Unlike regular electrolytics Tantalum capacitors don´t age and they cannot dry out. Instead they keep their nominal values for ages.
There´s no need for replacement at all.

majestyk wrote on 2022-08-27, 04:55:

Unlike regular electrolytics Tantalum capacitors don´t age and they cannot dry out. Instead they keep their nominal values for ages.
There´s no need for replacement at all.

In practice, tantalum caps can and do fail, often by going short, which can result in "fireworks" when a given cap heats up enough due to the short.

This is a known failure mode for older (80s and early 90s) tantalum caps. No idea if this is still a concern for newer ones.

yes, 50 year old tantalums can blow up, OP indicated 2x younger gear

How many shorted SMD-Tantalums have you guys seen going up in fireworks in your lives?

Early, radial Tantalums did have issues when operated near their nominal values (e.g. 16V Tantalum at 12V DC). Even erlier, the first Tantalums needed always a (serial) resistor to limit maximum current.
This isn´t an issue anymore today and I´m convinced it´already wasn´t an issue when SMD Tantalums were introduced.

majestyk wrote on 2022-08-27, 13:01:

How many shorted SMD-Tantalums have you guys seen going up in fireworks in your lives?

Early, radial Tantalums did have issues when operated near their nominal values (e.g. 16V Tantalum at 12V DC). Even erlier, the first Tantalums needed always a (serial) resistor to limit maximum current.
This isn´t an issue anymore today and I´m convinced it´already wasn´t an issue when SMD Tantalums were introduced.

My Tandy 1000 RLX in 2017ish (back board with the sound) magic smoked a tant. Then I got a bulk lot of ATI Rage DVD cards, and the first one magic smoked. Same cap, on the second, then third card, so I replaced them all on the entire lot (this year) and they all work like a champ. I'm worried about putting an expensive CPU over a little bank of possible TNT, 🤣!

Yes, I did find 2 already failed in compaq notebooks, blown hole.

Cheers,

TimWolf wrote on 2022-08-27, 01:53:

Going to recap the 22uf 25v button caps on my Zenith with some nice Wurth 105c 5000hr caps. While I'm at it, these other caps, some tants are under the cpu. Very low profile. They still work but as thru are 25yrs+ old I'm pondering doing them as well. Problem is what to use. New tants, ceramics with 2x the voltage, or??? I'm kinda limited due to the height.

Thanks in advance for the input!

I've a trick for you : under Windows 95 (only 95, and Not 98), if the ACPI power management system (normaly activated by default) does not cause any BSOD, it is show that these CPU caps are ok.
It shows in less than an hour of use.

pentiumspeed wrote on 2022-09-12, 18:59:

Yes, I did find 2 already failed in compaq notebooks, blown hole.

Cheers,

Good to know! I guess I was just lucky so far...

majestyk wrote on 2022-08-27, 13:01:

Early, radial Tantalums did have issues when operated near their nominal values (e.g. 16V Tantalum at 12V DC). Even erlier, the first Tantalums needed always a (serial) resistor to limit maximum current.
This isn´t an issue anymore today and I´m convinced it´already wasn´t an issue when SMD Tantalums were introduced.

Interesting... I always assumed that the mechanism for failure of tantalum and niobium capacitors was the equivalent of "tin whiskers", or crystalline growths shorting the cathode to anode.

But there's also something else... Distributors are very careful with moisture levels on SMD tantalums... In fact, there's an even industry practice of baking tantalums to dry them out. I wonder why all the fuss and worry about moisture ingress if they're so invincible, as you say.

Wow all good info. The machine was part of a 100 computer haul in a tireless uhaul for 25 years with a rotted out floor on the gulf coast, in a hurricane zone. Humidity was a major factor.