• State Resolved
  • Locked Locked
  • Replies 2 replies
  • Subscribers 31 subscribers
  • Views 475 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TMP116: Thermal pad connection

Arthur Ohl
Prodigy 125 points
Part Number: TMP116
Other Parts Discussed in Thread: LMT01, LMT87, TMP468, LMT70, TMP108, TMP468EVM, TMP144, TMP411

Hello everyone,

In one of my design, I need to sense MOSFET temperature or/and inductor temperature (TDSON package) which is NOT at ground potential. I first used a resistor divider with a NTC connected to the uC ADC. This solution works pretty well excepts it is dependent of the voltage on the voltage divider and it requires the transport of an analog signal in a noisy environment. I'am thinking replacing this solution by an I2C temperature sensor thus reducing the number of signals (cause I have multiple MOSFET temperatures to measure) and improving the accuracy of the measure.

I notice that the TMP116 had a thermal pad which is not connected to the ground, I thought connecting it to the drain via a copper track to the MOSFET could be a good solution. However, the datasheet clearly states that the thermal pad should not be soldered (at least not to something else than ground).

Does my solution makes sense or should I use an other sensing technique like remote sensing with a diode which would requires the use of an isolated thermal interface (thermal sheet)?

Thank you,

Arthur

  • 0
    TI__Mastermind 35835 points

    Hi Arthur,

    It's true you cannot connect the TMP116's thermal pad to a voltage potential other than ground (0V.) If your MOSFET heatsink is not at 0V potential, you could still use copper to aid in thermal transfer from the MOSFET to the TMP116's thermal pad without making electrical contact. For example, it's common for several vias to be placed under a MOSFET to carry the heat away. The sensor could be electrically connected to internal copper which flows around these vias but does not contact them.

    We're currently recommending that customers not solder the thermal pad on TMP116/7 at all due to shifts that were observed in temperature offset when the thermal pad was soldered. The shifts observed varied based on assembly process, and were less than 0.1C in all cases. If you can accept this accuracy, or if you have a process for calibrating it, there's no further concerns for soldering the pad.

    If your MOSFET was a through-hole package, I could suggest through hole sensors like our LMT01 and LMT87. Similarly, I could suggest a through-hole remote diode paired with a TMP4xx remote sensor such as TMP468 and 2N3904. These solutions would make use of thermal adhesives to attach them to the same heatsink as the MOSFET.

    Since your MOSFET is surface mount, I wouldn't recommend trying to fold a through hole sensor on top of your MOSFET. It seems like an assembly risk to me, but maybe it can be resolved. That said, there are some very very small transistors on the market which can be used with our remote sensors. Check out the 1.1x0.7mm MMBT3904LP transistors on the TMP468EVM. Placing these very close to your MOSFET may be adequate without any additional layout tricks. We also sell local temperature sensors in similar 1mm^2 sizes, but they are all DSBGA. If you can handle assembling 0.4 pitch DSBGA, check out LMT70 (analog) TMP144 (digital, daisy chain bus) or TMP108 (I2C, 4 address options.)

    thanks,

    ren

  • 0 in reply to Ren Schackmann
    Prodigy 125 points

    Hi Ren,

    Thanks you for your extensive answer, it helps me a lot !

    I'am gonna go for the solution with the MMBT3904LP and the TMP411.

    Have a nice day !

    Arthur