LMT70: thermal information

Felix,

What are your operating temperature range, operating supply range, accuracy requirements. How many parts do you plan on calibrating in production if you go with analog sensors? Calibrating each device seems to be expensive for production scaling. I would recommend looking at digital output sensors since the system accuracy is guaranteed (e.g. TMP112, TMP108, TMP103, LMT01, these have ~1 sec thermal time constant and are available in a variety of packages).

You are on the right track to look at precision references since the system accuracy can't be better than your ADC reference. There are many design considerations for an analog sensor design you must consider in order to get to the system accuracy you desire (ADC specs like INL, DNL, offset, gain error; voltage reference, supply noise, layout, input bias current offset, EMI etc.).  So if you can avoid analog go with digital.

What is the time frame of stability/drift are you looking for (e.g. 5 or 10 years) and the stability/drift specs?

Thermal conduction will be dominated by the metal leads or solder bumps (depending on package type). Using copper planes will help with the thermal response. The thermal conductivity of copper is ~300x higher than the package mold compound just to give you a sense. So you want to design the layout with that in mind.

A few other things that might affect your measurement stability are convection and radiation.  The faster the air flow the smaller the thermal time constant (faster settling).

-Kelvin

The operating temperature range is very limited, certainly less than 0 to 50 C, maybe as small as 15 to 30 C. The present supply voltage seems to stay stable to a few microvolts, maybe better. This is a very specialized piece of test equipment we are building. We normally build computer controlled systems to calibrate such equipment and some of the more complicated calibration runs take 12 hours. So, process time is not really an issue. The entire system specs will be determined by how accurately we can resolve changes in temperature of the main pressure sensor. The time frame for stability is about 1 year. That is our normal recommended recalibration schedule. But what I really need to know is whether it is better to heat sink through the pcb or add another heatsink to the top or side of the device to better heat sink it to the pressure sensor we are calibrating. It appears from the data for this particular sensor, the LMT70, that there is a much lower thermal impedance path to the outside through either the top or the sides of the device. In fact, in one of the TI application notes it recommends adding "underfill material" that not only goes underneath, but partially up the sides of the device.

Thanks for the information.

Regards,

Felix

The most thermal conductive side of LMT70 is the bottom as indicated in the app note.  The underfill material improves the conductivity from the pcb to the die by removing the air gap.  I would recommend placing the heat sink at the bottom of the pcb and add copper planes with vias to optimize conduction to the die. Also use thermal grease.

I will need to follow up with the stability spec and get back to you.

-Kelvin