed that the voltage across the dielectric layer is the trigger mechanism for the breakdown and that the switch-on current pushes the collapse to a catastrophic failure.

To prevent such sudden failures, manufacturers recommend:

50% application voltage derating against rated voltage using a series resistance of 3 ohm/V or using of circuits with slow power-up modes (soft-start circuits)

You have mentioned needing soft-start circuits before. Why take the risk?