Connect with us

soldering small surface mount discretes

Discussion in 'Electronic Design' started by Michael, May 12, 2007.

Scroll to continue with content
  1. Michael

    Michael Guest

    Hi - I am working on a board where I am quickly running out of space.
    The thing that is killing me the most is all the decoupling caps
    around an incredibly sensitive ADC. I need to keep the same number of
    caps. Currently I have them all as 0603s, but that's just too big.
    This is a prototype that I'll be hand soldering. I've done hundreds of
    0603s and I find them as easy as can be. I suspect 0402s won't be that
    much of a difference. Now - what about 0201s? Anybody ever tried hand
    soldering those buggers? I've also heard that 0201s have a nasty
    tendency to tombstone unless you are very careful with your
    temperature profiling. Any truth to that?

    Thanks!

    -Michael
     
  2. John Larkin

    John Larkin Guest

    How many caps are you using? Which ADC is it?

    John
     
  3. DJ Delorie

    DJ Delorie Guest

    Yup. http://www.delorie.com/pcb/smd-challenge/

    0201's aren't *that* hard (the 01005's are) but they're not as easy as
    0402s. Part of the problem is that the parts are on the order of the
    same size as your wire solder and iron tip. Even your pulse is enough
    to bump them out of place, so keeping them in place while soldering is
    trickier.

    What I do now (for 0603s) is manually put down solder paste, place the
    parts, and use the iron to reflow the paste while holding the parts
    down with tweezers or a toothpick. You don't have to get the iron as
    close as with wire solder; you only have to touch the copper trace and
    the heat is conducted under the part to melt the paste.

    If you can etch a paste stencil, you can put the paste down with that
    and use either a hotplate or iron to reflow it. Otherwise, it's still
    possible to manually paint the paste if the syringe is small enough.

    Note that with the really small parts, both pads heat up at pretty
    much the same time even if you only touch the iron to one side. This
    means the "solder one pad to hold it down" trick doesn't usually work;
    the first pad re-melts when you solder the second pad.
    With hand ironing, it's not just tombstoning you have to worry about -
    it's bumping the parts and messing up the alignment. The surface
    tension of the solder on one side can easily move these parts, and
    jiggling the solder with the iron moves them too. Even with parts as
    big as 0603, I have to hold the parts in place while soldering.
     
  4. Michael

    Michael Guest

    So you don't have any of those boards left, right? It sounds like a
    good idea to make a trial board like that.
    So with 0201s, is there still enough space to hold on to them with
    tweezers while soldering them down?
    Interesting - those buggers sure are teeny. At that size is there any
    worry about them getting damaged from the heat? I mean with larger
    components there is a whole lot more surface are to dissipate the the
    heat.

    I do the same - I always put a dot of solder on one side of the PCB,
    pick up the part with tweezers in my left hand, then heat up the dot
    again with my iron in my right hand, and place the part at the same
    time. Easy as can be.

    Thanks,

    -Michael
     
  5. Michael

    Michael Guest

    Currently 23 capacitors, 8 of which are fairly large tantalums. The
    rest are either 100nf or 1uf, so I think could be pretty easily
    replaced with much smaller parts. The ADC is an Analog AD7656.

    -Michael
     
  6. D from BC

    D from BC Guest

    How come you're not using a hot plate?
    D from BC
     
  7. Michael

    Michael Guest

    Along the same lines as my first question: anybody know of any small,
    reasonably priced large capacitors? My standard decoupling on a chip
    is a .1uf on every Vcc/Vdd and a 10uf cap on each supply line/chip.
    I've been using Digi-Key # 478-1751-1-ND, which is a 10uf 10V in a
    3216-18 package, costing $.54 in quantities of 100. A bit spendy, but
    oh well.

    But I want to go smaller. I was just looking through Digi-Key and they
    have some 0402 and 0603 tantalums with fairly good looking specs, but
    they cost $3 or more each! Ouch. Any options? The M size parts
    (1.6x0.85mm) seem like the best alternative, like the 511-1494-1-ND:
    6.3V, 10uf, $0.46 in quantities of 100.

    Suggestions?

    Thanks!

    -Michael
     
  8. Michael

    Michael Guest

    Actually I'm going to have to use a hot plate to do one of the
    components on this board (a BGA).
     
  9. To replace tantalums and small electrolytics, I use a lot of
    the 10 uF 10% 25 volt 1206 X5R Panasonic ECJ-3YB1E106K.
    These cost $0.55 each in quantity 10 from Digikey and have
    about a milliohm of ESR. I've never had to parallel them
    with a smaller value, yet.

    http://www.panasonic.com/industrial/components/pdf/ABJ0000CE4.pdf
     
  10. John Larkin

    John Larkin Guest

    Those decoupling requirements are insane. I doubt that the chip
    actually needs all these caps.

    John
     
  11. John Larkin

    John Larkin Guest


    We've sworn off tants for cost and reliability reasons, and off
    aluminums because of their low-temperature issues. So we mostly use
    ceramics. I think we're using mostly 0.33 uf 0603 for general
    bypassing, and for the big stuff we have

    0603 2.2 uf 0.09 each

    0805 10 uf 0.22 each

    1206 22 uf 0.63 ea

    or something like that.

    John
     
  12. Michael

    Michael Guest

    That's quite possible. Then again, I'm not going to argue with the
    guys who designed the part. I need every ounce of accuracy that I can
    get out of this thing.

    -Michael
     
  13. Do you know of some good 20uF or so at 75V? They seem to be just above
    what's available at Digikey etc. (especially the 75V requirement). I
    can put lots in parallel, of course, but it starts to get silly.



    Best regards,
    Spehro Pefhany
     
  14. DJ Delorie

    DJ Delorie Guest

    They're too small to make - practically - in small quantities. I
    originally made a batch of 100 kits (so, 100 of each part and shared
    FAB costs), and sold them at cost to folks on the gEDA mailing list,
    so that we could get some experience with the smaller parts. Each kit
    cost $2 plus S&H, but it would have been much more if I had made less.
    Even then, I added it to another board I made, to share the FAB costs
    across projects.

    I put the design files online in case anyone wants to add a dozen or
    so to their existing two-layer projects, for their own use. They're
    only 0.5" square.

    I've thought about doing another run, but I keep mentioning the
    project and people keep not asking me to make more ;-)
    Not really. The tweezers I have, have needle points. I use one point
    to hold the part down (i.e. press down on top, not hold from sides)
    while soldering the two ends. Even toothpicks need to be sharpened
    with sandpaper to be small enough.

    Here's a comparason of 0805, 0603, 0402, 0201, and 01005 parts:
    http://www.delorie.com/pcb/0805-01005.html

    Close-up of 0201s:
    http://www.delorie.com/pcb/0201-parts.html

    My first hand-soldered 01005 (right side is 0.4mm pitch IC):
    http://www.delorie.com/pcb/first.html

    Some microscope pix of various parts on laser printed footprints:
    http://www.delorie.com/pcb/smd-challenge/fp-pix/
    I've found the real problem is mechanical damage. It's really easy to
    try to nudge a part into position after soldering one end (i.e. while
    melting that end's solder), only to rip the entire cap off that end
    because the solder wasn't quite melted enough.

    Desoldering pretty much ruins the parts too, assuming you can find
    them afterwards.
     
  15. DJ Delorie

    DJ Delorie Guest

  16. Michael

    Michael Guest


    I have a friend that sells educational electronics kits for hobbyists.
    I'm going to try to convince him to make something similar.

    If you do another run let me know, I'll take a couple off of your
    hands.
    Ah - that'll be interesting. So with both pads clean and free of
    solder, you place the part and hold it down, then solder both sides?
    How do you get solder on to it? It seems to me that both of your hands
    would be busy. Just put a little on the tip of your iron maybe?
    I pretty much toss all surface mount components that I desolder -
    life's too short to try to keep track of them all.

    Well - I guess I'll making the big switch from 0603 to 0201 for now.
    Hopefully I'll survive. Do the electrical characteristics of the parts
    change much from 0603 to 0201?

    Thanks,

    -Michael
     
  17. Jeff L

    Jeff L Guest

    Caps do - the dielectrics are often very voltage dependant. High k
    dielectrics are the worst. Going from a 0603 to a 0402 cap with almost
    identical specs can show over a 40% - 60% reduction in capacitance at high
    percentages of working voltages. With 0402's, you are usually working at
    high percentages of the working voltages.
     
  18. Jeff L

    Jeff L Guest

    There are some that come with a double stacked cap with leads soldered on
    the side with high temp solder to hold them together. They were 100v and
    around the range in capacitance you were looking for. I can't remember the
    brand, but they were larger caps, may be 2010's, I think they were bought at
    Digikey.

    There is nothing wrong with placing several caps in parallel.
    http://www.speff.com
     
  19. mpm

    mpm Guest

    Honestly, I stuggle with 0603's.
    I suspect I don't have the right equipment and/or technique.

    I just wanted to add (for us Engineers anyway..) that maybe the Eye
    Doctors could update their charts. Instead of 20/20 eyesight, we
    could have 06/03, or 02/01.?

    Just a thought.

    -mpm
     
  20. John Larkin

    John Larkin Guest

    Do you need a very low wideband noise floor? If it's accuracy you
    need, oversample and average or lowpass filter. Any real-life sampling
    16-bit ADC system will have many LSBs of noise anyhow.

    That adc scares me. Some of the nodes are bypasses of active circuits,
    and whenever I see strange, unexplained bypassing requirements, I
    wonder if the datasheet is hiding ldo-type stability problems, or if
    the applications engineer was just being too enthusiastic.

    John
     
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

-