Is tearing off of SMD components common?

[lpc32]

How often do you see boards where SMD components were apparently torn off?

For example, C1427 and C1426 are missing from the bottom side of this graphics card.
A possible explanation, maybe there was a sticker on top that was removed forcefully.




But then, another card I happen to have. C1439 is missing. I don't think a sticker is a likely explanation here.

[Rerouter]

In both cases it looks like a fatigue fracture from the edge of the pads to mid height in the ceramic, generally caused by thermal cycling or the board flexing (modern large graphics cards should ideally be supported at the back corner rather than left swinging in the breeze)

To clarify, they likely fell off, rather than where torn off,

[Yansi]

You wanted to be ecologic, so there you have it. RoHS. That awesome solder, which does not work. (expect for the warranty period) 

[lpc32]

If fatigue, why don't they fall in droves? Just a few that ended up being soldered less ideally?

The one on the purple board is close to the bracket.

BTW, other than removing a similar capacitor from the board for measurement, any suggestions on how to decide what alternative capacitors to use? Any good generic guesses?

[Yansi]

Was both likely a decoupling cap. Possibly something between .1 to 10uF by 6.3V. You can carefully measure the voltage there (possibly less than 5V) and then place a fitting cap there. 1uF might do the job. Better than nothing.

[lpc32]

Is it better to overshoot or undershoot the original capacitance? Any other important parameters other than voltage?

[Yansi]

Both are not optimal. Small capacity does better HF supression, but cannot supply much curent. Big cap cannot supress much HF content, but can deliver heavy currents. (one reason why "hightech" stuff is blocked by multiple caps in parallel in almost all cases)

But if it works fine without any cap, it will /might work with almost any cap there :-)

Or just take off the board the other cap near and take a measurement. Very likely will be the same.

[mariush]

If those were for decoupling (most likely), then use 100nF  of type x7r or cog/npo or in worst case x5r. Even though the voltages are probably under 3.3v, use ceramics rated for 25v or higher ... 50v rated 100nF capacitors are easy to find in lots of sizes.  x7r and x5r ceramics capacitance varies with voltage, an 100 nF with - let's say - 5v on it won't be 100nF, it would be much smaller.

The common ceramics used are 10nF , 100nF (0.1uF) , 470nF (0.47uF), 1uF etc

[wraper]

If it is computer hardware, and apparently there are pads left from the missing parts (what means there is no soldering failure), there can be only one explanation. User with long screwdriver and hands growing from ass or technician who assembled/repared it with similar kind of hands.

[wraper]

If those were for decoupling (most likely), then use 100nF  of type x7r or cog/npo or in worst case x5r. Even though the voltages are probably under 3.3v, use ceramics rated for 25v or higher ... 50v rated 100nF capacitors are easy to find in lots of sizes.  x7r and x5r ceramics capacitance varies with voltage, an 100 nF with - let's say - 5v on it won't be 100nF, it would be much smaller.

The common ceramics used are 10nF , 100nF (0.1uF) , 470nF (0.47uF), 1uF etc

Expect 1+ uF, likely 4.7 uf, 10 uf or even higher capacitance at such size. For the first PCB easy to check, capacitor nearby (left) almost certainly is the same. Desolder and measure.

[nukie]

Hi research how ceramic capacitors are constructed then you will know why the pads break off easily. I have seen how these stuff is stacked and stored carelessly in a container then slosh around during transportation, parts rubbing each other will shear off the poor ceramics.

[Whales]

Only a week ago I picked up a graphics card missing several SMD caps

I've tried soldering on some random caps of the same physical size but it did not work.  Presumably the ones on the PCIE comm pairs need to be more specific than 'anything handy' so I'm going to remove some that remain and get their values.

Tell us if you have any luck

[G7PSK]

I have pulled stickers from boards and found SMD components stuck to the sticker and not the board, usually the ones that look like grains of sand.

[Neilm]

I have seen a board where if some animal was trying to get off one of the connectors and decided to use pliers to force it off it was almost guaranteed to take off a small but vital resistor. If that resistor was damaged, the whole board stopped working.

[lpc32]

I have pulled stickers from boards and found SMD components stuck to the sticker and not the board, usually the ones that look like grains of sand.

So smaller ones than the above?

type x7r or cog/npo or in worst case x5r.

Beyond temperature dependencies, should it matter if the caps are class 1 or 2?

use ceramics rated for 25v or higher ... 50v rated 100nF capacitors are easy to find in lots of sizes.

Any disadvantages to higher voltage ratings? Or just size and maybe price?

I've tried soldering on some random caps of the same physical size but it did not work.  Presumably the ones on the PCIE comm pairs need to be more specific than 'anything handy' so I'm going to remove some that remain and get their values.

Tell us if you have any luck

The green card above kinda works anyway. The missing capacitors are at the periphery of the GPU, on the underside. There's some instability at times, but I'm not sure it's the card because the test system is fishy by itself. I'll do some more testing in a known good system.

The purple one was in use for quite a while without problem. I suspect it was without the SMD cap all that time, but I'm not sure. The missing cap is near a power MOSFET, on the opposite side. The card's unstable now, but I think due to two bad electrolytics rather than the SMD.

[askjacob]

type x7r or cog/npo or in worst case x5r.

Beyond temperature dependencies, should it matter if the caps are class 1 or 2?

It is not the temps, it is the ripple capacity. Especially with larger capacity smd ceramics - they "lose" effective capacity depending on the application so get the right ones for the task

[aargee]

I see it quite a bit with our guys pulling cards from card cages. Usually it's larger caps or inductors that get clipped by other boards or the edge of the cage shielding.
"Oh and by the way, I found this floating around in the bottom of the card cage"

[malch]

Is C1428 leaking in the first pic?
The no-lead solder doesnt grab the component as tenaciously as lead solder.

[Yansi]

What shoud be leaking? No-clean flux that wasn't cleaned after reflow soldering? :-)

Nothing leakable in a cermic cap. (except magic smoke).