repairing broken traces on a PCB

[electronic-eng.com]

Well, I have a board here worth $100 perhaps, that certainly passed
all tests, but fails now. When I bend it, it works. I should add, that
after bending it ~15 times it stayed in failure mode. I found one
decoupling cap standing in the air, some flux rubbish (?) between a few
legs of a TQFP, but nothing that causes the failure. Ditched it
already.

I wonder what the dropout percentage is for tested boards, and if
it is worth paying for. Difficult to anwer of course, I simply have
excellent experience with untested boards, even with 7mil tracks/distance.
Seems that more goes wrong with the assembly of parts, imo
.

Hi

I debugged a similar intermittant failure a few years ago: Visually
inspected the PCB and solder joints with a microscope, found nothing.
Touched up all joints with a soldering iron anyway because I couldn't
see under the legs, no effect. As the via's weren't covered over with
solder resist I reflowed them with a little extra solder, this fixed
the problem. I had the via cross-sectioned and found cracks in the
barrel. The PCB manufacturer was drilling the via's with a 4 board
stack-up, which meant the bottom board was suffering burring from
excessive heat on the drill bit going through the boards above. They
changed it to drilling single boards and when we cross-sectioned the
new batch we had picture perfect via's.

Alan
www.electronic-eng.com

 

[Frank Bemelman]

Hi

I debugged a similar intermittant failure a few years ago: Visually
inspected the PCB and solder joints with a microscope, found nothing.
Touched up all joints with a soldering iron anyway because I couldn't
see under the legs, no effect. As the via's weren't covered over with
solder resist I reflowed them with a little extra solder, this fixed
the problem. I had the via cross-sectioned and found cracks in the
barrel. The PCB manufacturer was drilling the via's with a 4 board
stack-up, which meant the bottom board was suffering burring from
excessive heat on the drill bit going through the boards above. They
changed it to drilling single boards and when we cross-sectioned the
new batch we had picture perfect via's.

Hmm, worth a try. I'll try fishing it out of the rubbish bin then, and
see if there not too many vias and resolder them, perhaps. I didn't
think about the vias this time.

 

[colin]

Well, I have a board here worth $100 perhaps, that certainly passed
all tests, but fails now. When I bend it, it works. I should add, that
after bending it ~15 times it stayed in failure mode. I found one
decoupling cap standing in the air, some flux rubbish (?) between a few
legs of a TQFP, but nothing that causes the failure. Ditched it
already.

I was frustrated when I had this problem with an smd prototype after
changing a few components several times by hand, I too found flexing the
board can make it not work as well, ive found that ceramic smd capacitors
can go half open circuit, acting like a bend sensitive variable capacitance
!

I only realy got to the botom of this when I found a capacitor that changed
in value the tighter i held it in the probes of the capacitance meter. took
me a while to work out the connection.

Colin =^.^=

 

[electronic-eng.com]

I was frustrated when I had this problem with an smd prototype after
changing a few components several times by hand, I too found flexing the
board can make it not work as well, ive found that ceramic smd capacitors
can go half open circuit, acting like a bend sensitive variable capacitance
!

I only realy got to the botom of this when I found a capacitor that changed
in value the tighter i held it in the probes of the capacitance meter. took
me a while to work out the connection.

Colin =^.^=

Hi

Ceramic capacitors are notorious for cracking on a PCB subjected to
warping or shock, particularly around the outer edges. You should try
to place the footprint a little in from the edge as good design
practice. You will not usually notice the issue electrically if it is a
decoupling cap as the others on that supply are making up for it, also
the cracks usually take place where the ceramic joins the end cap and
are not visible under high magnification. Be sure the operators are not
misshandling or using a rough tool to remove the PCB from the panel
after assembly. A company call Syfer (www.syfer.com) manufacture a
'flexicap' which uses a flexible epoxy to join the ceramic to the end
cap therefore making them more resiliant to flexing. I handled an issue
with a cracked SMT cap on roughly 1000 PCB's by replacing the standard
cap with a flexicap so the PCB's could be used in production, while the
operators were doing that I moved the footprint in more from the edge
and didn't have any problems from the next batch on.

Alan
www.electronic-eng.com

 

[przemek klosowski]

More interesting is to hear about experiences in finding hairline cracks
in tracks.

I bought a signal generator that failed to power up. I checked the
power supply and it was OK, but the circuit wasn't energized. I followed
the power trace, and realized that the voltage stops on a fat trace near
the edge of the board, right after the area with a mounting screw.
I had to use a magnifying glass to see the crack---apparently the unit
was dropped and PCB cracked. I soldered a piece of wire across the
crack, and it all works now. I was lucky that it was a fat trace, and
that it was the supply and not some internal circuit.

 

[colin]

Hi

Ceramic capacitors are notorious for cracking on a PCB subjected to
warping or shock, particularly around the outer edges. You should try
to place the footprint a little in from the edge as good design
practice. You will not usually notice the issue electrically if it is a
decoupling cap as the others on that supply are making up for it, also
the cracks usually take place where the ceramic joins the end cap and
are not visible under high magnification. Be sure the operators are not
misshandling or using a rough tool to remove the PCB from the panel
after assembly. A company call Syfer (www.syfer.com) manufacture a
'flexicap' which uses a flexible epoxy to join the ceramic to the end
cap therefore making them more resiliant to flexing. I handled an issue
with a cracked SMT cap on roughly 1000 PCB's by replacing the standard
cap with a flexicap so the PCB's could be used in production, while the
operators were doing that I moved the footprint in more from the edge
and didn't have any problems from the next batch on.

Alan

Thanks, its interesting there are components more resiliant to flexing, as I
said its only a problem in initial prototyping where ive had to
change/resolder a component several times, and its a hand soldered hand made
pcb, I often find the end cap comes off, sometimes I just cant get the
solder to melt at both ends of the component with one soldering iron,
sometimes think I shld make a pair of tweezers out of two fine soldering
irons, maybe these exist already ?

I hate to think how the 'through hole' decoupling capacitors ive used would
cope with any stresses that they might endure, fortunatly mass production or
even long term reliability isnt required.

I think I might look for an alternative to smd/ceramic caps for some
instances as im trying to eliminate sensitivity to even small vibrations.

Colin =^.^=

 

[electronic-eng.com]

Thanks, its interesting there are components more resiliant to flexing, as I
said its only a problem in initial prototyping where ive had to
change/resolder a component several times, and its a hand soldered hand made
pcb, I often find the end cap comes off, sometimes I just cant get the
solder to melt at both ends of the component with one soldering iron,
sometimes think I shld make a pair of tweezers out of two fine soldering
irons, maybe these exist already ?

I hate to think how the 'through hole' decoupling capacitors ive used would
cope with any stresses that they might endure, fortunatly mass production or
even long term reliability isnt required.

I think I might look for an alternative to smd/ceramic caps for some
instances as im trying to eliminate sensitivity to even small vibrations.

Colin =^.^=

Colin

You should only solder one side of a ceramic chip capacitor or resistor
at a time. If you solder both sides simultaneously, upon cooling, the
constriction of the component and joints can cause cracking. Solder one
side, then after a few seconds solder the other. If you are changing a
component, get rid of all the old solder from the pads first, and apply
with new solder.

Alan
www.electronic-eng.com