Connect with us

Soldering surface mount components

Discussion in 'Electronic Basics' started by Daniel Kelly \(AKA Jack\), Jun 8, 2004.

Scroll to continue with content
  1. Clarence

    Clarence Guest

    Of course this may have actually happened, and Boy, you had some pretty lousy
    assembly people. I've seen the pre-prod units used for test assembled and
    soldered by hand and subjected to extensive testing. Never saw a solder
    related failure of a component. We tested for very long periods on many
    boards. Of course we also inspected the boards before applying power and
    checking for damage. Rarely had to retouch a board after the first three.
  2. Terry Given

    Terry Given Guest

    This is a VERY good point. Reflow ovens have very well controlled
    thermal profiles, slowly ramping temperature to a plateau, holding,
    slowly ramping up to final tmep, holding etc. Mostly to avoid this
    thermal shock related mechanical failure mechanism. High voltage
    ceramics are especially prone to this - hand soldering them is a risky

    I once used 2 x 15nF 1000V smt X7R caps in series across an 80-800Vdc
    supply for a smps application. During testing one smps failed
    catastrophically (two others ran fine). Detailed examination of the
    corpse showed a blast pattern radiating outward from one of the caps,
    which had ruptured. The resulting mess sprayed directly across the legs
    of one of the FETs, thereby toasting the unit. At the time it was
    operating at a DC bus voltage of around 400V, so the cap was nowhere
    near its rated voltage, more like 20%. One of the guys I worked with had
    extensive experience in this area (hi-rel smps hybrids for
    il/aerospace), and showed us what went wrong. We immediately replaced
    the capacitors, carefully using a manual hot air station, to both
    preheat and solder. The units operated continuously into a dead short at
    800Vdc, no problems - there were other issues of course, it was a
    pre-production protoype, but none of the explosive kind.

  3. John Larkin

    John Larkin Guest

    Maybe the parts are getting better, but I've never seen that happen,
    and some of our boards have lots of parts on the bottom,
    hand-soldered. I hand-solder all kinds of parts in the lab, and can't
    recall ever damaging one, unless it wasn't on a PCB, like soldered
    directly to a connector or on a copperclad breadboard, where it is
    possible to apply some bending forces and rip off the end caps.

    We have virtually zero returns from the field due to damaged passives.

  4. Howard Long

    Howard Long Guest

    "Steven McGahey" < (remove
    the obvious bits)> wrote in message
    have > to use a different approach, and try to keep the component cool (as
    FWIW, I use a standard temperature controlled iron with a small tip (I think
    it cost about $50), with very thin solder, desolder braid.

    I do everyting on a plain white tray with lips around the edge - there's
    nothing worse than losing that last 3.3k resistor you had on the carpet...

    I do have a small magnifying glass, but that's just to help me identify the
    components that actually have markings - I don't use it for anything else.

    For the chips with loads of pins at tiny spacing (including 0.5mm), I use a
    really tiny piece of bluetak (a bit like plasticine or playdoh, but sticks
    paper to walls) to fix the component into place, then I solder the
    component. I don't worry too much about solder bridges over the leads for
    now, but I am rather careful to use small amounts of solder, in case it
    creeps under the chip.

    Then I use the desolder braid to mop up the excess solder. After a visual
    inspection, I do a continuity check of each of the leads to ensure there's
    connection to the pcb as well as no shorts between adjacent leads of the
    chip. A pain, and almost every time it shows nothing untoward.

    Rework involving removal of chips takes a number of tricks, but primarily
    remember that you're much more interested in maintaining the PCB in good
    shape, at the expense of a trashed component.

    Good luck. Howard.
  5. Yes, of course.
    They were quite skilled and competent, generally. If you knew
    the circumstances under which the hand soldering occured, you
    might not be so willing to denigrate them. (But who knows?)
    To see the excess noise phenomenon, you would have to be
    looking at a circuit handling low level signals which would be
    affected by random parametric shifts. To see the drop in
    voltage withstand, you would have to be using parts at an
    appreciable fraction of their rated voltage, or subject them
    to conditions under which moisture would enter the cracks.
    So the fact that you never saw that is not much reassurance.
    But what were you testing for? Did the environment
    promote moisture ingression into the cracks? Was there
    thermal cycling? I must say, your failure to see that
    phenomenon is weak evidence against its reality.
    The damage I mentioned is nearly impossible to see without
    a microscope. Typically, the micro-cracks do not extend
    clear thru the part, and they tend to be closed, being held
    together by the unbroken material. I doubt your inspection
    would have caught that damage.

    The facts I have related regarding the failure mechansim,
    and the strong disrecommendation against hand soldering
    ceramic SMD capacitors, came to me directly from a well
    known and reputable supplier of such parts. You, or other
    "we got away with something, so it must be fine" kind of
    folks can disregard it and often not pay the price. Those
    who desire reliability will more likely heed it.
  6. Terry Given

    Terry Given Guest

    its the rate-of-change of temperature thats the real killer. Larger caps
    are worse, as the resulting dimensional changes are bigger.

    doesnt everyone have a binocular microscope? how quaint.
    There are also mechanical resonance related with the larger ceramic smt
    parts (Marcon have poublished several papers on this effect).

    Recently I have hand-soldered about 2000 0603 caps (prototypes). Perhaps
    2-3 caps failed immediately; as its a prorotype I dont care about
    medium-long term reliability, but no way would I give it to a customer :)

  7. Clarence

    Clarence Guest

    Not much moisture in a near vacuum!
    I'll take the 'disrecommendation' with a box of salt!
    I am delighted to hear you wouldn't give your work to a customer. Many
    prototypes are unsuited for the customer to see anyway due to the rework and
    handling in engineering test. I NEVER ship a 'prototype' to anyone. That is
    what a "first Article" is for!

    As for the work I cited! These were weather Satellite boards, RF, motor
    control, CPU, and digital communications, plus low level analog video, with
    analog to digital conversion. There will only be 18 final units built, and
    testing (with temperature cycling from -40 to + 80 Degrees C four times a day
    at 5 degrees C per minute) was eight times a day, total time of a complete test
    was 2,000 hours. MTTF predicted is 18 years. Also they must survive 50,000
    Kilorads exposure.

    Yes, Inspection under a microscope, 20 and 50 diameters magnification. Before
    and after tests. Yes All boards were also tested on a shake table, they must
    survive launch.

    The Customer is NASA, they are very particular, and will launch the first of
    these in 2006.

    Shooting one's mouth off when someone tries to help causes a loss of
    Since your really an amateur, live with your poor workmanship and cry about it!
  8. Jim Adney

    Jim Adney Guest

    I was on the Vishay web site recently and came across a writeup they
    have on hand soldering surface mount electrolytics. If you do any of
    this I believe it's worth looking up. They are rather cautious and
    suggest that if you spend more than 3 seconds on a junction that's too
    long and you should start over with a new part.

    Seems extreme to me, but they're the ones who are actually in a
    position to know.

  9. Terry Given

    Terry Given Guest

    Hi Clarence,
    Hear Hear!

    Reflow machines are best suited to soldering smt. If you have to do it
    by hand, use hot air. A bloody great soldering is the worst way.
    What makes you think I'm an amateur? I just dont want to hire a tech
    (labour laws become a real pain in the ass when you have staff).

  10. Roger, John, Howard and Jim,

    Thanks for your response, guys - much appreciated.

    I'll give this a go with the blutack (we had the same thing in Australia -
    wonder what the UK equivalent is...) and desoldering braid.

    I think I also need a smaller soldering iron tip - the last one was still
    too big and bulky with a 1 or 2 mm point.

  11. Clarence

    Clarence Guest

    Since that is what they were designed for, and that is what I use them for when
    it is appropriate.

    By the way, all these boards MAY be hand soldered in the limited production.
    There are components which can be reliably flow soldered, but these boards are
    populated on both sides, and depending upon weight MAY not remain in place
    going through the reflow process with an already soldered side down. Ordinary
    FR4 will usually work this way, but the aluminum cored boards get too hot on
    the bottom when the core conducts the heat through the board. (There is no
    "convection cooling" in space. Only radiation and conduction.)
    Your statements were a strong clue, then the lack of experience added fuel. I
    would hope I was wrong! I only maintain five consultants (1099) on call, all
    specialists! I do the initial design, and work along side of experienced
    specialists for a quality result.

    Sometimes I take my entire team into a customers facility, it helps to have
    people with all those training certifications and credentials.

    Thanks for the warning about Macon. I'll avoid them!
  12. Paul Burke

    Paul Burke Guest

  13. Howard Long

    Howard Long Guest

    I'll give this a go with the blutack (we had the same thing in Australia -
    It's the tiniest piece of blutak - too much and the chip's leads don't rest
    in contact with the PCB solder pads. This allows you to position the chip
    accurately by sight over the solder pads. The blutak then remains there ad
    infinitum (or until you have to rework the chip!).

    In the US blutak's like this
    Here's the iron I use

    I use the same sized tip that was supplied with the iron - it's pointy but
    not miniscule.

    Cheers, Howard
  14. Terry Given

    Terry Given Guest

    Actually its not Marcon per se, its just that the larger ceramic caps
    have mechanical resonances that can be excited electrically. square-loop
    ferrite has the same problem - witness the warnings in the Ferroxcube

  15. Clarence

    Clarence Guest

    "Terry Given" wrote
    Like the 2.2mF to 100mF units I normally use?

    Never seen any warnings. They are not piezoelectric.
  16. John Fields

    John Fields Guest

    Pretty large value for ceramics, eh?
    It's not necessary for a capacitor to be piezoelectric to exhibit a
    mechanical resonance or to be microphonic. All that's required is for
    the dielectric to be mechanically deformable by the forces exerted by
    the electric field across it or for the dielectric to be deformed by
    external mechanical forces.
  17. Terry Given

    Terry Given Guest

    you use 0.1 farad ceramic caps? those I'd like to see. Whats the
    dielectric? (or perhaps you use "mF" to mean micro-Farads - quaint but
    confusing, given the preponderance of SI units nowadays). I did once see
    a 100uF 200V (IIRC) NPO cap (mil smps). very very expensive - US$300 IIRC.

    AFAICR the piezoelectric behaviour was not the issue - just electrically
    exciting them at their mechanical resonant frequencies.

  18. Terry Given

    Terry Given Guest

    And if the mechanical resonant frequency is the same as the electrical
    excitation, significant (wrt the cap) forces can build up over time.
    2220 and bigger were noted in the Marcon paper.

  19. Clarence

    Clarence Guest

    C4532X7R2A225M By TDK 2.2uF 100V (since your into SI (stupid interference))
    Up to and including C4532Y5V1A107Z 100uF 10V
    And NOJC107M004RWJ 100uF 4V Y5V

    These are all 1812 SMT parts.
    Electrical excitation to mechanical resonance is the DEFINITION of
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day