Tantalum caps

You only work on crap designs, though.

Really? How narrow? Have you ever connected one to a network analyzer to see how it looks?

I did one load 8-channel cell conditioner board where I allowed the user to select between 5 and 10 volt excitation. I used a remote-sense closed-loop thing to servo the transduced excitation voltage. The final output was an opamp driving the adjust pin of an LM1117, regulating +12 down to 5 or 10. There was a 16 volt tantalum on the output. Changing the programming from 5v out to 10v out blew the caps.

John

...said AlwaysWrong who doesn't know any better. AS ALWAYS.

I've personally never seen a tantalum fail unless it blew up from surge current or was installed backwards. But I use ceramics whenever I can, for lots of reasons. A 22uF caramic is reasonable nowadays. Polymer aluminum caps are great, too.

Some of my stuff gets used cold, -20 or even -40C, and aluminums get huge ESRs down there.

John

Your parents could've done with.................

decoupling,

Right. As Spehro pointed out, adding a few feet of very skinny 1/2 oz copper trace will fix the problem.

John

normal electrolytics.

the wrong way...

I have. Your typical Ta cap looks like having about 3nH and

Jeroen Belleman

On a sunny day (Thu, 22 Sep 2011 07:14:54 -0700) it happened John Larkin wrote in :

decoupling,

Lest keep this in perspective, we are talking about rise-time of the voltage at the cap.

So if you connect your 24 volt supply with a 10000 uF parallel to its output directly, then the rise-time is short, and the current peak huge. You may argue that the tantalums should handle that, but near infinities created by their VERY low ESR argue against that. So it is a design issue, add some resistance, or a small inductor, that does not drop so much voltage, then you get i = t * U / L, and those infinities are gone. And as I stated before, in those case a fuse on board helps, also to protect your board.

high.

decoupling,

Hand waving. Do some math.

John

I was talking about Phil's comment about 100nF caps being very narrow response.

A vasectomy and tubal ligation before they met.

And abstinence, just to make sure.

John

The manufacturing department of one of the companies I have worked for connected a few tantalums backwards. We tried to locate those boards and fix them but for some it was too late as they already burned a hole straight through the PCB (15A PSU). IIRC it took a year before the last one blew.

That is true, but ceramics are a superior alternative to tantalum nowadays.

lose?).

Bad electrolytics are a typical failure of TVs, monitors and power supplies. However, I've come across a lot of failed tantalums in computer equipment where they where used for decoupling power supplies.

Had they any concience they'd have tied him in a sack of bricks and tossed him in the river at birth.

IIRC IBM used to have special 3-legged tants, they never put them in the wrong way round.

come lose?).

Sure, there was a spate of bad tantalums some years ago. It was an infamous industrial espionage thing. Most of them ended up in computer motherboards, where the manufacturers were trying to save every nickel the could.

I've seen problems with tantalums over the years, too. Most were production problems and some were defective components and fewer still, design. One rule we had was that tantalums had to be underrated 50% (resistors, too, but that's another issue). Only a few manufacturers were qualified and their quality was strictly monitored. This is likely why most of the issues ended up being production problems. They just couldn't seem to get the damned things in the right direction. ;-)

Over the past couple of years I designed most tantalums out because of cost or availability, not because I was afraid of them. As others have pointed out, ceramics in the 1-10uF range have gotten cheap. Larger caps have been a problem because aluminums are pretty bad WRT RoHS. Some high-temp aluminums have become available but only smaller ones and they're not cheap either.

Yes, and 4-legged. Since many boards had a hole every .100", it was still possible to get them a pin off. The 4-legged ones made it more obvious. ;-) If it was possible, it happened.

There were also big-leg, little-leg tantalums, meant to go into big-hole, little-hole on the boards. One day my boss got a call from production bitching that the "girls" were getting sore thumbs stuffing the capacitors into boards. If it was possible, it happened.