Hand-soldered SMD caps thermal and other issues

[abit]

I have multiple questions, thanks for reading!

I read a good number of literature about effects on the caps' aging and re-conditioning which I am not fully understanding. I suspect that a proper reflow takes care of most of those issues, but I am about to hand-solder many 1210 caps in parallel, and I am not sure what are the DO NOT DO THIS, or CERTAINLY DO THAT..

I have a fixture where I can upright these caps between some 30, 60 and 100 at a time, lay a bare copper 30 or 24 GA wire across their terminals and solder them in one move. Then turn the string around and connect the other terminals the same way.

First of all, should I leave air-gap between each cap, or solder them as close as they get to each other - at 10V max is there a chance for arching between terminals in the string?

Then, if I use the lowest quality ROHS with little silver and such, would I need to use some sort of post-heating as some datasheets explains? Now, this might be a really stupid question: will a toaster oven, preheated to an 'exact' temperature will do?

Or, should I just forget about it, because the caps are really just to store energy - they are not for filters of any kind, and the idea is to restore or maintain their initial rated storage capacity - no audio or other high-frequencies matter...

Should i just solder stress-free joints the best way I can, and forget about all else?

Thanks!

[Rerouter]

The fact your using ROHS solder means this may be harder than normal as lead free needs higher temps,

The main things your going to face are mainly from the layout stage of the pcb, as opposed to the soldering, e.g. how is the board secured, are components mounted over a flex / sag point, could the weight of a component flex the board, etc, as the main issues faced with ceramic caps come from thermal and force cycling,

For thermal cycling, say you had one side with a flooded ground plain, and the other side loosely placed 10 thou traces, the board would change between concave and convex slightly as the ground plane expanded and contracted, so the more towards the center of this interaction, the less stress on your components.

For force cycling, are there any board mount connections that are likely to have a large number of connection cycles, if so is the board supported so it will not flex if someone doesn't insert / remove it dead straight, etc




For component spacing, when parallel, i prefer to leave a 10 thou gap between parts, to leave wiggle room (say 1 cap fails, and its pinned in place by its 2 neighbors, which are pinned in place by there neighbors, etc), when series, I've gotten into the habit of leaving a gap large enough to easily fit a solder iron tip between, it may not be needed for re-flow, but it has been a nice rule of thumb that i have worked off for a while now,


For heating with a toaster oven, you will have hot spots, your massive rows of caps will take time to heat up meaning everything else will cook before they have re-flowed, meaning a longer pre-heating period may be required,

If you want to maintain there original capacity, start by getting the right dielectric, as some are plain horrible, then worry about what might crack them off the board, and then you can worry if your solder fillet was to NASA standards (this sounds like a joke, its not, its a very good guide)

Also just because its bulk power doesn't mean its not KHz /  MHz pulses, and frequencies, it ties in with decoupling, but even if your just turning on and off an led, a mechanical switch can easily create a current pulse at a few Mhz, as can a regulator or transistor,

[abit]

Thanks!

I was thinking to route a 0.61" PCB and drop the 0.61" height string of cap bodies into the rout, and leave long-enough soft thin copper leads to pick up any stress. I was even considering enclosing both sides of the PCB in some sealant around the assembly area, since ceramics do not need venting and since it is for outdoors as well...

I do not worry about failed caps, the bank will be replaced entirely - as of today I am still a lousy solderer anyways, I need to get better at it, or I am going to ruin a lot of those caps.

I am actually trained on most IPC standards and up to level III, but I do not per see work directly along those lines so I do not fully appreciate them This is a learning project of mine@home.

Using a toaster oven is then out of question.

The IC is going to have a decoupling cap at input, and that is going to be the one suggested my the manufacturer of the IC. The MHz-noise thing I do not fully understand, but again, it all depends on the IC I am going to use.

I was more worry about sparking and arching from terminal surfaces not joined by solder but very close to each other.
I was also worried that the EM field by the caps somehow require a minimal air-gap when stacked - apparently not so.

The ROHS solder is a worst case scenario, as I do not want to use any leaded stuff - much of what I build probably end up in the recycling centers anyways, its a personal preference thing. Thanks again!

Update:

Of course I meant 0.061" and not 0.61".

[IanB]

The ROHS solder is a worst case scenario, as I do not want to use any leaded stuff - much of what I build probably end up in the recycling centers anyways, its a personal preference thing. Thanks again!

Except that lead is really not dangerous. It comes out of the ground and it goes back into the ground. Heck, they used to make water pipes out of the stuff!

[abit]

I do not argue along those lines - working with ROHS is certainly more challenging and more in demand, and I would like to gain some hands-on experience with it. As I said, it is a worst-case scenario. If I keep burning my caps with 270 I am going to drop back to Pb, just to get my project working and off the ground. ROHS or not, I return all my electronics junk to a recycling center and let those entities decide what is what and what to do with them

[Rerouter]

I think i would need some kind of reference by what you mean by the 0.61" pcb, a 1210 package pads would be about 0.2" wide, meaning about 0.2" on either end of the cap to the edges?

For sealant, as long as its not in contact with the capacitors you should be OK, as most silicon sealants tend to shear parts off pcbs over time,

if you plan on only doing small batches, look into a cheap hot air station, in aus there are clones for $50 that work lovely for this kind of stuff, for larger batches, (100+) you may be better talking to itead or similar to produce and assemble for you, (i have no idea on there quality of assembly, but like there boards)

As for the IPC stuff, i myself have not been professionally trained in it, but after 4 years of re-work on smd boards, designing my own,  and much reading i think i have at least a firm grasp in where things go wrong,

 Terminals arcing can still be a problem with lead free, but not suddenly, its called tin whisker formation, otherwise a 0.2mm separation gap that is not filled with solder should be enough for less than 20V, but plan wider if possible and you will minimize the chances of tin whiskers,

EM will be contained if there is a reference plane underneath it and its conductors for the most part, unless you mean cross-talk which is another thing again, i'll have to dig it back up, as i was reading over the TI? reference guide on this earlier to confirm the opposite (i had a possible issue with noise getting in)

The noise i was refering to before is part of decoupling, a 1nF cap right across the pins of a micro slows down the rate of the edge on the changing current, from there the change in current propagates along back to your bulk storage caps, which in turn slows down the change in current back to your supply, the noise is still there, but at a lower amplitude and frequency, from the micro could be harmonics up to a Ghz, after the local decoupling, something like 80Khz, and after the bulk storage something like 300Hz (for equivalent sine wave edge rate)

If your concerned about EM you need to be specific in what you are trying to achieve, e.g. a screen grab of the part of the pcb, etc. or a scribble in paint,

I'm willing to help, but there is only so much i can generalise,

[Rerouter]

Burning 1210 caps on 270 degrees.. either your temp control is way off, your using non-fluxed solder, or your not pre-tinning the joint, even with lead free you should still be able to re-flow a joint well before you burn the package (I could likely hold my iron at 330 against a cap for 40 + seconds before it seriously discoloured),

Sounds like a hot air rework station is for you then.

[abit]

Rerouter - Thanks, nearly everything you wrote I never considered or never heard of before.

Hot air - I need to pay closer attention to people using it at work, I never considered it before.

There is nothing I can do about 'tin whiskers' until I actually solder the cap bank and see it happening - as you suggest it depends on the solder, fluxing, and more. For now let's assume there are no gaps between the caps, and all terminal surfaces are fully soldered.

But the 'reference plain' dealing with EM - you mean a ground plane? As I am planning, I am going to route a 0.061" PCB (I know, it sounds insanely bulky) to some 0.068" width, to the length of the capacitor bank, and the two soft copperleads from that bank going to be about 0.25" long, each entering the PCB on different sides, and the 1210 cap bodies are sunk into the route. That is, the rather heavy PCB prevents any warping, protects the cap bank, and still ends up relatively flat - some 0.05" for the copper leads and fillets on each side - 0.071" thick. Now, the maximum number of banks is going to be 10 - and the PCB is happen to be the same size as the 0.031" PV PCB. How is this going to affect thermal or force cycling, EM fields, and is this a good idea?

I am not sure if this makes sense, I would love to provide a picture but I am not even that far 'designing' this thing yet. Thanks for your time!

[Rerouter]

I really think i am failing to understand you on the pcb, atleast im my head, those dimensions would translate to this PCB, if you built such a monster, thermal and force issues would be ruled out, but who you could get to build such a thing is lost on me,

By reference plane i did mean a ground plaine in your context (a plane doesnt have to be at ground, and for EMI issues it only has to be low impedance)

For tin whiskers, you probably wont see it happen, there finer than human hairs in most cases, and takes between months and years to form, was just giving you a run down, it comes solely down to there being tin in the solder with no lead to weaken the chains and prevent the hairs from growing,

Sketch up a rough concept of what your planning if not what i drew up and work from there,

[T3sl4co1l]

I'd be interested to see the whole circuit, or even as much of the system design as you have right now (I'm assuming from your other thread that you're working on many aspects of this at once?).

Ceramic caps are rarely the best method for bulk energy storage, for a variety of reasons; and that it's for PV sounds even more bizarre to me.

Tim

[abit]

Rerouter - Here it the picture of the cap bank, I have a hard time making graphics because I am really terrible with graphics programs, and Google Draw just do not work under vbox as expected - and I rarely go anywhere outside of vbox...

T3sl4co1l -- Ceramic caps TODAY are rarely the best storage options, yes. In five years this choice might not be so odd, when I have a design ready - and you do not. Still find it bizarre?

[wraper]

Still find it bizarre?

I find it 
BTW as I see it, capacitors closest to the sides will experience huge mechanical stress. And overall, quiet likely there will be more stress than just soldering them on the PCB.

[abit]

Would more extended leads of the copper wire alleviate the stress? Or some other way? Is this solvable? Thanks!

[wraper]

Extended leads = flexing/bending = stress on the caps located on the sides. Basically what happens is that 1 cap on the each side holds all of the weight of the whole construction. This is viable in some way only if capacitors are held is some another way and leads are very thin.

[wraper]

Also using ceramic caps as energy storage is a bad idea not only because of the big size. But also because of the huge capacitance change with voltage applied. It can be as high as >80% capacitance decrease at rated voltage for some X7R capacitors and even worse for Y5V, and this besides temperature variation. Also, if you change the temperature of the charged capacitor, voltage will change with changing capacitance too. Heating -> capacitance decrease -> voltage rises. Cooling -> capacitance rises-> voltage drops.

[abit]

wrapper - Thanks! This is something I was vaguely familiar with, and I am still not understanding it entirely - but I know that no other caps (including double-layers and so called super-caps) do any better. Again, some way some day some hybrids might come along, or going to be purely EM-based as I suspect. I am also keeping an eye out for LiFePo - but the reality of that is pathetic at best. I might give up on ceramics and go with chemical or another, but what I learn here and now is going to be useful. Thanks again!

[wraper]

I'm not familiar with supercapacitors but never heard about them suffering from such negative effects. What do other capacitors do not do any better? All other types are basically free from capacitance change with temperature/voltage. IMO just take some quality electrolytic capacitor, that would much better fit the purpose. Comparing LiFePo and supercapacitors with ceramic capacitors is just plain stupid. They are in (orders of magnitude) completely different leagues.

[Rerouter]

Now as you kind of pulled agro a bit before i'm going to skip it and let you figure it out yourself, here is how you could layout your board, you would solder your busbar either on the top or the bottom, and you could pack about 2.8 times as many caps per side by having them vertical, so about 6 times as many caps by having both sides packed vertically,

So assuming you went and used the largest capacitance i can find in that package, you would have 56x 220uF, so a 12320 uF (or 0.12F) cap rated at 6V, (but would drop in capacity as it approached that voltage), with an extemely low ESR of something in the micro ohms, in a size of 26mm x 9mm,

I'm not going to shoot it down, as there are places in the world such a circuit could be used, just not sure on solar,

[wraper]

12320 uF (or 0.12F)

0.012F

[wraper]

When using highest possible capacitance in particular package, that likely would mean something like 80% capacitance drop for X7R or more for Y5V (need to refer to the data for particular capacitor model). So 12000uF 6V capacitor bank could easily end up like 1500-3000uF if charged up to 5V, And this is even without heating the thing. If heating Y5V to something like 70^oC, this likely will end up being less than 1000uF.

[T3sl4co1l]

Ceramic actually do quite well on energy storage, but you have to look for high voltage C0G.  Big, big bucks, and that's if they survive soldering and vibration.

Not sure what "five years" means; I'd be surprised if the suggested construction survives that long in an outdoors environment.

Ordinary type II dielectrics (X7R, Z5U, etc.) are about a third the energy storage of electrolytic, and only a little bit faster.

Also, so much for "please, shoot it full of holes", I guess you actually don't want your baby improved, huh?...

Tim

[abit]

Reading through I begin to realize WHY I am NOT going to be an electrical engineer. This is insanity.

Tim,
“high voltage C0G.  Big, big bucks, and that's if they survive soldering and vibration.”

Noted. i just started reading datasheets, it is going to take me a while more to understand them.

As you duly noted, ‘if they survive’ - and that is a BIG IF with hand-soldered caps, so I opted for an 8000 reel of 1206 1.6x1.6 mm just as an experiment. I truly do not care if the whole 8000 must be scrapped, but I need to get out of SPICE world into real-world. Wish me good luck!

I read the entire thread, thanks, see you at Projects! None more insane here than I am, thinking of the invention of the wheel, gears, worm-gears and the gyroscope. Oh -- LED indicators turn into commonwealth illumination, caps turn into power/energy storage? Bizarre?

[Rerouter]

The real world is bigger than any of us, but happy to help you get your footing, I'll be blunt and say that pick any one subject and if you get deep enough into it it will start sharing a large number of common threads with many other fields, helping you get your footing in them, and so on,

If you want to play with the real world, material science, and TI layout and design guides are some of the best places to get a firm grasp on how things can go wrong and how to make things last,

[abit]

Thanks Rerouter, read you at Projects...

[T3sl4co1l]

Indeed, the richness of the world beyond a certain depth of study is a rich and wonderful thing.  Learning differential equations (and all that entails -- calculus, curves, graphs, geometry, polynomials, and yes, linear algebra as well -- matrices and eigenvalues) and wave mechanics (most often E&M and quantum physics) will take quite some time to study (these encompass the better part of a full undergrad degree, and maybe a bit more than just one!), but is highly informative of a great many practical and theoretical subjects.

The Fourier transform alone appears naturally in everything from QM (solutions of "matter waves" are often done in the frequency domain, or more specifically, momentum or k-space), to signal processing and information theory.  The duality between time and frequency, or momentum and position (Heisenberg Uncertainty Principle), is a direct consequence.

You can come up with a great many interesting and fantastical ideas, but if they aren't grounded in, and informed by, these fundamental aspects of the world as we know it, you'll have a hard time (see: all those Kickstarters that attempt to scoff at these laws!).

Tim