Soldering surface mount components

I have yet to see an area where Y5V caps are actually worth using.
Absence of proof is not however proof of absence; I assume there are in
fact very good reasons why you chose Y5V, and was kind of hoping you
would share those reasons, so that I and others could learn from your
experience.

I am however constantly surprised by how few engineers seem to know
about the voltage characteristics of Y5V, or for that matter the
temperature characteristics - odd when you consider most manufacturers
supply that information (although TDK seems to want to hide the voltage
coefficient, Philips (or whatever they are now called), AVX and others
do not).

**YOU** brought up the piezoelectric effect, then incorrectly defined
it, and erroneously said X7R and Y5V are not piezoelectric.

I'll concede I'm a pedant

(but the resonance was enough of a red herring to warrant correction)

I originally stated:
"AFAICR the piezoelectric behaviour was not the issue - just
electrically exciting them at their mechanical resonant frequencies."

I am merely asserting that forces act on capacitors, regardless of any
piezoelectric effect. I had a look, but cant dig up the Marcon paper; I
suspect that you are probably right about the piezoelectric effect being
the predominant issue though.

Nor do I, except they provide a fairly graphic demonstration that forces
act on capacitors.

Not at all. I greatly enjoy technical discussions, I often learn a lot
from them - for example I didnt really know much at all about
microphonics until following discussions here - in smps use the
piezoelectric effect of caps is pretty much negligible (unless, of
course, your large caps break due to exciting the mechanical resonance).

The OP is of course about soldering smt parts. The TDK app notes here:

http://www.component.tdk.com/ceramic.asp

are quite useful. "Common cracking modes" is particularly pertinent. TDK
state that the main causes of cracking are mechanical damage - impact
damage or pcb flexure. But they also talk about how uneven soldering
causes stress concentrations that exacerbate flexure related cracking
(probably a lot less of an issue on Aluminium cored boards than FR4).
They very briefly (one sentence) mention thermal stress, in-circuit
testers (impact I presume) and H2 absorption as possible failure modes.

Something I did not know about high-K caps was the ageing effect - the
"what is the capacitance of this capacitor" document talks about it.
Another thing due to the use of Barium Titanate. And the ageing is a
reversible process! See, lots of interesting things to learn.

I can share a funny story about pcb flexure - our production dept.
wanted to reduce the build time on a gatedriver pcb (6 per product) so
smt'd it, and panelised about 20 onto an A4 sized FR4 pcb. Rather than
routing the pcb and using breakouts as advised, they V-grooved it and
got the mechanical workshop to produce a guillotine-like device to cut
the individual PCBs out. As the guillotine sliced the PCB it bent
significantly, and all the caps and resistors disappeared off the PCBs,
leaving only their endcaps behind. The really sad part is that
production abandoned the product, rather than simply routing the
panelised PCB. ISTR the original PCB hand assembly cost about the same
as the parts themselves, and the smt pcb was almost half the price.
Adding in the extra money for routing would have added around 10% to the
cost of the smt pcb, so it was still a lot cheaper.

The same guys also built a Jig to insert a pcb into a press-fit plastic
enclosure, by pressing on the top of a 1206 resistor. We had a 100%
failure rate on the first couple of hundred PCBs, but the tech never
mentioned that every single resistor broke in half when he desoldered
them. The real funny bit was there was huge amounts of space, the mech
guys never looked but miraculously picked the one spot there was a
resistor - that Murphy fellow again....


Cheers
Terry

 

Any particular brand of capacitors? Any particular dielectric? Over years
we hand assembled a number of prototypes and small production runs that used
plenty of ceramic capacitors. These included high voltage, tuned and timing
circuits where a partial value failure would show up. At up to 30 power
magnification there were no visible cracks. The dielectric materials varied
from NPO to Y5V. We use Metcal MX with 700˜ tips and organic flux. The
environment is geophysical instrumentation which can be pretty demanding.

We had two component failures directly attributable to hand soldering. One
was stacked film chip capacitor where there was a drastic drop in value due
to internal stack disconnects from the end termination. The other was a
resetable fuse where the end termination cracked. The fuse was still
operational.

Regards,

Boris Mohar

Got Knock? - see:
Viatrack Printed Circuit Designs http://www3.sympatico.ca/borism/

void _-void-_ in the obvious place



Regards,

Boris Mohar

Got Knock? - see:
Viatrack Printed Circuit Designs http://www3.sympatico.ca/borism/

 

I'm sorry that I cannot recall, precisely. It was one of the
Japanese producers. A representative of theirs gave a
presentation to (some of) our engineers (including myself)
where the effects of that abuse where shown in micro-
photographs. From my visual memory of it, I suspect
the magnification was well in excess of 30, and the cracks
were somewhat subtle in appearance because they were
closed and on the corners.

I am sure it was NPO. I expect all the vendors are using
similar ceramics, at least with respect to the mechanical and
thermal properties that relate to thermally induced microcracking.

We were doing Doppler processing where minute phase
or amplitude shifts would be interpreted as meaningful.
The cracked capacitors would create artifacts. These
were not much above the noise floor, so they would
have been 50 to 70 dB below the signal that had been
modulated by the parametric noise to make the artifact.
(This, too, is a demanding application.)

We would see this failure maybe in 1 out of 5000 parts.
Perhaps the parts in your prototypes that had to be very
stable were lucky, or the person doing the soldering was
applying the iron to the pad, which is gentler thermally.

Even after seeing that effect, I hand solder SMD ceramic
capacitors for initial prototypes. But I take care (or have
care taken by the assembler) to not apply the iron to the
endcaps. And the finished article is not fielded. I cannot
consider them reliable because of potential cracking.

 

---
I would think so, especially that 100milliFarad unit you normally use?
---

---
The piezoelectric characteristics of the part are not normally
something which warrants a "warning", since operation within the
specified electrical and mechanical limits for the part will not cause
a failure because of the piezoelectric nature of the dielectric.
Rather it's an inherent characteristic of the device which must be
taken into account when using the part and its contribution "designed
out" if necessary.
---

---
????? I don't believe I said that it did.
---

---
Yeah, I see you have a remarkable grasp of the obvious, but so what?

I mean, what point are you trying to make?

You've already demonstrated that you know nothing about the
manufacture of ceramic dielectric capacitors as evidenced by your
ignorance of the fact that barium titanate is a commonly used
dielectric, yet here you are throwing around platitudes as if they
were rare gems. In all fairness though, perhaps they are... to you.
---

 

Ditto. And I *NEVER* re-use ceramic smt parts. I just buy values 1,000
at a time. But me and my METCAL MX2 (and my binocular microscope) have
soldered many, many thousands of 0603 parts (when a prototype uses 1400
parts, and you re-build it 5 times, thats a lot of dead parts) over the
last month or so, and no parts have failed (yet). But of course it aint
going to a customer.

I mostly break 0603 parts by moving the tweezers after soldering one end
only; both Cs and Rs are susceptible to this, but its fairly obvious
when you do it. perhaps 1% break like this.

1206 quad pack resistors are a pain to solder, especially when I lay out
my pcb so closely the tweezers are obstructed Self inflicted, so no
sympathy I guess

Cheers
Terry

 

No, I said they were not damaged by a piezoelectric effect while hand
soldering.

I only miss quoted the definition. So what?

As I said:

You are also something of a pest. I got nothing from this except going over
very old ground which was not the topic.

 

I once had a knarly argument with a guy who thought I
was irresponsible for using a high K capacitor in a circuit
whose accuracy relied on charge balancing rather than
the particular capacitance or even its linearity. He had
adopted a rule resembling "Y5U's are crap". (I think he
may have eschewed them even for bypass applications.)

[snip]

Methinks that common ceramic capacitors are not
actually piezoelectric. The forces that occur with
applied voltage are different in character from what
is defined as piezoelectric effect. Consider that the
dielectric of a charged capacitor is compressed
regardless of the polarity of the charge. This differs
from what happens with (properly poled) piezoelectric
ceramics, where one polarity compresses and the other
decompresses. For similar reasons, a capacitor with
no charge on it will not transduce mechanical strain
into an electrical output, while a piezoelectric device
will transduce without an ostensible external charge.
(The poling produces a situation similar to applied
charge, but it is not available at the terminals unless
the device temperature is elevated.) So, to sum up,
a material that *can be used* to obtain piezoelectric
effects (when poled) is not necessarily piezoelectric.
Commonly sold ceramic caps are not piezoelectric.

Regards,

 

No you didnt - read your own posts. The discussion was about reliability
of hand soldering smt caps. I pointed out that mechanical resonance
related issues can also arise with large smt caps (although I omitted
the word "issues"). You come back with:

"Like the 2.2mF to 100mF units I normally use?
Never seen any warnings. They are not piezoelectric."

when in fact you meant 2.2uF - 100uF that ARE piezoelectric. mF instead
of uF indicates vintage, the "not piezoelectric" indicates lack of
knowledge re. Barium Titanate, of which you later state:

"However AFAIK piezoelectric quartz, barium titanate,
or other piezoelectric materials are NOT routinely used in the
production of capacitors."

So I go to the TDK website, click on ceramic caps and what pops up - PDF
documents stating they are made from Barium Titanate. Have you even
looked at the manufacturers datasheets for the components you use?!

Your knowledge is clearly, demonstrably incorrect, as evinced by your
own statements.

Without any accompanying verbiage it is reasonable to assume you are
blaming the mechanical resonance problems on the piezoelectric behaviour
of the caps - so a couple of people point out that forces act on all
capacitors, and that X7R, Y5V, Z5U are piezoelectric.

by "miss quoted" (sic) you actually mean "got it completely, totally and
utterly wrong, whilst simultaneously missing the point entirely"

I presume the pesky parts relate to pointing out technical errors. Dont
be so hard on yourself - everyone makes mistakes.

I find it very interesting that you never seem to respond to technical
comments - for example why use Y5V caps. Although my X7R argument is
sufficiently persuasive (and demonstrably factual) that I am not
surprised you cannot refute it.

It would also appear that your grasp of written english is less than
profound. We could all take lessons from John Popelish - his
pump-and-dump LED pulser paragraph was a marvel of clarity and brevity,
and contains all of the pertinent technical data.


Cheers
Terry

 

Hi Larry,

To be honest I just avoid Z5U/Y5V (and in fact seek them out for
criticism/removal during design reviews), but in theory I would be quite
happy with using one in a charge-balancing circuit. In practice I would
be concerned about having them on the shelf, because some bright spark
in production is bound to use them in the wrong place (hey, lets put
them in this active filter...).

And when it comes to bypass applications, I still havent seen a valid
reason for using them - by the time you work out what the actual
capacitance is under DC bias and operating temperature range, its so low
that you can invariably get an X7R cap for the same price that has about
the same (if not much more) overall capacitance, and its usually the DC
bias that is the strongest factor - who runs caps at 10% of rated
voltage? thats just wasteful.

But, if you have a good reason, I'd love to hear it.

Something to think about. Thanks.

Cheers
Terry

 

I think, you think wrongly here. Ceramic capacitors with high K materials
are often enough piezoelectric to make them a bad idea for small signal
work. They make a voltage if you thump on them. This makes your circuit
into a bad microphone.

 

This is where it gets interesting. Testing the microphonics is
relatively straightforward, but would need to be done for a variety of
DC bias conditions and temperatures, and of course differing dielectrics.

How to test the forces generated though...and the mechanical resonant
frequency (that might be easier by calculation given the fairly simple
geometry) of the device under consideration. What about some form of
model using relatively easily measured data to predict the mechanical
effects?

Cheers
Terry

 

... or boring depending on your point of view. Or this may even a way to
start another flame war on the subject.

It is easier to put in NPO capacitors or design the need for capacitors
out than it is to ensure that the high K ones are not microphonic. The
high K capacitors I had trouble with had no bias voltage on them and were
microphonic at room temperature. I could get a few tens of microvolts
signal from them by dragging my fingernail over the PCB near them.

Even if we tested 100 capacitors of a certain type, we could not really be
sure. What would happen if the next batch were made during a thunderstorm
or just after the crew ate lunch?

I think the problem will be the loading of the capacitor. Ceramics have a
high "Q" mechanically. It is likely that the PCB loads the mechanical
vibration enough to be the controlling factor in the amplitude of the
stress.

An interesting thought: Tin-lead solder is very lossy mechanically. This
no no-lead stuff may not be as lossy. The no-lead folks may have created
another problem.

 

If you define "piezoelectric" loosely enough, then your "bad
microphone" is relying on piezoelectric effect. By such a
loose definition, even vacuum is piezoelectric. But by the
more discriminating definition Terry quoted, and according
to the usage among people who exploit the piezoelectric
effect, ordinary interactions among separated charges, such
as occur in a charged capacitor when thumped, should not
be deemed "piezoelectric".

What do you think happens when you thump a capacitor
that is not biased? If it was piezoelectric, (and read Terry's
quoted definition carefully), it would produce an electrical
signal. But it does not.

I do not deny that there are electric/mechanical interactions
in capacitors, or that high K capacitors can be misused.
But nothing you have said goes to show that I am wrong.

 

Absolutely. Although for most of the stuff I do its the tempco and DC
bias that cause the biggest problems. I use a fair bit of X7R, but avoid
Z5U/Y5V, and recently re-worked an actve filter design to drop an X7R
cap down to an NPO.

It probably depends on the rate of change of force, too. Thanks for the
data though - sounds like (sorry, couldnt resist) mV signals ought to be
quite feasible..

yes. I am constantly impressed at how damn hard mechanical engineering
really is (if you do it properly). Electronics is often easy by
comparison, and usually devolves into mechanics anyway (thermal etc).

doh. This sort of metallurgy is way beyond me - but I guess it boils
down to the malleability of the resultant alloy?

Cheers
Terry

 

I'm gonna get a steak and cheese pie, then whack up a little test
circuit - inverting amp, cap from -ve i/p to 0V, 100k feedback and bias
comp resistors, +/-5V supplies, and do a few tests. I have some nice
10uF X7R caps, and some 1nF NPOs. I think I have a few Y5Vs too...

anything wrong with this setup? other than the fact I amp applying a
VERY small bias to the cap, around the offset voltage of the opamp - in
this case about 4mV. How can I reduce this further?


And the corollary here is of course thumping a biased cap WILL produce a
signal, as the pressure wave propagates through the cap there will be
deflection of the plates, therefore VdC/dt current must flow. Stiffer
caps have smaller dC/dt, so generate smaller signals - eg film or npo
caps versus air caps (eg trimmers) or electrolytics (there are probably
other things going on inside electrolytics, but I imagine the pressure
wave will actually move the electrolyte much, much more than the atoms
of say NPO will move when thumped).

Cheers
Terry

 

It depends on the operating environment. Does your product have to detect
mV signals while being dragged down a dirt road, from the back of a pick
up truck? If not, your life may be easy.


[...]

I have met at least 5 people who's busines card said "mechanical
engineer". As far as I can remember, I have actually only met one
mechanical engineer and that is not what his business card reads. The
others were mechanical dunder-heads. I've very often ended up doing my
own mechanical design. It is not the fastest way to get the job done.

 

True but I was not defining it loosely as indicated below.

[...]

As stated elsewhere, I got tens of uV of signal from a non-biased
capacitor by dragging my fingernail along the PCB material near it.

How about now?

 

2SK170
30mH 0.1u !--- ect
--------+----------- -------->!
( ! )( !--- ect
( --- )(
( --- )(
( ! )(
--------+----!!----- ----------- ect
C2 SP-4
Triad
1K:200K


The 30mH has a SRF above 10KHz, is air cored, and has a resistance of
about 13 ohms. It is made from 2 15mH coils pointing in opposite and
matched so that their distance source pickup is under 1 turn on a 3
sqr-inch area worth. The coil is then placed in a 3 layer shield can.

 

I think the problem with mechanics is that it initially is more
accessible than electronics. By that I mean it is easy to get a handle
on say torque - lean on a bar, whereas getting a handle on an electron
is a bit trickier - you cant see it, feel it etc. most of the
electronics people I meet have a fair understanding of second order
systems, stuff like that. but few so-called mechanical engineers do. I
have been fortunate enough to work with some brilliant mechanical
engineers, people who are every bit as clever and creative as the
smartest electronics guys I know (some guys I very briefly worked with
at Penn State were incredible), but most seem a bit thick. Hell, try
getting a sheet metal shop to fold up a box accurately.

Then when you really get into it, mechanics is a lot more complex than
(most) electronics - nothing is isotropic, or homogeneous, or perhaps
even well characterised. Everything is as non-linear as all hell, and
the measurements are a lot harder. I suspect all the clever mech
engineers go work on the really tricky stuff, and leave the rest of the
work to the metal-shop dropouts.

Cheers
Terry

 

Yes I agree. People should be able to get a feel for the basic lever
problem just from living in the universe. You don't see electrons in the
play yard much.

[...]

Or put the PEMs in the right way.

And the units of measure are all weird. "Shore D" indeed.

I think I'll use a 3/8th cable over a 10 CM pully.