AD598 consumes up to 18mA at +-15V supply voltage which results in a heat disspation of considerable 0.54W. Heat dissipation due to excitation drive is about 50mW, additionally, when chosing the recommended LVDT. So, let's take a worst case heat dissipation of 0.6W.

Thermal resistance between junction and ambient is 80°C/W, which makes the die temperature rise up to 80°C x 0.6W + 85°C = 133°C at 85°C ambient temperature, which is quite a lot!

The increase in failure rate due to higher die temperature can be estimated by the help of Arrhenius equation. The acceleration factor AF, by which the failure rate increases is:

AF = exp (Ea/k x (1/Tu - 1/Ts))

Assuming an activation energy of Ea = 0.7eV and taking k = 8.63 x 10^-5eV/K, Tu = 273K + 85K = 358K, Ts = 273K + 133K = 406K yields an acceleration factor of AF = 14.6. So, the failure rate will increase by a factor of about 15, when the die temperature is 133°C instead of 85°C.

On the other hand, temperature cycling by periodically turning the chip on and off will stress the chip even much more than a permanent higher die temperature! The only remedy is to find a cooler place for the chip and/or to decrease the heat dissipation by choosing a lower supply voltage and/or by cooling the chip in a suited way (fan, heat sink).

But don't forget, that the AD598 is designed to be very reliable. Wafer tests at 55°C have shown a mean time between failure (MTBF) of 3,122,147,185 hours (60% confidence level). Using the Arrhenius equation again translates this to a MTBF of 393,067,365 hours at 85°C die temperature. So, at 85°C die temperature the MTBF is about 45000 years and at 133°C about 3100 years.

Or by other words: After a life time of 10 years at 85°C die temperature you will observe about one of 4500 devices containing a damaged AD598, and one of 310 devices when running the AD598 at 133°C die temperature. I guess, when using power cycling, on the other hand, all will be damaged after 10 years...

Kai