Long-term reliability of SMD aluminum electrolytics

[jklasdf]

Anecdotally, there are lots of reliability issues with SMD electrolytic capacitors in older Tek scopes, Keithley multimeters, Sega video game consoles, and many other older devices built around the transition from through-hole to SMD parts. A common recommendation on many of these models is to just replace all SMD electrolytics on-sight -- regardless of if all of them are visibly leaking or not -- because some of them may be leaking underneath and eating into traces on the multi-layer boards. For some instruments built during the transition from through-hole to SMD with different versions (such as the processor board on Tek 2465), there is a huge difference in reliability between boards with SMD electrolytics (which are almost always leaking) and even older boards with through-hole electrolytics, most of which are still fine despite being older (at least the capacitors are not leaking everywhere).

So anecdotally at least, there's a lot of reason to question the long-term reliability of SMD electrolytics...but the plural of anecdote isn't necessarily fact. What's the oldest thing you've taken apart that has SMD elecrolytics that are *not* leaking (at-least visibly)? Are modern ones more reliable? Probably hard to know I guess, since there's usually more failure with age, and most of these are pretty young.

And finally, assuming they're less reliable, why are they less reliable? If you've ever taken a pair of side-cutters to them, their construction is pretty similar to their through-hole counterparts, just with the leads going through a black plastic base to SMD pads. There are through-hole versions that look very similar to the SMD ones, without a sleeve.

Some possible reasons I can think of include the following:

• many of these SMD electrolytics do not have emergency vent cuts on the top (some do though), so maybe they were designed to vent downwards?
• reflow soldering exposes the entire capacitor to higher thermal stress
• flux cleaners may be more aggressive for SMD processes

Any other reasons?

[T3sl4co1l]

Too many to have any control over -- you'll never spot the difference, at least not without destructive testing, microscopes, chemical tests, and extreme in-depth knowledge of industry formulations and techniques.

Electrolyte: solvent, salts (or acids or other ionic compounds), additives, impurities
Separator: thickness, material (cellulose paper or other?), possible reactions with electrolyte or electrodes
Electrodes and leads: thickness, grain structure, impurities, etching
Seal: thickness, material, impurities, mechanical fit, susceptibility and diffusion of solvent through material
Leads, assembly: base metal, plating, solder, flux

Probably even more things than these, and these are just high level outlines, not specific items and how to test them.

Probably a lot of things aren't very important, or never are (maybe sealing materials are well enough understood, and standardized, that they tend not to have problems by themselves?), and can be dismissed with fortunately little evidence necessary.

But other things are huge, and impossible to tell for any mere mortal.  Supposedly the mid 2000s capacitor plague -- that affected many electronics manufacturers, big and small alike -- was a consequence of poor electrolyte formulation.  I don't know the details, and a lot of that will be trade secrets anyway.

So, quantitatively speaking, it really is hard to beat simply having an epidemiological history of some product.  You'll have no idea about the health of an individual -- has it been used for years of total operation, or did it spend most of its life in a dry, cool storage area? -- but you can at least anticipate likely problems from frequent offenders (like the mid-age classic Teks).

Tim

[Ice-Tea]

I think the reason is fairly simple. When the SMD electrolytes came out, most of them came out with 2kh@85C specs. I think a lot of designers that used them never really considered it. Just "oh, yeah, SMD caps, sex on a stick, w00t!". Where at the time, TH electrolytes had much better ratings. So, if you never worried about lifetime of caps (when I started out, this was weirdly not a topic where I worked), being bit in the ass by TH was much less likely than with SMD caps..