A maybe not so well known error mechanism in mainboard circuits.

[ZMacZ Furreh]

Most people involved in electronics already know that metal expands when it's heated.

They also know that it happens on a circuitboard, and that under normal circumstances this is normal and

won't cause breakage.

But...

When a circuit is being heated a very long time (without pause), and itself is relatively long,

the metal will force the cohesive layer which attaches it to the circuitboard to comply with it's lengthening..

Then, when the device is shut off, the metal circuit will cool down, and will try and force the cohesive layer to comply again..

The cohesive layer will not comply this time, and tension increases, until the circuit breaks..

This happened to me yesterday when I switched off my monitor yesterday after having been in continuous operation

for several months/years.

Since it's systematic failure, albeit a slow one, I decided to post here..

Think of it what you will, I won't monitor, since my interest in the replies is not high..

I also added the solution (fix) for this, which is a simple one, and will try a situational explanation of it as well..

Laterz,

ZMacZ Furreh..

Keep in mind, that the the longer the duration of the heating effect (power supply, processor, or whichever raises the local temperature),

the greater the effect, as well as the length of the circuit.

Also, fast cooling down amplifies this even further..

basically Effect = diff Temp x length circuit x duration of operation x (r)coolrate..

(since it takes months for the cohesive layer to comply, the effect of the fast cooling will amplify but not as much as the

diff Temp, length circuit, or duration of operation..hence the modifier (r)..)

 

What I persoanlly learned from this is that devices may heat up quite a bit over long durations, np, but the problem is with the cooling down..

In order to completely nullify the effect I'd say roughly a cooldown of equal measure, if at all, or no cooling down at all..

Other solutions may involve less sticky cohesive layers that allow for 'drift'..(albeit minute drift..)

(You don't want ur circuits floating off into the horizon..which could be fun to watch once though..)

 

 

Alternately, you can employ S shaped 'breaks' in the straight lined circuits..which looks snakey..

(- does not like snakes nor cucumbers..except for the latter finely chopped..)

 

(NOTE: I reversed the importance of the rate of cooling and the duration in the picture..sowwies.. '^^'..)

Edited May 29, 2018 by ZMacZ Furreh
(UGH TYPO'S !!!)

[Strange]

1 hour ago, ZMacZ Furreh said:

Keep in mind, that the the longer the duration of the heating effect (power supply, processor, or whichever raises the local temperature), the greater the effect

It is well know that heating and cooling can causes stresses between materials with different thermal characteristics. I don't see why the length of time would be significant. The general advice is to avoid turning systems on and of to minimise the effects of thermal stress.

I think you would need some good evidence that this is the case, rather the failure of one old monitor.

[hoola]

I have never seen a copper circuit expand in my repair business, but the substrate board might at some point, shrink. I have never noticed that other than that the board can burn at a failed junction if the joint is in a high current location, if you can term that "shrinkage". What I have noticed is that the solder seems to evaporate, not that I have direct evidence of that, other than seeing it thin out in a ring, circling a high current connection. As the solder mass shrinks, the temp accordingly goes up due to increasing resistance, hastening the process of joint failure, causing an open circuit. I have seen an instance or two of a situation where I soldered loose connections, only to have the set come back 10 years later to me, and the solder had done a similar "disappearing act" at the same locations. Also, general heat seems related to the copper adhesive to loosening up, sometimes finding them loosened near the failed joint. Never have I seen where the copper had extended itself any observable amount. I have noticed that all solders are not the same. There are differing qualities and have recommended avoiding cheap solders that may have contaminants, which seems to be unhelpful in a long lasting connection repairs...actually this topic is more suited to be in amateur science, than in classical physics...

Edited June 19, 2018 by hoola

[ZMacZ Furreh]

Thermal Expansion Coefficient:

Copper expands @ 17E-6/K
Resin mainboard expands @ 4E-6/K

Difference in expansion 13E-6/K

So, yes, the copper expands more than the mainboard's resin..
And thus the part that 'glues' the two together will have to give as
reaction to this expansion.

Over time it will settle at the new length of the copper, and then
rapid cooling will break the copper circuit.

It's fact, not supposition.

The circuit was repaired later by bridging the connection alternately..
(And it was truly broken, with this error mechanism being the cause..)
(And yes, the 'fix' can help alleviate such breakage, since chance for breakage
is directly linked to the length of the circuit, in a straight line,
across the mainboard.)
(The longer the circuit the greater the sensitivity for rapid temp fluctuations..)
(It's already being tested and has already shown appreciable alleviation of the error mechanism..)

Laterz..

Note: I will not be watching this any longer..

 

@ hoola: It's not visible to the naked eye, the amount of stretch/shrink required to sever the circuit is in the order of microns.
Re-examine the figure, and you'll figure the reason why..
You'll also see why the zig-zag will prevent a circuit from beraking when it's cooling rapidly,
or at least you oughta be able to..
Also, the srinking of the resin happens very slowly over time as it tends to dry out still more,
thus shrinking..but that's  a 'permanent' shrinkage, which doesn't involve the cooling 'cycle'..
In engineering for circuitboards meant for use in space, the 'zig-zag' may prevent many repairs..
In space things heat rapidly, while being on the nightside, it will cool also rapidly..
The only alleviation of heat in space is heat emision, through light, not heat transfer,
like atmospheric, and yet, sunlight or no sunlight will still rapidly heat or cool..

Anyways, it's usefull, even in Earthly settings, where the temp delta is well in the range
of 60K, specially with stuff like fans around..

I also believe that this error mechanism may also occur on Si-SiO processor types,
but in a different way, more like contributory to electron migration.
(electron migration can be mitigated this way, and thus also processing equipment
for use in outer space may be 'hardened' a little that way, albeit
at a cost of a tiny bit of processing power, in the order of 2-3%..
But THAT right now for me is just a calculated guess..

For those who are interested:
A very long circuitboard (1 meter long with a single circuit, in a straight line, and one more
with the 'zig-zag'heated to 363K, over a month, then rapidly cooled..)
The straight one failed, the zig-zag held..
Each circuit was 0.5 mm wide..

The stretch was visible to the naked eye on the test subject.(I didn't measure it though,
I was more interested in the result of the rapid cooling..)

 

Edited June 26, 2018 by ZMacZ Furreh
more addition..

[Enthalpy]

On 6/26/2018 at 11:34 PM, ZMacZ Furreh said:

Copper expands @ 17E-6/K
Resin mainboard expands @ 4E-6/K

Wrong.

Epoxies expand much more than that, but the proportion of glass fibres in boards is adjusted so the resulting thermal expansion is 17ppm/K to match that of copper.

[swansont]

1 hour ago, Enthalpy said:

Wrong.

Epoxies expand much more than that, but the proportion of glass fibres in boards is adjusted so the resulting thermal expansion is 17ppm/K to match that of copper.

In the x and y directions. But the material can't just not expand as it wants to, so it has a larger coefficient in the z direction. The situation gets worse if you go above the glass transition temperature.

 

see e.g. http://www.frontdoor.biz/HowToPCB/HowToPCB-Thermal.html

[Enthalpy]

The original post was clearly about the in-plane directions.

It is also about long operation of a circuit, which takes place below the glass transition temperature.