short circuit in a decoupling capacitor path

[lynx]

Hi

I want to get an opinion...

the graphics card in the photo seems it
has a short circuit in the second row of
capacitors (as we look it from the connector side).

the first row has no issues but the second row measures short between the positive and ground rail...

I have disoldered every cap but outside the circuit it measures fine...there are also some SMD that look like capacitors removing some of them the short still remains the silkscreen names them NC601,2,3 but there are also SMD capacitors named C501,2 3

NC stands for what?

I haven't removed every SMD around the rail since the ground plane must be too large and its hard to desolder ( the soldering iron has hard time to melt the unleaded joint.

any help why I might measure short circuit? I believe smd caps don't fail more often than electrolytic and solid caps

**broken link removed**

 

[Nigel Goodwin]

I believe smd caps don't fail more often than electrolytic and solid caps **broken link removed**

You probably believe wrongly

S/C SM capacitors are relatively common.

Presumably you know what voltage the rail is?, so a common technique is to apply that voltage to the board from an external high current PSU and see what' gets hot', 'explodes', or 'sets on fire'

 

[lynx]

hi thank you for the reply

just now corrected the photo i uploaded, the previous was wrong..

the graphics card has already been tried in a working PC but there was no
output, the GPU remains ice cold..right now i'm not sure i should try it
again on that working PC

i have removed each one of the MOSFETs but the short circuit remains

Is there a possibility i'm looking at a circuit topology that i don't understand and
makes me wrongly think there's a short circuit?

[EDIT]

a very high amount of SMD and through hole caps seem shorted inside the circuit.
there are too many components impossible i desolder each one of them..just to find
out they are okay outside the circuit.

the solid capacitors which are in the problematic rail say 2.5V so that rail
should come out from a regulator? i can't find a <2.5V regulator there's only
one 78M05, can it be a VRM controlling the MOSFETs?

can you help me out... what i'm looking for?

 

[Nigel Goodwin]

Without a schematic you're pretty well stuffed for fault finding, it could be anything.

 

[lynx]

hi Nigel

You are the only helpful guy around?? btw...know you for years thank you for
your advices all this time.

I don't think graphics cards and mainboard schematics are readily available as it happens with other commercial products.

so I guess I couldn't get my hands on the schematic of XFX's implementation radeon 6870.. i have some schematics though from some laptop motherboards etc.

now about the issue with the graphics card I've noticed if I heat certain areas of the PCB the resistance in the short circuit rails drops, after alot of effort I believe the problem is in the area in the following picture.

but still I cannot detect the specific component because after i reflowed the GPU the heat seems it stressed more the affected component and now the resistance is almost not measurable ( few mOhms)

also freeze spray proved handy to locate the problem area..but now neither that can be useful to isolate the specific component.

how can I find it inside this mess of tiny SMDs??


P.S what I'm doing for most computer service shops in my area either they don't know (can't do it), or its not worth the working hours spent, so they only replace the graphics card for their costumer.

for me its outside my job matter..I'm just trying to help a friend in difficult economic situation to save a graphics card once costed 170 EUR.

me I wouldn't spent so much for a graphics card my onboard covers my needs perfectly, but gamers its different story, they need these extra FPS!

 

[tomizett]

I'll share something that's caught me out before (and made me look like an idiot into the bargain!).
Some very low voltage (2.5V, 1.2V etc) ICs appear to measure as a low Ohmic resistance across the power supply. In my case is was a DSP (on a 2.5V supply, I think) - I don't know whether a GPU would be the same, but it seems possible. The resistance I was measuring was of the order of 20 Ohms and varied significantly with temperature.
Once I had eliminated all the decoupling capacitors in the vicinity I concluded that the DSP IC must have failed catastrophically and gone short. Only, of course, it hadn't - the fault was elsewhere in the system.

So, what I'm saying is that unless the short is *very* low resistance (maybe <1 Ohm) it's just possible that it's not a short at all. I don't know whether anyone else can shed any light on this...

As Nigel has said, surface mount multi-layer ceramic caps can and will go short - from what I've read they are very susceptible to mechanical and thermal stress.

 

[lynx]

So, what I'm saying is that unless the short is *very* low resistance (maybe <1 Ohm) it's just possible that it's not a short at all. I don't know whether anyone else can shed any light on this...

hi

very useful what you said... that's why i was suspicious and i asked if i was wrongly think
i'm dealing with a short circuit.

the resistance across the rails was around 2 Ohms (temperature dependent) meaning more cold gets higher <10 ohm and with more heat gets lower <2 ohm.

the area affecting the measured resistance its in the back side of the graphics card exactly behind the GPU, as i'm showing in the previous photo and seems heat directly on the SMDs than on the GPU affects slightly more the resistance.

caps far away from the GPU which seem they are too in short circuit measure around 15 ohm

the short circuit is tied to the black wires of the 6 pin power connectors, i have no way to detect where's the positive side...probably a regulator or a VRM circuit?

 

[tomizett]

2 Ohms seems like a pretty hard short. Can you verify that there is no voltage present across the caps (that is, on the supply rail you believe to be shorted) when the card is powered?

 

[lynx]

2 Ohms seems like a pretty hard short. Can you verify that there is no voltage present across the caps (that is, on the supply rail you believe to be shorted) when the card is powered?

i don't have a spare PSU with 6 pin connectors...there are some resettable fuses on
the PCB maybe they prevented the short circuit before something blows? or maybe not
if the current on these rails is too high, the card is in perfect condition no signs of
damaged components.

as i said i believe the problem is isolated in the area i'm showing in the second photo, problem is i can't figure
out which SMD could be the problem, especially after the resistance dropped to mohms or maybe it could be
a shorted BGA ball from the GPU behind it, maybe a tin whisker, although it shouldn't since i guess it's rare
phenomenon and also the GPU has been reflowed.