FR4 time*temperature discoloration

If you are concern, there is also FR5 to use. that is a higher temperature material and also higher voltage handling.

P.S. It does cost more, obviously..

Jamie

Use black solder mask.

Cheers

Phil Hobbs

Excellent suggestion! Thanks.

A good way of hiding it, but what if forced to use the typical green color?

Cheers, John S

Space hot things, like wirewound resistors, up off the board.

Put a topside copper pour below them. And use really big multilayer pads/pours for the terminations, to spread out the heat.

r?

remove the solder mask underneath the hot part?

-Lasse

It's the FR4 that discolors. Better to dump a big copper pour under the hot part, with thermal vias to other-layer planes. Best would be solid copper, nickel-gold, no mask.

My experience is that solder ages slightly faster than FR4, and copper traces age almost as fast. Your customers are likely right to call boards defective when they have dark spots.

It was common in 1980s home theater systems to have extremely hot voltage regulator transistors that eventually destroyed their solder joints. You could probably find one and measure the temperature. I'd guess nearly 200C based on what they do to your finger tip.

I looked into this some years ago. The only information I could find was submitted by Isola, referring to UL746 which has figures for the lifetime versus temperature:

It does not however state the point at which the discoleration occurs

The rule we used for FR4 temperatue was max 95 degrees, which would be on the conservative IMHO.

Regards

Klaus

On a sunny day (Sun, 22 Dec 2013 16:43:49 -0800) it happened Kevin McMurtrie wrote in :

In the old Ampex broadcast video recorders the tubes that drove the recording heads were mounted upside down, when the fans stopped working and the air flow detector did not work (often) the tubes unsoldered themselves and fell to the bottom, no more signal.

Indeed when FR4 get colored, the solder joint are probably damaged too. Better to have power transistors and other hot components, like power resistors, mounted against the metal housing with some wires to the board. more expensive, but lasts forever. Not always possible, in RF for example, then use heatsinks at the spot.

Many thanks. This is a number to keep in mind. Every degree lower is better, I know.

Cheers, John

I've also made some preliminary searches, and I am surprised at the lack of accessible information regarding this subject.

I've tried to find graphs of loss in weight as a function of time and temperature and failed.

Another aspect of discolouring is that any charring can give rise to a conductive path, and have know plastics to become conductive after an arc has carbonised the surface.

Thanks to all who contributed. I think I have a goal now.

Merry Christmas.

John S

there is something about the resin if FR4 starting to change state at ~130C causing a big expansion in the Z direction stressing vias etc.

-Lasse

So: I don't know the answer for this, but reading Kevin's comment about high temperatures accelerating the breakdown of solder made me think.

There's probably a MIL- handbook out there, possibly with the words "reliability" and "electronic" in its title, that addresses all of this. Maybe this would help narrow your search enough that you'll find an answer?

I know that there are pretty firm rules for computing the reliability of a PCB given various parameters: I used to work at a place where one of the steps in the design process was to give the schematics and PCB layouts to Leo (uh -- per ISO 9000 that's "the reliability guy", but we all said "Leo"). Leo would then withdraw into his cubicle for a period of time marked by odd-scented smoke, purple lightning, and strange cries, after which we'd get a completed reliability estimate. The janitorial staff would clean up the chicken entrails later.

I'm sure that the standard that he was working from in such a scientific manner included how to treat hot components, and I'm sure that you could work backwards from that to how to reduce the heating from same.

Probably a MIL-HDBK-217 analysis. I've done just one full FMECA analysis and was successful in getting accepted by the planning yard. I swore I'd resign before doing another. The other big standard I'm aware of is the commercial standard for Bellcore/Telcordia.

The base 217 does contain the necessary tables and factors but nowadays nobody runs the numbers by hand, it's usually a mildly expensive package like Relex (now sucked into the PTC borg as "Windchill"). The full 217 is out there for download.

The expansion of the resin would be the same in all directions, except the glass fibers limit the dimensional change in the X & Y directions due to the layup.

One rule of thumb to prevent delamination is to keep the operation

premium, of course).

Best regards, Spehro Pefhany

Wow! Thanks. Best info yet! First I need to find a chicken. Then I need to find out if my cleaning lady understands to clean up what's left after... er, what do I do with the chicken?

Heating a chicken is no problem. There are a number of ways. But, don't tell the ASPCA.

We do MTBF calculations when customers request same, using 217 or Bellcore numbers. One of our guys uses a spreadsheet, and it's no big deal.

I don't think the numbers are especially useful. A good design will have a field failure rate 20x better than calculated, and a bad design can be much worse than calculated. But some people want to see it.

As Phil pointed out, adding a thermal or current limiter, or ESD protection, increases the parts count and makes the calculated MTBF worse.