Yep, it certainly does - very similar construction to the Apple machines of the era. I expect that type of capacitor was common for the period - I know the Amiga 4000 has them as well (and is plagued with similar issues). Pete

FWIW the NeXTcube has the same shoddy SMT can caps, but last time I gave it a peek to see if they were actually leaking, they were not- whereas all the cans in its contemporaries had spilled electrolyte. Perhaps a different MFG or SKU of cap? Pete, were yours actually leaking? Or just the same shoddy type?

On Mon, 27 Apr 2015, Ian Finder wrote: Modern made top brand SMD aluminum electrolytics (such as Panasonic, Nichicon, Rubycon) also don't tend to exhibit this failure mode. There were two primary causes of the end-seals in the late 1980s to mid 1990s era SMD aluminum electrolytics failing. One was that the "rubber" compound used for the capacitors end-seal didn't hold up well to the higher temperatures used in reflow soldering and would deform. Another was the use of chlorinated solvents to clean boards after soldering. Manufacturers solved the end-seal issue by switching to rubber compounds which could better handle the higher temperatures encountered in reflow soldering. This is why modern parts soldered at even higher temperatures using lead-free solder aren't failing like these early SMD aluminum electrolytics did. Chlorinated cleaning solvents OTOH, are still a problem for aluminum electrolytic capacitors, even for through-hole parts, and most capacitor manufacturers explicitly state in their datasheets not to use them. Not only can chlorinated solvents damage the rubber compound used for the capacitors end-seal, but they can wick into the capacitor and cause the aluminum itself to corrode. It wasn't that those early SMD aluminum electrolytics were shoddy or defective, it was that the electronics industry didn't understand how to handle them during assembly. This is similar to what happened in more recent times when manufacturers first began switching from tin/lead solder to lead-free solder. Instead of replacing process equipment which had been designed for tin/lead process with equipment designed explicitly for lead-free process, many board houses at first tried just running their reflow ovens and wave soldering machines at higher temperatures. In hindsight, the results of doing this sure look to have been predictable.

On Mon, 27 Apr 2015, TeoZ wrote: Seconded. I don't own a hot air system (yet) and I do plenty of SMD work, including 0402 stuff and sometimes smaller. The single most important thing to have is liquid flux such as flux pen. You also really want a wide tip (chisel, knife, or hoof) on the soldering iron to carry molten solder to the joint. This is totally different from through-hole work since you are applying flux directly to the joint before applying solder. With enough flux, the right amount of solder just wicks right into the joint (yes, really!). Even fine pitch QFPs don't give me any trouble. The only thing I've yet to touch are BGA packages, and I'm really not looking forward to the day when I have to... Still...I see a lot of people using SMD tantalum parts to replace SMD aluminum electrolytics, and after seeing so many of these SMD tantalums burned to a crisp after they developed internal shorts, I certainly wouldn't use them for general purpose bypass work. SMD solid polymers which do not exhibit that type of failure mode are readily available in the same case sizes (round cans) as the original aluminum electrolytics and are also available in the same case sizes as SMD tantalums. Not only that, but solid polymers are less expensive than tantalum parts.

<snip> </snip> The trick to replacing the aluminum cans with Tantalums is to over-provision substantially on voltage. A tantalum will eventually react much more violently to transient spikes than an aluminum can, as such anomalies slowly degrade the oxide layer in the cap, eventually leading to thermal runaway. A popular example are the 47uf 16v SMT cans that are common in Apple gear. People replace them with 16v tants but they should really be going up to about 25v. If you leave adequate headroom in the voltage spec, you will typically not suffer this failure. Still agree that if you wanna be truly safe polymer is the way to go. Cheers, - Ian