Connect with us

troubleshooting method for micro-controller board ?

Discussion in 'Electronic Repair' started by robb, Nov 5, 2007.

  1. robb

    robb Guest

    I can only work on this in short segments of time and am usually
    tired ... so i am moving backwards.

    Anyway the revisit.

    The last diagnostic consensus was to remove voltage regulators
    and inject a good 5v power into the 5v lines and look for
    warming components to get a fix on problems to repair.

    Since my 5.2V *aged* lantern battery power supply did not impress
    many i cobbled a simple 5v power supply from a 18 VAC/2.5A wall
    wart, a 100V/1.5A bridge and 7805 5V regulator plus some
    resitors and a 6V bulb to test.

    I connected the 5v (through 450 Ohm) to the PCB 5v and 0v lines.
    i measured ~60 mA curent nothing warmed ...i idecremented the
    resistance by 100 for each iteration of testing the measured
    currernt increased slightly until i made it to (100 Ohms) and
    then the current spiked to ~+ 1.5A and the 7805 fried ?

    but still no heat on the board anywhere, no trace no component no
    IC ?

    do i need to do something different with power supply test ?
    thanks for any help,
    robb
     
  2. Arfa Daily

    Arfa Daily Guest

    Do you mean that the 7805 got hot and shut down ? One of those should never
    "fry", because they are SOA protected. I suspect that something else
    happened there, because if you do the math, it is impossible for 1.5 amps to
    flow through a 100 ohm resistor, with only 5v driving it ... Apart from
    that, most variations of the 7805 are only rated to 1 amp, although there
    are exceptions.

    When you cobbled together your test supply, did you mount the regulator on a
    heatsink, and most importantly, did you place a 4u7 cap, paralleled with a
    0u1 cap directly across the output and ground pins of the regulator, as
    close to the device as you could get them? This is *very* important, to
    prevent the regulator bursting into vicious ultrasonic or higher
    oscillation. It may be just that you reached a current level where this
    happened, and the meter didn't know what to do with what it was measuring,
    so just displayed some meaningless nonsense. These regs do get very hot when
    they oscillate. Although not strictly necessary, it's considered good 'belt
    and braces' to put a 0u1 cap between input and ground pins as well.

    The more I think about the 'problem' that this board has, the more non
    'real-world' it seems. I can't really see any reason why three unrelated
    fuses, on three unrelated rails, should have blown, unless there was a
    serious and easily visible problem. The only thing that did occur to me was
    input voltage. You're not by any chance firing the thing up on UK 240v
    power, and its transformer is actually rated for 100 or 110v ? That would do
    it ...

    Arfa
     
  3. robb

    robb Guest

    Thanks for help Arfa,
    I am describing from memory, the last thing i recall was
    decrement too the 100 Ohm resistor and the amp meter's numbers
    were jumping around and 1.5 is the largest number i caught with
    my eye.

    as for fried 7805 ... the 100 ohm resistor started to smoke and i
    think i saw the 7805 let out a puff of smoke and now the 7805
    does not produce output ? so i assumed it was fried

    this all happened very quickly
    the pcb i took it from had a small flat square of aluminum bolted
    to it 1.5cm x 1.5cm x 1 mm
    i'll try the test again with those mods
    it is 115v plugged into 115v supply
    i suppose if i could address the issue of the seemingly unrelated
    supply rails i could probably figure it out

    if this info helps ....

    the machine *was working* when i replaced a VFD display driver
    chip. That was easy problem to fix as it was very specific. I
    started with closest components, checked pin signals with oscope
    and i had a datasheet for guidance. Everything was working i just
    needed to bolt it together.

    *BUT* i found a yucky ring in peaks of the clock signal and
    thought i should do something to repair that (find the source of
    ring) again more very easy specific tasks, just follow traces and
    check for expected behavior what goes in and what comes out

    well i forgot to put the oscope probe tip guard on (to prevent
    shorting pins) and i bridged the 32v VFD supply line to the 5V
    rail (a little snap) and all sorts fo jitters came out of the
    connected devices, i disconnected all those devices and i found
    one lone fuse between a bridge and the transformer was blown i
    think the 8V or 16V line that supplied the voltage regulators and
    servo controler chips {L387,L298}

    i replaced fuse , powered on and all three fuses blew

    now i am here with very nebulous problem with no specific ideas
    about how to repair

    looking for help

    thanks again for your help arfa,
    robb
     
  4. robb

    robb Guest

    Thanks for help Arfa,
    I am describing from memory, the last thing i recall was
    decrement to the 100 Ohm resistor and the amp meter's numbers
    were jumping around and 1.5 is the largest number i caught with
    my eye.

    as for fried 7805 ... the 100 ohm resistor started to smoke and i
    think i saw the 7805 let out a puff of smoke and now the 7805
    does not produce output ? so i assumed it was fried

    this all happened very quickly
    well the regulator had a small flat square of aluminum bolted
    to it 1.5cm x 1.5cm x 1 mm when i pulled it from some other pcb
    i'll try the test again with those cap mods
    it is 115v plugged into 115v supply

    i wish i knew enough to speculate the connection , for now i
    can only report what i see and do. i suppose if i could address
    the issue of the seemingly unrelated supply rails i could
    probably figure it out

    FWIW, if this info helps ....

    the machine *was working* when i replaced a VFD display driver
    chip. That was an easy problem to fix as it was very specific. I
    started with components that were related to the VFD (ie. driver
    chip) , i checked pin signals with oscope and i had a datasheet
    for guidance. Driver chip had 6 bad lines that would not allow
    user button presses to be seen by main board. I replace the DIP40
    with a PLCC chip and Everything was working i just needed to
    re-assemble and bolt the pieces together.

    ***BUT***
    i stumbled across a "yucky" ring in the peaks of the clock signal
    to the VFD driver chip and i thought i should do something to
    repair that (ie find the source of ring) again more very easy
    specific tasks, just follow traces and check for expected
    behavior (in vs outs) what goes in and what comes out

    well i forgot to put the oscope probe tip guard on (to prevent
    shorting pins) and i bridged the 32v VFD supply line to the 5V
    rail (a little arc snap) and all sorts of jitters came out of the
    connected devices, i disconnected all those devices and i found
    one lone fuse between a bridge and the transformer was blown. i
    think the 8V or 16V line that supplied the voltage regulators and
    servo controler chips {L387,L298, etc}

    i replaced the one fuse, powered on and all three fuses blew

    since then i have checked transformer output (ok), desoldered the
    L298/L387 ics, continuity tests on various parts of PCB (73 Ohms
    between 5v and 0V all around the board), tried to feed 5v into
    the 5v line as suggested on the group to feel for warming
    components,

    now i am here with a very nebulous problem with no specific ideas
    about how to repair

    and looking for help

    thanks again for your help arfa,
    robb

    BTW there are pictures of all this stuff on
    (alt.binaries.schematic.electronics)
     
  5. robb

    robb Guest

    Thanks for help Arfa,
    I am describing from memory, the last thing i recall was
    decrement to the 100 Ohm resistor and the amp meter's numbers
    were jumping around and 1.5 is the largest number i caught with
    my eye.

    as for fried 7805 ... the 100 ohm resistor started to smoke and i
    think i saw the 7805 let out a puff of smoke and now the 7805
    does not produce output ? so i assumed it was fried

    this all happened very quickly
    well the regulator had a small flat square of aluminum bolted
    to it 1.5cm x 1.5cm x 1 mm when i pulled it from some other pcb
    i'll try the test again with those cap mods
    it is 115v plugged into 115v supply

    i wish i knew enough to speculate the connection , for now i
    can only report what i see and do. i suppose if i could address
    the issue of the seemingly unrelated supply rails i could
    probably figure it out

    FWIW, if this info helps ....

    the machine *was working* when i replaced a VFD display driver
    chip. That was an easy problem to fix as it was very specific. I
    started with components that were related to the VFD (ie. driver
    chip) , i checked pin signals with oscope and i had a datasheet
    for guidance. Driver chip had 6 bad lines that would not allow
    user button presses to be seen by main board. I replace the DIP40
    with a PLCC chip and Everything was working i just needed to
    re-assemble and bolt the pieces together.

    ***BUT***
    i stumbled across a "yucky" ring in the peaks of the clock signal
    to the VFD driver chip and i thought i should do something to
    repair that (ie find the source of ring) again more very easy
    specific tasks, just follow traces and check for expected
    behavior (in vs outs) what goes in and what comes out

    well i forgot to put the oscope probe tip guard on (to prevent
    shorting pins) and i bridged the 32v VFD supply line to the 5V
    rail (a little arc snap) and all sorts of jitters came out of the
    connected devices, i disconnected all those devices and i found
    one lone fuse between a bridge and the transformer was blown. i
    think the 8V or 16V line that supplied the voltage regulators and
    servo controler chips {L387,L298, etc}

    i replaced the one fuse, powered on and all three fuses blew

    since then i have checked transformer output (ok), desoldered the
    L298/L387 ics, continuity tests on various parts of PCB (73 Ohms
    between 5v and 0V all around the board), tried to feed 5v into
    the 5v line as suggested on the group to feel for warming
    components,

    now i am here with a very nebulous problem with no specific ideas
    about how to repair

    and looking for help

    thanks again for your help arfa,
    robb

    BTW there are pictures of all this stuff on
    (alt.binaries.schematic.electronics)
     
  6. clifto

    clifto Guest

    That's pretty good. If you then shorted the PCB 5V and 0V lines, you would
    have 450 ohms across 5V and should read 11.1 ma.
    To get 1.5A through 100 ohms you need to apply at least 150V to it.
     
  7. robb

    robb Guest

    thanks for reply clifto,

    i will have to believe you on the numbers

    and although the numbers may seem incorrect from a electronic
    formulaic analysis i can only report what i see on the equipment
    i use and with the setup i use.

    The DMM amp reading numbers were fluctuating wildly on my last
    test and 1.57 ???was something i thought flashed by as highest #
    maybe it was .57 but i do know that whatever the number the 100
    ohm resistor had a stream of smoke going up and the 7805 appeared
    to have a puff of smoke released before i could dis-connect now
    i am hopping somebody can help me decipher what went wrong with
    my testing

    even operator errors would be useful as long as it helps nme
    omove toward a repair/
    robb
     
  8. Guest

    I'm reading this thread with some interest, but I'm more puzzled as it
    goes on, there should be something we are (and the OP is) missing.

    hmm 78xx don't smoke easily, maybe if you apply a too high input
    voltage (> 30V I believe) or if it goes into self-oscillation if an
    output capacitor is omitted. A 100 ohm resistor, assuming we are
    talking about a 1/4 W one or bigger, shouldn't smoke easily with only
    5V even on a short to ground, it should become quite hot but not smoke
    instantly, if it smoked I believe the input voltage was higher.

    Power supply problems shouldn't be that difficult to troubleshoot,
    even if once I had almost the same problem with a simple three tubes
    amplifier, but that was another story :)

    Regards
    Francesco
     
  9. ian field

    ian field Guest

    AIUI the OP is trying to find which/how many 5V rail devices on a heavily
    populated logic board are S/C after a short between 36V & 5V feeds.

    A slightly risky strategy would be to use the 5V O/P from an old (pre-3.3V)
    AT PSU, which typically can supply 200A or more via progressively lower
    current limiting resistors, at some point the current will be high enough to
    distinctly heat any S/C components enabling rapid identification, obvious
    pitfalls are S/C Unobtainable Logic Arrays, firmware ROMs the supplier
    refuse to sell as spares, the risk of blowing circuit traces (unrepairable
    in multi-layer boards) and the risk to eyesight from exploding S/C
    capacitors!
     
  10. clifto

    clifto Guest

    What I was trying to get you to realize is that there is probably a
    problem with your equipment, or with your understanding of its use.
    When the PCB supposedly draws over five times as much current as a
    dead short, it should tell you that what you think you're seeing isn't
    what is actually happening.
    If you used a 1/10 watt 100 ohm resistor, putting 5 volts across it
    should dissipate 2-1/2 times its capability, or 0.25 W. That might
    make it release smoke.
     
  11. ian field

    ian field Guest

    That's probably because most US flags are made in China these days!
     
  12. robb

    robb Guest

    Thanks for help clifto,

    i do realize what you are saying which is why i said, "i
    consider operator error to be a possible problem".
    this amateur horse brain needs some help getting itself in front
    of the cart.

    i've been told that the 73 Ohm resistance between the 5V and 0v
    is not so bad and is somewhat hopeful that i did not incinerate
    all the ICs.

    if i can diagnose without removing the ICs then that would be
    great, otherwise i may need to fall back to brute force
    diagnostics and remove ICs one at atime and check things and
    iterate .

    i was hoping to learn something a little more sophisticated.

    i was uing 1/4 watt resistors.

    i plan to try the capacitors to prevent oscillation,

    thanks for help and for any more ideas you may have,
    robb
     
  13. robb

    robb Guest

    thanks for help Francesco,

    i do not under-estimate my ability to mis-understand.
    so i consider myself missing something as distinct possibility.

    but what ?

    source is a 18-24v ac 1000mA wall wart that feeds the bridge and
    then 7805.
    someoner suggested that i add caps to prevent oscillation i will
    try and i hope to give more accurate report and numbers after i
    make the change.

    thanks for help francesco and any other ideas you may have,
    robb
     
  14. robb

    robb Guest

    Ian has the problem packaged precisely in a nutshell.

    but i don't remember the explosion risks better get my safety
    glasses.

    having said that ian i think i will desolder the rom, ram, and
    8031 just in case they are ok and are not replaceable as you
    point out.

    the board was mfg mid 80's so i am pretty certain it is only two
    layer board.

    thanks for the nutshell version,
    robb
     
  15. ian field

    ian field Guest

    In my experience of situations like this (such as regulation failure in a
    SMPSU) its quite common for a small number of devices - sometimes as few as
    1, to go S/C and "crowbar" the rail effectively protecting the remainder of
    devices. Occasionally a few devices will blow O/C instantaneously after
    going S/C, which can lead to further problems after the S/C devices are
    replaced - if the Vcc bond wire to the S/C die blows, the I/P & O/P pins cn
    the S/C die can load the O/Ps of other chips connected to it, this can be
    tedious to trace on bussed logic, its possible to snip the tops of DIL chip
    pins to isolate the chip while leaving enough pin on the chip body to solder
    the pin back on to (you can always replace the chip later if you want to)
    with the pin isolated you can use a pulse injector and scope to find if the
    snipped chip or something else on the bus line is still holding it down.
     
  16. Arfa Daily

    Arfa Daily Guest

    That is the most important thing that you must do before going any further.
    If not properly bypassed, 78 series regulators will *readily* oscillate
    violently, particularly if you have long leads attached to them. Connect 0u1
    between "in" and ground, and 4u7 in parallel with 0u1 between "out" and
    ground. It is *essential* that these are soldered *directly* to the pins of
    the device, not some distance away down the connecting wires.

    Arfa
     
  17. clifto

    clifto Guest

    Might as well add the back-bias protection diode with anode at "out" and
    cathode at "in" while he's at it.
     
  18. Arfa Daily

    Arfa Daily Guest

    Yeah, if you like, though probably an unnecessary complication for this test
    setup. Going back to the story of how this unfortunate situation arose, I
    think that there is a valuable lesson for Robb to add to what he's learning
    here, and that is "If it ain't broke, don't fix it ..." The fact that it was
    working after replacing the VFD driver chip, should have been an end to the
    job. It obviously didn't matter that the clock was a bit scruffy, as
    observed on the 'scope, as the item was managing to use it satisfactorily.
    Depending on how fast the clock is, and how good a probe was being used on
    the 'scope, the signal may not even have been 'wrong' in the first place. It
    might just have *looked* as though it was. It's pretty likely that any clock
    inputs on the chips are internally schmitt'd anyway, to make sure that a
    clean signal is being passed on into the chip's circuitry.

    Arfa
     
  19. Eeyore

    Eeyore Guest

    Not any more.

    1.5A is the norm. Look at any up-to-date data sheet.

    Graham
     
  20. Arfa Daily

    Arfa Daily Guest

    I wouldn't say it was "the norm". A quick look at Farnell's catalogue shows
    at least as many 1 amp variations of the 7805, as 1.5 amp ones. Maplin,
    probably the biggest supplier in the UK to the amateur market, as well as
    having a professional supply division, doesn't even list any 1.5 amp
    versions of the standard 78xx and 79xx series devices, although they do keep
    a small range of 2 amp devices. I would also contend that any device older
    than 5 years is pretty much certain to be only 1 amp rated, and Robb's
    original fitted to a board from the 80s will definitely be.

    Arfa

    Arfa
     
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

-