SMD aluminium electrolytics (non-polymer): advantageous?

[Whales]

I bought a few 1000uF 16V 10x10mm SMD electrolytic caps (non-polymer) to test out on a few non-production projects.  My experiences so far:

Pros:

• Look cool
• Less holes to drill in PCB (but not by an amount that really matters, as opposed to ICs)
• Pick-and-placeable

Cons:

• More expensive than equivalent TH parts (am I not looking widely enough?)
• Take up more 2D boardspace than than TH parts (probably due to the need for bigger solder pads & less mechanical leverage)
• Hard to skillet reflow due to their heatsinking ability (not sure about IR, hot air, etc)

It seems that they are only a good choice if you want to reduce TH hand-assembly on your product.  Alas many of the things I think of still need TH connectors.

Does anyone enjoy using SMD aluminium electrolytics in their products?  Do they have some advantages I'm not aware of?

[Siwastaja]

They are slightly more expensive than equivalent TH parts even though manufacturing isn't that much more complex, only involving adding the small plastic "plate" under the cap and forming the leads.

There are some cost-savings in reducing number of TH parts and if you can go full SMT including all connectors in SMT as well, the difference can be significant as you can completely omit a production step. These savings may totally offset the price difference of the parts.

I use them extensively and have seen absolutely no problems, be aware that largest sizes are not mechanically as sturdy as a TH part which is also easier to glue down for added protection against vibration. I prefer smaller sizes than 10x10mm. Very handy in providing that damping in DC buses with a lot of MLCC taking all the ripple, or as a board side bulk cap on a board full of 100nF caps, yet fed with already regulated DC so you are fine with, say, just 100µF or so.

Yes they waste a bit more area on a PCB especially if you draw the courtyard as a simple rectangle. OTOH, you don't necessarily have to, I'm happy to use odd-shaped courtyard and place other components accordingly. Best to ask your CM beforehand whether such configuration is a problem in P&P.

[Faringdon]

If longest lifetime is your goal, then thru hole Al Elec caps can be found  with longer hours lifetime at higher temperatures than SMD ones.
eg "105C at 10000" hrs etc.

Al Poly does have longer lifetime than Al Elec aswell. If  you want the damping, you can always put a series 1206 (say) resistor in series with an AL Poly.

An AL ELEC cap doubles its lifetime with 10degc temp fall
An Al Poly cap ten times' its lifetime with a 10degc temp fall.

In short SMD AL ELEC's are cheap and thats it (mounting cost and part cost)...and if your product doesnt have a long warantee...then people shovel them in.

[Siwastaja]

I guess the lifetimes are seemingly better with through hole parts because they might be a tad larger for the size, i.e., lower energy density, higher surface area to ripple rating ratio. If I'm right on this, then it would mean if you just derate the SMD part a tad more than you would derate the TH part, the reliability would be the same.

For demanding high ripple applications, indeed choose a TH part, and that would be easier to heatsink as well with lower thermal impedance connection to multiple planes (say one full GND plane, another large Vcc pour) on a multilayer PCB.

For example, in an input of 4-channel (4-phase) synchronous buck with 12V input and 4 x 4V, 30A outputs, I obviously chose leaded electrolytics from Panasonic FR series, put enough of those in parallel, and arranged them so that they are fan cooled as well, and the first components where the cold air hits before warming up from the MOSFETs and inductors. I would definitely not do this with SMT parts, although now the polymers more affordably priced, I could consider using lower-capacitance, smaller size polymer solution with SMT parts.

And this is the important difference between standard al elcap and the polymer cap; the former is often chosen by ripple current rating (i.e., ESR) and ends up being whatever capacitance it is because it's far more than the calculated minimum capacitance anyway. With the polymer type, ripple rating being so much better, you usually use a lot less capacitance, sometimes just the minimum you calculated from charging-discharging ripple voltage equation.

But in many applications, they are just for damping or bulk storage and see very very little ripple current at all. In that case, the hour ratings at the rated extremes are not that helpful, and it's very hard to extrapolate what happens when the cap is mostly just sitting with some DC bias over it and only seeing occasional power-on current surges. No one then knows if there is any difference between a 2000 hour and 10000 hour parts.