maximum trace temperature on FR-4

Question

What is the maximum temperature you would recommend for an isolated trace on the surface of FR-4? (I.e.: not an inner-layer track.) Is there an IPC recommendation? Does it really matter if it is past the glass transition temperature? (What is this value for FR-4? Wikipedia says "above 120C".) I mean, even if it's soft, will the track still stick with high reliability? 'Suppose what I'm looking for is a MTTF vs. temperature (vs. trace area) set of curves, but experienced guesses are encouraged.

This is useful when calculating trace widths vs. temperature rise. I've always limited traces to the lower of about 100C or whatever the connected parts can manage.

[For the Mech-E, Chem-E, and Mat-E: I guess the question can be boiled down to the intersection of shear stress between copper and FR-4 caused by differences in thermal expansion, which would be related to track width, and the decrease in shear modulus as temperature rises.]

Answer 1

This is not my area of expertise so I cannot point you towards any MTBF data but I work with tools that routinely operate in environments up to 175°C so I can tell you from experience what I watch out for.

1. Voltage regulators with thermal protection. These usually cut out about 125°C.
2. Solder. Depends on the lead/tin ratio this can vary greatly. You have to check the rating on each type but you can typically get solder that doesn't melt until around 220°C
3. Components vary greatly. You have to read the datasheets carefully. For instance, a microcontroller might be rated to operate at 125°C but that is 100 MHz. If you only run it at 12 MHz it might run much hotter. There is a good article here that talks a bit more about this.

I haven't experienced any trouble with traces separating from the boards before these other failures occurred.

Answer 2

AFAIK, safety regulations (e.g. UL) require that the temperature on solder joints and traces be lower than 130°C for regular FR-4 material.

Individual parts may become hotter (e.g. a MOSFET or diode in a TO220), but the solder joints (and the traces) must be at or below 130°C.

This maximum temperature of 130°C seems to be reasonable not only with regard to safety approvals, but also with regard to what the board really can take. I have taken boards out of the climatic chamber with some parts' temperature readings at ca. 170°C, and dames en heren, that smell was bad and that board was black!

Answer 3

I suspect you will render any ICs on the board useless due to high temperature long before the copper and FR-4 become a problem.