Author Topic: Having 3.3V short to GND in complex board.Looking for method to solve quickly  (Read 3296 times)

0 Members and 1 Guest are viewing this topic.

Offline muthukural001Topic starter

  • Regular Contributor
  • *
  • Posts: 211
  • Country: in
Hello All,

I am working on a board which is having 3.3V short to ground and the impedance between them is 5 ohms. Since this 3.3V is going to many places, removing components one by one is time consuming.My colleague told me an idea that give 100mv @ 0.5A  on the 3.3V path and look where the voltage is low to find the short. I have done the above method but the voltage is 70mV everywhere on the 3.3V path. Plus, I don't understand the logic of the above method that different voltage drop on the same path. Please let me know if the above method makes sense to anyone.

Also, if there is any other ways that I can try, please suggest.

Thanks and Regards,
Muthu
 

Offline langwadt

  • Super Contributor
  • ***
  • Posts: 4857
  • Country: dk
the Louis Rossmann way is to power the board and pour some alcohol on to see where it gets hot and evaporates
 

Offline KrudyZ

  • Frequent Contributor
  • **
  • Posts: 295
  • Country: us
5 Ohm isn't exactly a short.
At 3.3 V you will have 660 mA going through your rail and dissipate over 2W of power. Something is bound to get warm.
You can use a thermal camera to see what gets hot. Lacking that, you can use your finger to feel it or your nose to smell it.
Never tried the alcohol thing, but in theory it should work.
An alternative is to measure the voltage drop on the rail itself using a micro volt meter.
You would be able to track the direction of the current in the 3.3V rail by measuring the voltage drop over its own resistance and zero in on the location of your "short".
 

Online nali

  • Frequent Contributor
  • **
  • Posts: 732
  • Country: gb
Something doesn't add up in your figures: If you apply 100mV then 5 ohms will draw a whopping 20mA which shouldn't give any appreciable voltage drop on a power bus. Or, if you're using a supply operating in constant current mode at 0.5A you should get about 2.5V if you do in fact have 5 ohms across the bus.

Are you sure the 5 ohms measurement isn't an artifact of whatever else is on the bus?

If you *do* have a partial short as others have said apply 3.3v and see what gets warm. If you don't have a thermal camera try the alcohol method or maybe freezer spray (or air duster turned upside down) and see what defrosts first.
 

Offline bob91343

  • Super Contributor
  • ***
  • Posts: 2675
  • Country: us
A 4-wire Ohmmeter moved around will usually pinpoint the trouble spot.  Make careful resistance measurements that can resolve milliohms and you can find the lowest resistance place.  Use a stylus to make consistent contact, or whatever works best.  All this with power disconnected.

Another way is to pump a small current in and measure potentials around the board.  The lowest potential is closest to the short.  This in millivolts or lower.

The brute force method of putting in lots of current and looking for smoke or, at least, heat, will work but might destroy good parts.
 

Offline Apollyon25_

  • Regular Contributor
  • *
  • Posts: 66
  • Country: nz
I normally just use a current limited supply and set a high-ish current, then feel around the board for hot spots. Or if you have one, use a thermal camera.
Once the hot spot is found either replace the part/solder blob etc, or if its copper related but not easily accessible, I'll try to splat it clear with a charged cap across the short's approximate location... usually works 95% of the time.
 

Offline Niklas

  • Frequent Contributor
  • **
  • Posts: 404
  • Country: se
I have seen similar resistance readings on cracked 1206-sized ceramic capacitors. It was located on the output of an LDO, so the current was limited and it did not give a clear temperature hot spot.
 

Offline xavier60

  • Super Contributor
  • ***
  • Posts: 3034
  • Country: au
With laptop main-boards, if the short is a very low resistance, I use about 0.5A and trace the shorted component by reading voltage drops across the ground plane rather than the actual shorted rail because it's usually to difficult to follow.
The method works best for finding shorted MLCCs, not so good for when the short involves  large BGA ICs.
HP 54645A dso, Fluke 87V dmm,  Agilent U8002A psu,  FY6600 function gen,  Brymen BM857S, HAKKO FM-204, New! HAKKO FX-971.
 

Online voltsandjolts

  • Supporter
  • ****
  • Posts: 2549
  • Country: gb
Start with a close and thorough visual inspection, preferably with microscope.
 

Offline IanJ

  • Supporter
  • ****
  • Posts: 1818
  • Country: scotland
  • Full time EE & Youtuber/Creator
    • IanJohnston.com
6.5 digit multimeter, resistance mode and probe around the board with respect to the most local GND to where you probe and the closer you get to the problem the lower the resistance will be.

Ian.
Ian Johnston - Original designer of the PDVS2mini || Author of WinGPIB
Website: www.ianjohnston.com
YouTube: www.youtube.com/user/IanScottJohnston, Odysee: https://odysee.com/@IanScottJohnston, Twitter(X): https://twitter.com/IanSJohnston, Github: https://github.com/Ian-Johnston?tab=repositories
 

Offline xavier60

  • Super Contributor
  • ***
  • Posts: 3034
  • Country: au
Also, with about 0.5A of bias current, a shorted MLCC can be confirmed by measuring between the solder fillet and the top of the termination cap.
The DMM needs to have 10µV or better resolution.
« Last Edit: December 05, 2019, 09:31:31 am by xavier60 »
HP 54645A dso, Fluke 87V dmm,  Agilent U8002A psu,  FY6600 function gen,  Brymen BM857S, HAKKO FM-204, New! HAKKO FX-971.
 

Offline muthukural001Topic starter

  • Regular Contributor
  • *
  • Posts: 211
  • Country: in
Hi All,

I checked the impedance everywhere on the 3.3V path, it is 4.7 ohms but actually it should be around 3.7K. @nali,I am not getting your point that it should drop 2.5V if my supply can give 0.5A. I set 100mV and max current limit is 0.5A in my power supply. The 100mV is not going down at power supply terminal and the power supply current display tells the load draws only 20mA when I connect my board to the power supply.


Thanks,
Muthu
« Last Edit: December 07, 2019, 12:38:13 pm by muthukural001 »
 

Online nali

  • Frequent Contributor
  • **
  • Posts: 732
  • Country: gb
Hi Muthu

Sorry, I wasn't sure if you meant you were either supplying a constant 100mV to the board, or using a PSU in constant current mode to force 0.5A to look for something getting hot.

As you're supplying 100mV and the board is taking 20mA - are you saying you're measuring 100mV at the PSU yet only 70mV on the board? That's quite a voltage drop for such a low current, although it's not really relevant to your problem.

It may help if you let us know what test equipment you have access to e.g. if you have a high resolution DMM you can look for voltages or voltage drops, or if you only have basic equipment then the "get warm" method is probably your only approach.

Personally, as it's not a direct short I'd just apply 3.3V after using an air duster can held upside down to cover the board in frost then the offending area should warm up pretty quick. A 5 ohm load will only generate a couple of watts heat enough to create a hotspot but not high enough to cause damage (maybe limit your supply to 500mA in case the bad part gets worse)


 

Offline james_s

  • Super Contributor
  • ***
  • Posts: 21611
  • Country: us
Set the power supply to 3.3V with .5A current limit, 100mV is much too low to push enough current through the partial short. Using 3.3V will cause something to get warm and should make it easy to locate.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf