Stray current causing IC chip crashes detection techniques?

Discussion in 'General Electronics Chat' started by steward, Oct 29, 2014.

  1. steward

    Thread Starter Active Member

    Jan 21, 2011
    38
    0
    I was hoping by buying a Fluke 771 that goes down to .01 milliamp ... that it would be helpful in the pursuit of stray voltage that can cause IC chips to crash. That is ... going beyond just checking for less than 1 ohm between a printed circuit board ground & the main electrical panel earth ground isolated ground. The other advantage to he Fluke 771 over a logging oscilliscope is that it eliminates the need to use probes because it is a less invasive clamp on style ammeter.

    From what I gathered using a fancy $8K+ electrometer that logs extremely low amounts of stray voltage or current can be a solution. Not sure if the cost-benefit $$$ equation looks good enough though ... plus they & higher end oscilloscopes require probes vs the less invasive clamp on approach.

    Any thoughts or am I just consumed by some unproductive wishful thinking for this type of troubleshooting?

    Mike
     
  2. MrChips

    Moderator

    Oct 2, 2009
    12,432
    3,360
    That would be my answer.

    There are 101 reasons why an IC could fail. Measuring stray voltage or current is not my way of detecting the problem.
     
  3. Papabravo

    Expert

    Feb 24, 2006
    10,137
    1,786
    I'm not sure what you actually mean by crashing, but if it means that they fail to operate properly then it is far more likely that one or more of the "Absolute Maximum Ratings" has been exceeded. The most likely culprit in my experience is static discharge which can involve thousands of volts. Even if the failure is not complete, a part can be weakened or compromised by a static discharge. The next most likely is excess current causing a thermal overload. I should point out that in a career spanning half a century that the number of actual IC failures I've seen can be counted on the fingers of one hand.
     
  4. steward

    Thread Starter Active Member

    Jan 21, 2011
    38
    0
    MrChips & others,

    Mainly, I am concerned about smaller control boards that have one or two very small intel IC chips with embedded logic. We have ability to do some programming within a keypad that is also used for an operational interface. Maybe focusing mostly on EMF caused by other components etc. will help narrow down to less than 101 reasons. The only explanation I can think of is that stray voltage or current gradually builds up & makes the IC chip lose some of its programming along with crashing. That is why it seems like using a clamp on ammeter with the right capabilities could be helpful.

    Mike
     
  5. takao21203

    Distinguished Member

    Apr 28, 2012
    3,577
    463
    high frequency noise can travel over DC lines and it just ignores capacitors, and can pass air gaps too.

    You need to shield supply lines and earth the shield + install a small choke too.

    If you have serious stray currents induced you need to consider full shielding. This can be caused by microwave circuits, tesla-coil like circuit, large motors or ballasts, and high voltage generator circuits.

    Or you can consider to shield the noise source and their supply wires.

    These currents are difficult to measure- for one, you need to simulate an "antenna" at the right place, and in some cases, these can be UHF harmonics very difficult to measure at all.

    Often it is enough just to shield the supply wires to a sensible IC (twisting a wire around them can be enough), and to inline a small choke for instance 22uH stops UHF from passing.

    I found a circuit where just one of these measures wasnt enough, both were needed.

    If you get rid of the problem, you dont need to measure anything.

    Oh yes its that kind of circuit, patching a LM2576 dc/dc with exernal gate driver transformer...then triggering the photo flash is enough to saturate the small ring core with an EMF pulse- effectively the chips shuts down and the large lamp goes off...

    If it doesnt show a reaction anymore, why measure a thing?
     
  6. Papabravo

    Expert

    Feb 24, 2006
    10,137
    1,786
    To apply the scientific method you need to have a hypothesis which you have and you need to perform an experiment which you are trying to do. Based on my half century of experience I believe your results will be inconclusive. My number one suspect is a wiring or fabrication error on the board which is causing this corruption. In order to test that hypothesis we'll need a schematic and links to the datasheets for the chips.
     
  7. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    You also might need to sacrifice a goat in a satanic ritual to find the solution.
    Seriously though, I dont really undestand what you mean by
    How and why would current and or voltage gradually build up? Where would it come from and where would it be stored? I think you are approaching this from a completely wrong end.
    You seem to imply that the stored program gets lost. In what kind of memory is it stored?
    What interfaces to outer world does this mesterious device of yours have? What kind of power supply is it using? Are there any galvaincally isolated parts?
     
  8. Papabravo

    Expert

    Feb 24, 2006
    10,137
    1,786
    Magical thinking is what is left when no rational suggestion seems plausible.
     
  9. joeyd999

    AAC Fanatic!

    Jun 6, 2011
    2,674
    2,724
    I supposed that depends upon what you mean by "failure". Back around '84 or '85, I was responsible for diagnosing the root cause of loss of calibration on a board using the veritable Intersil 7106. The boards would be calibrated outside of the enclosure, and once installed, the calibration would change.

    It turned out that flexing the board during assembly into the enclosure also flexed the 7106 IC package which change the output of the on-chip Vref. Intersil had to buy back and replace hundreds of their chips, which were quite expensive at the time.
     
  10. Papabravo

    Expert

    Feb 24, 2006
    10,137
    1,786
    Anecdotal evidence is just that. I'm just trying to suggest that the OP should pursue other avenues of inquiry before spending big bucks on an instrument that is unlikely to produce the required insight to solve his problem.
     
  11. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    5,779
    1,103
    The jaws of the Fluke771 seem to be about 0.5" wide. Isn't that much too big for probing when you are "concerned about smaller control boards that have one or two very small intel IC chips" ?:confused:
     
  12. takao21203

    Distinguished Member

    Apr 28, 2012
    3,577
    463
    It riddles me why you cant identify something using an 8K+ apparatus, when it affects a probably just ordinary IC.

    OP needs to show the thing or not much can be said at all. Often it would be quite clear what kind of problem it has.

    About these goats, sometimes the IC manufacturers do that to solve these problems in advance, maybe OP also did some ritual already, and the spells cancel out.

    How about measuring it on a Friday, have a large mirror in the room and cover it with a skulls scarf and light 7 black candles at 3.40 PM. That'll will most likely do it, among with placing 5 garlic bulbs around the circuit board and murmuring "Daraniss Akar" 3 times.
     
  13. steward

    Thread Starter Active Member

    Jan 21, 2011
    38
    0
    Alec & others,

    Below is a photobucket link to a hi res pic of one of the boards that seem to sometimes not last for more than 6 months without developing intermittent problems that require the turning off of power to it ... & then turning the power back on after about a minute usually temporarily makes it OK again. Notice the 15 wire connector plus a place where the ground goes to the board. I have tried both the main panel ground & a dedicated wet earth isolated ground. So you see there are over 15 external wires that I with "my not too large of a brain" was hoping that the .5" diameter jaws Fluke 771 with its .01 milliamp detection would give a clue to an external stray current or EMF or lack of proper current or whatever. The boards that I send in to be rebuilt tend to be more vulnerable but also brand new $$$ boards can sometimes be a problem.

    http://i26.photobucket.com/albums/c138/mjwalsh/PCBHiResPic.jpg

    Mike
     
  14. takao21203

    Distinguished Member

    Apr 28, 2012
    3,577
    463
    Could be transients on the high voltage side.

    The board looks pretty old fashion to me. Is there a 5.6V zener diode on the chip? Normally the 7805 (guess this part is used) should cover transients but maybe it doesnt. Switching loads and having these relays + traces on the PCB can cause problems. There are no EMI filters, there is no choke on the uC supply, there is no transient arrestor, no compensation capacitors (to some transients for a small time they appear almost like a short circuit). How are the relays switched?

    The cables from the relays to the controlled devices also can pick up transients of all kinds pretty much like an antenna. Are they shielded?

    Maybe some of the capacitors are old + have gone bad so it just merely works on the margin now.

    Also these oldfashion transformers provide total isolation chances are, the circuit has no earth reference. Switching supplies do have small capacitors 1nF or so connecting to mains.

    Many possible reasons, still difficult to say without to know the complete schematics + the kind of wiring/length used + the age of the board.

    There are at least 20 possible issues, and possibly a few real causes, which through some of the issues, can promote in such a way they are able to affect and to disrupt the circuit.

    You could try firing a photo flash close to the circuit, if it hangs, you certainly know it's bad. Soldering stations and TRIAC circuits also provide transients + harmonics.

    Probably you can get rid of the issues by a 5.6V Z-diode, a small choke inlined after the regulator, a small capacitor providing an earth reference, and by shielding external cables + compensate external devices.

    I'd just check for these standard issues and not try to measure induced currents directly, it may only happen sometimes, and it may be difficult to measure them at all.
     
    Alec_t likes this.
  15. steward

    Thread Starter Active Member

    Jan 21, 2011
    38
    0
    I rigged up the below linked inline filter where the 120vac goes to the controls part of the machine. It did help some making it so there were fewer "off for a minute" episodes needed. Being pretty limited in my knowledge ... I am not sure that this is the best choice of an EMI or EMF filter for the 120vac that powers everything except for a 1/2 HP motor that is controlled by another close-by PCB with a mounted mechanical Class C dry contact relay with 10 amp rated contacts. Any thoughts on this specific filter & if it is optimal in this situation. There are two 2 amp fuses as part of the controls so I figured the in line filter's 6 amp rating would be close enough.

    http://www.newark.com/te-connectivity-corcom/6eeb1/rfi-power-line-filter-6a-380ua/dp/70K9990

    Mike
     
  16. takao21203

    Distinguished Member

    Apr 28, 2012
    3,577
    463
    well, large motors switched by relay, cables to them, there you have it.

    The relay may turn on at any point of time (AC voltage goes from negative to positive maximum).

    One approach would be to use a TRIAC and turn on the motor close to zero crossing (the moment when the AC is zero volts).

    Still you can get pretty awful EMI and transients from a motor (which basically is a large inductor), connected with cables of some length. The cable and the motor coils can pick up transients, and the motor generates EMI and harmonics. You need a filter before the motor.

    A switching supply on the PCB is also better, as some kinds of EMI and transients cant pass through the inductor or at least they are attentuated.

    If the chip is loosing its stored program, not just hanging up, you have some serious issues going.

    Microchip PICs have a WDT so if the chip hangs, it can self-reset. Its still no good, only a last resort, hanging controllers should be avoided by all means.
     
  17. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    5,779
    1,103
    I agree with Takao21203.
    1/2HP = ~3A running current. Start-up current would be much higher. 10A contacts may be under stress and could be arcing badly, generating EMI.

    Edit: If that's a slide-switch left of the transformer it might be worth excersising it with a squirt of switch-cleaner in case there's an intermittent connection.
     
    Last edited: Oct 30, 2014
  18. MrChips

    Moderator

    Oct 2, 2009
    12,432
    3,360
    Fault identification and reliability testing depends on your situation and the work/product you/your company produces.

    Are you:
    1. Semiconductor or IC manufacturer
    2. OEM (original equipment manufacturer)
    3. Systems integrator
    From the information you have provided I assume you are 3) Systems integrator.

    The diagnostic approach you choose will depend on the environment your product is employed:
    1. Residential
    2. Commercial
    3. Industrial
    I assume your environment is 3) Heavy industrial.

    Is the failure mode temporary or does it result in permanent failure?

    To diagnose the failure mode you need to make the device fail.

    1) For reliability testing in heavy industrial environment, you need to subject the DUT (device under test) in severe and extreme conditions, temperature, pressure, humidity, over/under voltage, EMI/RF interference.

    2) You need to monitor and record the incident of failure and events leading up to failure. This requires monitoring equipment, digital/storage oscilloscopes, logic analyzers, chart recorders, data loggers, etc.

    Until you can pin-point the exact cause leading up to the failure there is little else that can be done to mitigate the problem. Anything else would be shooting in the dark.
     
  19. MrChips

    Moderator

    Oct 2, 2009
    12,432
    3,360
    One important note: ESD (electrostatic discharge)

    Many OEM and board assemblers do not take ESD prevention seriously.

    How do you know that your boards and ICs received the proper treatment and care during assembly, transportation and installation?

    Component and board failure months and years after installation can be the result of improper or non-existent ESD policy and procedure during board assembly.

    For example, are your boards shipped in anti-static containers?
    Do your personnel wear anti-static wrist bands during installation and testing?
     
  20. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    I think that OP is not a system integrator, I think he works in some kind of repair shop, based on the lack of detail about the PCB itself, and the fact that the board seems to be made around 2008 judging from that Micrel IC datecode.
     
Loading...