Forums

MLCC soldering and dielectric cracking damage

Started by Mr.CRC October 23, 2011
Hi:

I frequently assemble PCBs by hand soldering MLCCs of the 0603, 0805,
1206, and a few larger ones (and other parts too--as those
capacitor-only circuits aren't much fun!).

My hand soldering technique is to use 0.015" wire solder,
1. first adding a tiny bit of solder to one pad.
2. Then I flux both pads, and tack the part to the pad with the added
solder.
3. Sometimes at this point, I put my tweezers on top of the part to
create a downward force, and reflow the tacked joint to make the part
seat squarely on the pads.
4. Then I solder the other pad-to-part joint.
5. If necessary to make it look nicer, I add flux and reflow the tacked
joint.

The question is this:  How much should I worry about thermal stress
cracking or otherwise damaging the MLCC dielectrics?

I have heard that using a pre-heated PCB, and soldering the MLCCs both
joints at once using hot air is the preferred approach, to avoid damage.

Also, that MLCCs with thinner dielectrics are more susceptible, like
high values in small sizes.

Yet in practice, I've never noticed a bad part.  Then again, since most
of them are bypass caps, it's hard to notice a bad part.

Any experiences with soldering causing MLCC damage?

How about board flexing?  What amount of flex causes trouble?  Ordinary
fondling?  Assembling CPU coolers onto PC motherboards is always a
treacherous experience!  I wonder how many MLCCs survive that experience.


-- 
_____________________
Mr.CRC
crobcBOGUS@REMOVETHISsbcglobal.net
SuSE 10.3 Linux 2.6.22.17
Mr.CRC wrote:
> Hi: > > I frequently assemble PCBs by hand soldering MLCCs of the 0603, 0805, > 1206, and a few larger ones (and other parts too--as those > capacitor-only circuits aren't much fun!). > > My hand soldering technique is to use 0.015" wire solder, > 1. first adding a tiny bit of solder to one pad.
Why not soldering the part onto that pads right away? That how I do it.
> 2. Then I flux both pads, and tack the part to the pad with the added > solder.
I never use extra flux when soldering such SMT parts. The solder I use it 0.015" Kester 8806 No-Clean, leaded.
> 3. Sometimes at this point, I put my tweezers on top of the part to > create a downward force, and reflow the tacked joint to make the part > seat squarely on the pads.
Tweezers are metal, can scratch and stress the part. I use a toothpick, usually. A fresh one :-)
> 4. Then I solder the other pad-to-part joint. > 5. If necessary to make it look nicer, I add flux and reflow the tacked > joint. >
Suggest to try to practise until they look good without re-fluxing.
> The question is this: How much should I worry about thermal stress > cracking or otherwise damaging the MLCC dielectrics? >
IME it becomes a concern for packages larger than 1812. This goes for normal PCB thickness.
> I have heard that using a pre-heated PCB, and soldering the MLCCs both > joints at once using hot air is the preferred approach, to avoid damage. > > Also, that MLCCs with thinner dielectrics are more susceptible, like > high values in small sizes. > > Yet in practice, I've never noticed a bad part. Then again, since most > of them are bypass caps, it's hard to notice a bad part. > > Any experiences with soldering causing MLCC damage? >
I've seen failed parts but (so far) never on boards I hand-assembled.
> How about board flexing? What amount of flex causes trouble? Ordinary > fondling? ...
Fondling? Hey, that can get you in trouble with the laws :-)
> ... Assembling CPU coolers onto PC motherboards is always a > treacherous experience! I wonder how many MLCCs survive that experience. >
That one I don't understand. How can such coolers stress the caps? -- Regards, Joerg http://www.analogconsultants.com/
On 23 Okt., 19:12, Joerg <inva...@invalid.invalid> wrote:
> Mr.CRC wrote: > > Hi: > > > I frequently assemble PCBs by hand soldering MLCCs of the 0603, 0805, > > 1206, and a few larger ones (and other parts too--as those > > capacitor-only circuits aren't much fun!). > > > My hand soldering technique is to use 0.015" wire solder, > > 1. first adding a tiny bit of solder to one pad. > > Why not soldering the part onto that pads right away? That how I do it. > > > 2. Then I flux both pads, and tack the part to the pad with the added > > solder. > > I never use extra flux when soldering such SMT parts. The solder I use > it 0.015" Kester 8806 No-Clean, leaded. > > > 3. Sometimes at this point, I put my tweezers on top of the part to > > create a downward force, and reflow the tacked joint to make the part > > seat squarely on the pads. > > Tweezers are metal, can scratch and stress the part. I use a toothpick, > usually. A fresh one :-) > > > 4. Then I solder the other pad-to-part joint. > > 5. If necessary to make it look nicer, I add flux and reflow the tacked > > joint. > > Suggest to try to practise until they look good without re-fluxing. > > > The question is this: =A0How much should I worry about thermal stress > > cracking or otherwise damaging the MLCC dielectrics? > > IME it becomes a concern for packages larger than 1812. This goes for > normal PCB thickness. > > > I have heard that using a pre-heated PCB, and soldering the MLCCs both > > joints at once using hot air is the preferred approach, to avoid damage=
.
> > > Also, that MLCCs with thinner dielectrics are more susceptible, like > > high values in small sizes. > > > Yet in practice, I've never noticed a bad part. =A0Then again, since mo=
st
> > of them are bypass caps, it's hard to notice a bad part. > > > Any experiences with soldering causing MLCC damage? > > I've seen failed parts but (so far) never on boards I hand-assembled. > > > How about board flexing? =A0What amount of flex causes trouble? =A0Ordi=
nary
> > fondling? ... > > Fondling? Hey, that can get you in trouble with the laws :-) > > > =A0 =A0 =A0 =A0 =A0... Assembling CPU coolers onto PC motherboards is a=
lways a
> > treacherous experience! =A0I wonder how many MLCCs survive that experie=
nce.
> > That one I don't understand. How can such coolers stress the caps? >
they are usually attached to the board with some contraption made of springs clips and standoffs that need to snap into holes in the board, it can take scary amount of force and bending of the board to make it happen -Lasse
On Sun, 23 Oct 2011 10:03:52 -0700, "Mr.CRC"
<crobcBOGUS@REMOVETHISsbcglobal.net> wrote:

>Hi: > >I frequently assemble PCBs by hand soldering MLCCs of the 0603, 0805, >1206, and a few larger ones (and other parts too--as those >capacitor-only circuits aren't much fun!). > >My hand soldering technique is to use 0.015" wire solder, >1. first adding a tiny bit of solder to one pad. >2. Then I flux both pads, and tack the part to the pad with the added >solder. >3. Sometimes at this point, I put my tweezers on top of the part to >create a downward force, and reflow the tacked joint to make the part >seat squarely on the pads. >4. Then I solder the other pad-to-part joint. >5. If necessary to make it look nicer, I add flux and reflow the tacked >joint. > >The question is this: How much should I worry about thermal stress >cracking or otherwise damaging the MLCC dielectrics? > >I have heard that using a pre-heated PCB, and soldering the MLCCs both >joints at once using hot air is the preferred approach, to avoid damage. > >Also, that MLCCs with thinner dielectrics are more susceptible, like >high values in small sizes. > >Yet in practice, I've never noticed a bad part. Then again, since most >of them are bypass caps, it's hard to notice a bad part. > >Any experiences with soldering causing MLCC damage? > >How about board flexing? What amount of flex causes trouble? Ordinary >fondling? Assembling CPU coolers onto PC motherboards is always a >treacherous experience! I wonder how many MLCCs survive that experience.
The literature sure says it's not good to hand solder MLCC parts. However, I haven't had any problems with sizes from 0402 to 1812. We use Metcal soldering irons which can be used with lower temp tips than ordinary irons since the tip is induction heated. 0.010" diameter solder is the way to go for 0402 parts. I don't use flux as it isn't necessary for these parts. I have seen problems with cracked caps due to them being too close to the edge of the board (depaneling issues) and bad profiles on an assembly line.
Because if you have only two hands hands, you cant hold the part,
solder and the solder iron all at the same time. I have tried.
Evolution is not fast enough to grow the third hand.
> > My hand soldering technique is to use 0.015" wire solder, > > 1. first adding a tiny bit of solder to one pad. > >Why not soldering the part onto that pads right away? That how I do it.
> Tweezers are metal, can scratch and stress the part. I use a toothpick, > usually. A fresh one :-) >
Heh
On Sun, 23 Oct 2011 10:03:52 -0700, "Mr.CRC"
<crobcBOGUS@REMOVETHISsbcglobal.net> wrote:

>Hi: > >I frequently assemble PCBs by hand soldering MLCCs of the 0603, 0805, >1206, and a few larger ones (and other parts too--as those >capacitor-only circuits aren't much fun!). >
The whole key is temperature soakings and keeping them and their degree to a minimum. ANY reflowing of chip caps detaches end terminations and the chip may read correctly but not perform to spec or even fail in use.
>My hand soldering technique is to use 0.015" wire solder, >1. first adding a tiny bit of solder to one pad.
The tinier the better, but you must be able to hit the bead with the chip in place, *without* hitting the chip with the iron.
>2. Then I flux both pads, and tack the part to the pad with the added >solder.
Only reflow the bead long enough to seat the part square and flat and allow seizure. Reflux that pad, and then solder the OTHER pad first,very quickly, only adding heat to the PCB pad, not the chip termination 'end cap' (of whatever variety). Never go back to that pad again. It is done. Now solder the other pad to the right fillet size to match the first. Never reflow either pad. The key is to add no further heat OR thermal stress to the part internals or the solder terminations. That heat transfers directly into the chip. These failure modes are very easy to illustrate with HV caps (MLCCs) They read correct and act correct at low voltage after thermal damage, but at higher operating voltages, they fail from end termination detachment issues.
>3. Sometimes at this point, I put my tweezers on top of the part to >create a downward force,
It must be VERY light force, and the tweezer "noses" must form a flat, NOT pointed face to press on the chip with, or you can introduce tiny micro-fractures in the chips.
> and reflow the tacked joint to make the part >seat squarely on the pads.
your first bead should be so tiny that you can place it out of the way of the squarely placed part, so that when you reflow that tiny bead, it and the flux and the capillary attraction will lock it down. Only add heat for the 'moment' of that reflow act, and avoid contact with the actual terminations of the chips themselves. You are making it tack down, not be integral yet, so add no more heat than needed to perform that function. Heat kills chips.
>4. Then I solder the other pad-to-part joint.
That is the correct process step order.
>5. If necessary to make it look nicer, I add flux and reflow the tacked >joint.
if the tack is small enough, it *should* be your second joint anyway, just to make the proper electrical connection. The 'tacking' process step should have been minimal in both solder and the heat infused to do it.
>The question is this: How much should I worry about thermal stress >cracking or otherwise damaging the MLCC dielectrics?
A lot. That is why you take serious steps to develop and follow a strict process like that which I iterated to you. You may want to differ from what I gave you, but that is your choice. The killer is heat. Theprocess is where it gets introduced. One solder joint construction flow (for effect) is all these chips should be subjected to (there is a severe shock). They are meant for automated processes where the temp differential at reflow time is a mere 20 or 40 degrees and the terminations do not flex to their breakage point. So, a hot air reflow (done quickly) with solder paste is an option as well, but adjacent peripheral components get heated as well. For hand assembly, short of preheating the board to a couple hundred degrees and the chips in a hot pot... if you are soldering them cold, as it were, the shock *will* be severe, so you have to perform the ops very quickly and only once, if possible.
> >I have heard that using a pre-heated PCB, and soldering the MLCCs both >joints at once using hot air is the preferred approach, to avoid damage.
Yes, but do NOT try to do "both joints at once" using soldering irons. You are actually doing the exact thing you do NOT want to do at that point.
>Also, that MLCCs with thinner dielectrics are more susceptible, like >high values in small sizes.
It took them many years to be able to even make them. Pretty likely that they are easier to break or cause to shift from stated specs.
>Yet in practice, I've never noticed a bad part.
ceramic caps always appear fine on metrological inspection. "At voltage" is when it counts though. HV caps MUST be "good". When a multiplier cap in an HV circuit fails, one typically replaces all of them because the time savings is cheaper than trying to find which "good reading" cap is "really bad".
> Then again, since most >of them are bypass caps, it's hard to notice a bad part.
Unless they cause a problem in that location.
>Any experiences with soldering causing MLCC damage?
MLCC damage is usually from soldering. If you strike and break one from physical contact, it has nothing to do with a normal electronic industry process. It was some dumb dope who should learn some material handling practices before he (or she) handles delicate components again.
> >How about board flexing?
Solder joint physics area different animal. Depends on the pad/fillet shape, and the type of terminations the part you buy has.
> What amount of flex causes trouble?
If your post process assemblies are warped, you have a process issue other than broken terminations when a stupid tech or manager flexes a warped, solid board back toward its proper design shape. He needs to learn some physics. They need to be held FLAT while they are still hot, all the way until they cool, and if they are still warping, then the PCB maker has induced it in their MFG process and it needs to be addresses even further upstream.
> Ordinary >fondling?
If you grab a board and flex the fucking thing, you lack knowledge and experience about soldering, solder creep, and PCB assembly practices. If it is warped, it is a failure, if it does not fit form fit and function *WITHOUT* attempting to "undo" the flexure and warpage. That is so decidedly NOT "ordinary", by any measure. In some circles, you would be looking for a new position.
> Assembling CPU coolers onto PC motherboards is always a >treacherous experience!
Not really. Could be an aptitude thing.
> I wonder how many MLCCs survive that experience.
Do you walk down the aisle of a crowded train car, shoving folks out of your way with your elbows as you tromp toward the exit door? You one of those asshole dock workers that/who toss packages full of electronic devices up on the dock because they are too dumb to understand what the term FRAGILE means? You do not assemble electronic assemblies with FORD tools, idiot. In this case that references your brain. You MUST be a Ford owner! You sport a Ford owner mentality!
On Sun, 23 Oct 2011 10:12:55 -0700, Joerg <invalid@invalid.invalid>
wrote:

>> ... Assembling CPU coolers onto PC motherboards is always a >> treacherous experience! I wonder how many MLCCs survive that experience. >> > >That one I don't understand. How can such coolers stress the caps? > >--
Learn to read idiot. It is the act of ASSEMBLING CPU coolers onto PC motherboards that this guy does wrong, and you fail to discern. He is obviously striking then physically somehow. Your diagnostic capacity is like nil.
On Sun, 23 Oct 2011 10:03:52 -0700, "Mr.CRC"
<crobcBOGUS@REMOVETHISsbcglobal.net> wrote:

>Hi: > >I frequently assemble PCBs by hand soldering MLCCs of the 0603, 0805, >1206, and a few larger ones (and other parts too--as those >capacitor-only circuits aren't much fun!). > >My hand soldering technique is to use 0.015" wire solder, >1. first adding a tiny bit of solder to one pad. >2. Then I flux both pads, and tack the part to the pad with the added >solder. >3. Sometimes at this point, I put my tweezers on top of the part to >create a downward force, and reflow the tacked joint to make the part >seat squarely on the pads.
That can tiddly-wink them into neverland. I grab them with the tweezers.
>4. Then I solder the other pad-to-part joint. >5. If necessary to make it look nicer, I add flux and reflow the tacked >joint.
All OK, except that I don't bother with extra flux. If I sometimes wind up with a big ball of solder on one end, I just wick a little of it away.
> >The question is this: How much should I worry about thermal stress >cracking or otherwise damaging the MLCC dielectrics?
Not much. Surface mount ceramic caps are, well, made of ceramic. They are tough. We often put a few bypass caps on the bottom of a board, like around an FPGA when the top is solid traces and there's no room up there. If there are a modest number of them, manufacturing adds them by hand. I've replaced zillions of ceramic caps myself, debugging boards. No problems.
> >I have heard that using a pre-heated PCB, and soldering the MLCCs both >joints at once using hot air is the preferred approach, to avoid damage.
Overkill.
> >Also, that MLCCs with thinner dielectrics are more susceptible, like >high values in small sizes.
Most ceramic surfmount caps are about 20 mils thick. Some exotics may be different.
> >Yet in practice, I've never noticed a bad part. Then again, since most >of them are bypass caps, it's hard to notice a bad part. > >Any experiences with soldering causing MLCC damage?
If you are too rough removing one, you can rip the end cap. Any reasonable soldering method seems fine.
> >How about board flexing? What amount of flex causes trouble?
Not in my experience. John
On Sun, 23 Oct 2011 19:03:00 -0700, John Larkin
<jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

>On Sun, 23 Oct 2011 10:03:52 -0700, "Mr.CRC" ><crobcBOGUS@REMOVETHISsbcglobal.net> wrote: > >>Hi: >> >>I frequently assemble PCBs by hand soldering MLCCs of the 0603, 0805, >>1206, and a few larger ones (and other parts too--as those >>capacitor-only circuits aren't much fun!). >> >>My hand soldering technique is to use 0.015" wire solder, >>1. first adding a tiny bit of solder to one pad. >>2. Then I flux both pads, and tack the part to the pad with the added >>solder. >>3. Sometimes at this point, I put my tweezers on top of the part to >>create a downward force, and reflow the tacked joint to make the part >>seat squarely on the pads. > >That can tiddly-wink them into neverland. I grab them with the >tweezers. > >>4. Then I solder the other pad-to-part joint. >>5. If necessary to make it look nicer, I add flux and reflow the tacked >>joint. > >All OK, except that I don't bother with extra flux. If I sometimes >wind up with a big ball of solder on one end, I just wick a little of >it away.
That is a thermal no no. You should apply the right amount or LESS, never so damned much that you have to remove some.
>>The question is this: How much should I worry about thermal stress >>cracking or otherwise damaging the MLCC dielectrics? > >Not much. Surface mount ceramic caps are, well, made of ceramic.
No shit, dip tracy.
> They >are tough.
You are an idiot. The ends are terminated to the termination platings precariously at best. Thermal shocks cause detachments, and most of those cannot be detected with simple test instruments.
>We often put a few bypass caps on the bottom of a board, like around >an FPGA when the top is solid traces and there's no room up there. If >there are a modest number of them, manufacturing adds them by hand. >I've replaced zillions of ceramic caps myself, debugging boards. No >problems.
Is that you method of "debugging boards"? It is a wonder any of your crap ever worked long enough to get sold.
>> >>I have heard that using a pre-heated PCB, and soldering the MLCCs both >>joints at once using hot air is the preferred approach, to avoid damage. > >Overkill.
You're an idiot. You being allowed to live way back when you should have been flushed was underkill.
>> >>Also, that MLCCs with thinner dielectrics are more susceptible, like >>high values in small sizes. > >Most ceramic surfmount caps are about 20 mils thick. Some exotics may >be different.
U-be-dumb.
>> >>Yet in practice, I've never noticed a bad part. Then again, since most >>of them are bypass caps, it's hard to notice a bad part. >> >>Any experiences with soldering causing MLCC damage? > >If you are too rough removing one, you can rip the end cap. Any >reasonable soldering method seems fine.
With seems being the key term to take note of. Good job, Johnny. Good job of showing folks your near nil grasp of the subject.
>>How about board flexing? What amount of flex causes trouble? > >Not in my experience.
He asked for a quantity, not a yea nay, ya dopey ditz.
On Sun, 23 Oct 2011 19:26:29 -0700, TheGlimmerMan
<justaglimmer@thebarattheendoftheuniverse.org> wrote:

>On Sun, 23 Oct 2011 19:03:00 -0700, John Larkin ><jjlarkin@highNOTlandTHIStechnologyPART.com> wrote: > >>On Sun, 23 Oct 2011 10:03:52 -0700, "Mr.CRC" >><crobcBOGUS@REMOVETHISsbcglobal.net> wrote: >> >>>Hi: >>> >>>I frequently assemble PCBs by hand soldering MLCCs of the 0603, 0805, >>>1206, and a few larger ones (and other parts too--as those >>>capacitor-only circuits aren't much fun!). >>> >>>My hand soldering technique is to use 0.015" wire solder, >>>1. first adding a tiny bit of solder to one pad. >>>2. Then I flux both pads, and tack the part to the pad with the added >>>solder. >>>3. Sometimes at this point, I put my tweezers on top of the part to >>>create a downward force, and reflow the tacked joint to make the part >>>seat squarely on the pads. >> >>That can tiddly-wink them into neverland. I grab them with the >>tweezers. >> >>>4. Then I solder the other pad-to-part joint. >>>5. If necessary to make it look nicer, I add flux and reflow the tacked >>>joint. >> >>All OK, except that I don't bother with extra flux. If I sometimes >>wind up with a big ball of solder on one end, I just wick a little of >>it away. > > That is a thermal no no. You should apply the right amount or LESS, >never so damned much that you have to remove some. > >>>The question is this: How much should I worry about thermal stress >>>cracking or otherwise damaging the MLCC dielectrics? >> >>Not much. Surface mount ceramic caps are, well, made of ceramic. > > No shit, dip tracy. > >> They >>are tough. > > You are an idiot. The ends are terminated to the termination platings >precariously at best. Thermal shocks cause detachments, and most of >those cannot be detected with simple test instruments. > >>We often put a few bypass caps on the bottom of a board, like around >>an FPGA when the top is solid traces and there's no room up there. If >>there are a modest number of them, manufacturing adds them by hand. >>I've replaced zillions of ceramic caps myself, debugging boards. No >>problems. > > Is that you method of "debugging boards"? It is a wonder any of your >crap ever worked long enough to get sold.
ftp://jjlarkin.lmi.net/Cymer_award.jpg ftp://jjlarkin.lmi.net/NIF3.jpg Actually, my favorite is a thank-you letter from the Skunk Works. I have it around here somewhere. Our field failure rates are healthy multiples of Bellcore calculations. Ceramic cap failures are rare. John