PCB track fault - infuriating one

It could be a combination of the things you mentioned all working together, each contributing to the failure.

Reflow of the joint sometime in the past could weaken the copper adhesive; age-related deterioration of the adhesive, or long-term heating of the pad by a nearby component could also weaken the adhesive.

Another possibility is that the component was pushed around physically on the board in the past, which would stress the pad.

HTH

sing

or the flux was a tad too active and thinned the copper, or the solder over-hot and dissolved same

That's the most likely cause. This happens a LOT with connector pins that go straight into a PC board, and very rarely with connector pins that take a right angle bend before going into a through-hole. Large components (like electrolytic capacitors) can be misseated or can dislodge during wave solder and that makes subsequent strain on the solder pad. Vibration after the manufacture and test causes the failure.

On Tue, 22 Dec 2009 11:31:46 -0000, "N_Cook" wrote:

:Intermittant cutting out of an amp , something to do with the uPC1237HA :protection IC. Number of control inputs (o/p DC fault sense etc) that takes :o/p low of main speaker relay which clicks over. In fault mode this o/p goes :to 60V via the relay, all quite normal and in spec but for a small SIL IC :with 3.4V on the supply pin, a bit disconcerting. Incidently what does Vcc :mean wnen an IC is rated for 60 volt Vcc (from the NEC datasheet) but 3.4V :the recommended supply voltage : :Next time coming across similar fault I will change the 100uF ,3 second :power-on-delay timer , temporarily, for 1 or 10uF as although the main fault :would operate immediately it would take 3 seconds to reset making tracing :infuriating as not immediate on/off. : :Changed the IC and caps around the IC and just about to change Rs around it :although nothing has checked bad - still the problem persists. I decided to :change the 4.7uF electrolytic (power off muting one) for the second time :(most twizzle sensitive to inducing fault near this C) and this time the :solder pad came away with the molten solder. :Of course everything all nearby and elsewhere concerning soldering and pads :looked fine and tested for continuity fine. : :Perfect ring of solder with a white pad (like overlay printing) : showing under the solder blob , with the cap wire still in place passing :through. Track , perfectly wide about 2mm, had failed in a neat circle :around the pad. Only ever touching contact between track and solder pad or :resonance vibration effect tearing in use ? bearing in min only a very small :electro. Only 2V or 0V on this cap and less than 20mA , so I assume :manufacturing fault. What led to the fault and what to visibly check for (if :possible) with other pads before returning amp to owner ? :

On a particularly badly designed smps in the old Westinghouse Canada W842 terminals I often noted similar intermittents. In the end I resorted to resoldering every pad and this revealed poor soldering technique or process.

When I touched the soldering iron tip to the pads and melted some new solder with resin core flux the old solder immediately ran back from the copper pad into a blob around the tip. The copper pad had a greyish oxide film over it thus telling me that either insufficient heat and/or flux had resulted in a failure to wet the copper pad when it was initially soldered. While the pad had obviously been clean enough to cause the solder to cover the bare copper so that it looked normal, over the years air had permeated the inadequate bond between the solder and the copper pad and caused oxidation which then resulted in intermittent connections.

takes

goes

Vcc

fault

it

to

pads

or

small

(if

process.

solder

pad

it thus

failure

so that

between

That seems about the likely cause The components looked hand populated despite 2005 manufacture, including resistors canted over and touching one another. A propbably related effect is the solder run reinforcement for current carrying along the main rails. There are bare sections with what loooks like oxidised dull brown copper instead and looks like fractured ends to the remaining solder. I found no loose bits of solder inside so probably had dropped away at manufacture, so passing final assembly inspection, all in all a bad job. I will explore the bare copper rail sections a bit more and the saved white faced pad on the C lead.

IC

tracing

around

decided

time

the

passing

for

W842

in

like

white

curiouser and curioser For anyone else coming across similar it is a Kustom KPS PM300, 300W mixer amp, 2005 Been exploring with a microscope. Not broken away pieces of beef-up overlain solder runs along the main rail traces, incomplete soldering in the first place, covered in the overall thin conformal coating, the bare copper going dark brown underneath in just 4 years.

As for the failed C pad This pad had survived original soldering, my first desolder and resolder and came away with second desoldering leaving bare board where the pad was. I expected to see copper colour after scraping into the white circle but only silver colour of solder seen - no copper pad there ever? so thin it disolved ? when I desoldered/resoldered initially there was a normal pad of solder there

Most likely the pad came away with the soldering iron tip on the last reflow. It's probably sitting on your tip cleaning sponge with all the flux residue, or in your vacuum desoldering tool. It's very easy to do and easy to miss.

Anytime you reflow a pad more than once, you have an increased likelihood that this will happen. The copper adhesive isn't intended to be melted over and over...

and

of

the pa and ps share the same hs so I put the 2 caps on the track side of the board, mainly so easier to temporary change the timing 100uF to 1uF if necessary. the 4.7uF smooths the rectified ac for power-off sensing but with it intermittant contact , when that IC pin goes less than 0.7V in next mains cycle, it triggered muting.

Anyway second time of desoldering the small cap, the pad came away, with the usual few grams of force when blob molten. As a wired C , i desoldered with normal soldering iron. Amp back with owner now.

Glad it all worked out.