Connect with us

Video Switch Circuit

Discussion in 'General Electronics' started by Quack, Sep 4, 2003.

  1. Quack

    Quack Guest

    Hi,

    I have made a few 'video switching' circuits that are failing,
    hopefully someone can give me some tips .. perhaps my board layout is
    not good for this purpose ? (not enough ground plains perhaps ?).

    Firstly the board layout can be viewed at
    http://www.quack.cc/video.jpg
    p15v332-Q PDF can be viewed at http://www.quack.cc/P15V.PDF

    The 18 pin IC in the middle is a PIC 16f84 (DIP Through-Board) running
    with a 4mhz ceramic resonator, which receives serial signals from a
    host and configures its outputs (portA and portB, mixed) to control
    the video switch chip, which is a P15v332Q (QSOP surface mount package
    chip).

    Perhaps the resonator is causing some interference in the circuit ? -
    i have never dealt with any RF or video circuits before, maybe im
    missing some really simple rules that must be followed for a good,
    clean signal ... ?

    The problem:
    After soldering them up, they work nicely for a brief period of time.
    After about an hour or so, the video signal gets visible iterference,
    it seems to accumulate. After a day or so its completely unusable.

    Some of them i did a re-soldering job on the P15V (surface mount
    component), basically i just melted the solder again, and this makes
    it 'work' cleanly once again. some of them continue to work forever,
    some of them stuff up again.

    But all of them have a slightly 'dirty' signal, even the 'working'
    ones.

    I have not used SM components before, so perhaps my technique is not
    good.
    I have tried quite a few techniques, but i find the following to be
    the easiest to do and achieve what i think is a good result. I dont
    have special tools or anything for surface mount.

    The technique i use it to make my PCB using BLUE PNP paper printed
    from a laser printer (part# PNPB100 - www.www.techniks.com), i iron
    that on, peel it off and etch my boards in Ferric Chloride. This gives
    a fairly good result, and no shorts and good, solid tracks.

    I then coat the board in "flux 140" which is a dispenser pen which has
    both flux and a protective coating for the board to stop oxidization.
    This has worked wonders for all other work i have done. (part#
    363-6539 @ £7.29 from RS components, www.rswww.com if anyone is
    interested)

    Then to install the Surface Mount video switch, i wet each pad with
    just enough solder to 'paint it silver", (using 60/40, flux loaded
    solder).
    (i read somewhere that silver loaded solder is better for SM
    components because it conducts the heat faster and therefore minimises
    component damage because its soldered on quicker, and something todo
    with metal bonding ?? could this be my problem perhaps ?)

    I then position the chip under magnification and hold it down while i
    lightly and quickly touch the top of each pin with a fine solder tip
    at about 400' degrees. this melted the pads just enough for the legs
    to stick nicely, and leaves a good looking clean job. I then proceed
    to test its connection with a continuity tester, to ensure that the
    top of each leg does infact have connectivity to its track, and that
    there are no shorts. these tests work out fine.

    (even after the 'signal degradation', the continuity test still shows
    no problems, even when they completely dont work !! what is going on
    ?!?)


    Im at a loss why these are failing ? i cant see any problem.

    Any hints or suggestions would be great! :)

    Alex.


    PS : i cant change 'video switch' chip brands, due to the large stock
    i now have of them, the P15V332 must be used, and in SM form. I wish i
    could have got them in through-hole form, but unfortunantly they just
    didnt come that way. I chose these largely because of their
    availablitity and price. They cost only $2.10 from www.digikey.com.
     
  2. Robert Baer

    Robert Baer Guest

    Oops! Typo-- should have been size 1206 chip capacitor!
     
  3. Quack

    Quack Guest

    Hi Robert,

    Wow, great info, thats for the help :)

    Okay, so silver loaded solder wont help me much ... thats good to know
    :).
    I cant remember where i heard about it ...
    Sorry, i *think* i understood what your saying, but that would imply
    im using a double-sided PCB ? its only single side, the PIC and other
    components are 'through-hole' and the video switch is mounted on the
    bottom of the board.

    By 'copper ground planes' you mean leaving a lot of board area
    un-etched and lots of attachments from that to any other ground in the
    circuit ?

    Is my circuit visibly lacking in ground planes and vias ?

    (www.quack.cc/video.jpg)
    oooh.. i dont have any capacitors in this circuit at all :(. should i
    have ?
    i noticed in my PIC book it recommends a .1uF cap going to ground from
    the pic's ground pin. is this what you would call a 'bypass capacitor'
    ?

    Yup, i have soldered plenty of SM resistors and caps ... theyr easy,
    not much chance of shorting their pins :).


    ahh, this is sounding likely .. although i dont understand how
    exactly.
    all the pins, PORTA and PORTB are set as outputs in the PIC program.
    Although not all pins have any load or connection at all - are these
    the ones that could be 'floating' ? and if they are, how would they
    (being unused) affect the rest of the circuit ?

    Or are you saying that the i/o pins i am using to control the video
    switch (p15v) may be floating around ? - they are connected to the
    inputs of the other chip and set as outputs only (100% of the time),
    surely this would not allow them to 'float' ?
    Or maybe, i dont know much about this ... if this was the problem,
    what can i do to fix it ? use a pull-up resistor or something ?

    right over my head on that one!
    Yup, 100 checks ... not *All* of them have this problem. some seem to
    be working nicely. so the circuit must be functional ... Which leads
    me to believe it must be a physical problem. Someone suggested i use
    'flux remover' after soldering (not included in the process i
    described above) as i am using flux-loaded solder and flux-protectant
    on the board.
    I didnt have any cleaner, so i used some 'isopropyl alcohol', i think
    its the main ingredient anyway - and i got a bad one to work!. perhaps
    the flux was conducting ?!?

    I havent had this problem on my other boards because i dont use any SM
    components so things are farther away and i guess the flux residue
    doesnt affect them at that distance ...

    Im about to make a new batch today - but this time i am scrapping the
    P15V chip and going with some NPN transistors to switch the inputs
    on/off. I made a small test on a breadboard, seems to work well -
    we'll see how well it works as a completed circuit.

    Thanks :)

    Alex.
     
  4. Robert Baer

    Robert Baer Guest

    I *did* mean a 2-sided PCB, as with a certain trace width, the result
    is a transmission line near 50 ohms (it has been over 30 years so i do
    not remember the correct width, but i think it should be 0.1").
    For traces longer than 1/8 wavelength, that becomes significant, and
    long transmission lines should be terminated in their characteristic
    impedance - else one gets standing waves at constant frequencies.
    Pulses get reflected at unterminated ends, which could give voltages
    greater than "nominal".
    Having a ground plane also reduces coupling of signal between parallel
    traces.
    However, if the traces are all short relative to 1/8 wavelength of the
    highest frequency (look at risetime and fall times for that if pulses
    are there) then a ground plane per-se might not be needed.
    Having ground traces everywhere, especially for bypassing can help in
    that case.
    There is a "dirty" trick to make PCBs with very little etching, that
    can produce a ground trace around everything else: using paper and a
    drafting pen with black ink, draw "islands" that are the traces you
    need; put dots where you want holes. This gives you a paper negative
    that can be archived.
    You might be able to shave off some length of the PCB and shorten many
    of the traces to the SOIC by moving it closer to the 18 pin device.
    At least add in bypassing on that SOIC and the 18 pin device with
    ceramic chip capacitors....
    **
    Regarding RF levels and DC bias on each and every input pin of the
    SOIC (i get the impression that is where the problem is):
    1) Say the supply is +5V and that is the only supply.
    2) Say it is CMOS.
    3) Say logic inputs are DC biased at logic levels near 0V and 5V because
    the inputs are being driven by CMOS logic devices.
    Now all one needs is to have long and unterminated traces. Rapid rise
    times will produce reflections, leading to inputs being driven well
    above the supply voltage.
    4) Say signal inputs have a nominal DC bias of 2.5V, and signal levels
    are RF, peak-to-peak of 4.9V.
    Now all one needs is that bias level to increase more than 0.1V *or*
    the RF amplitude to increase more than 0.1V --- the signal input then is
    driven more positive than the supply.

    Either condition #3 or #4 can lead to the symptoms you mention.
    If you take that SOIC unit that is rather bad and heat it for a decent
    time, the damage will anneal out and it will work OK (for a while).
    ***
    A capacitor from a supply pin to a good ground trace is a bypass
    capacitor.
    Ground pins are already connected to ground...
    ***
    *ALL* unused input pins should be connected to ground, if that biasing
    is OK for the circuit. If a supply bias is recommended, use a 1K to 10K
    resistor to the supply and add a bypass capacitor to ground. A floating
    input pin can "pick up" RF and become driven above the supply voltage,
    causing the same problem i previously mentioned.
    ***
    Input pins to the SMD that are driven by a programmable pin that
    (after programming) acts like an output may or may not cause problems.
    Some programmable I/O devices are "open drain" which require an
    external pullup; look carefully for this possibility.
    When in doubt, add a 10K resistor to the supply; if an open drain, the
    rise time may be a bit slow, but it will work for all cases. The
    resistor certainly will protect both inputs (ESD, etc) when there is no
    power, and before programming to "out".
     
  5. Quack

    Quack Guest

    Hi Robert,

    Wow thanks for all your help :)

    Although i have found these other chips, MAX4141's, and they do check
    the sync signal on all 4 inputs, have buffers or something and switch
    cleanly.

    I have ordered their 'eval' board so i can see how its implemented and
    basically copy that into my circuit. Although it seems simple enough
    from reading their PDF .. (although thats what i thaught about the
    p15v chip!).
    Any comments about that ?

    Also on www.digikey.com it lists their package as SOIC, i found a
    description on google somewhere, saying "small outline plastic
    packages" for SOIC.. is that right ? - doesnt seem too descriptive..
    is it surface mount ? :)

    Thanks again!,

    Alex.
     
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.
Similar Threads
There are no similar threads yet.
Loading...
Electronics Point Logo
Continue to site
Quote of the day

-