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.

Strange humidity sensitivity on MSP430F5328 based product when in LPM3 mode

Other Parts Discussed in Thread: MSP430F5328

We are using the MSP430F5328IZQER as the controller on multiple products and each of these products require low current consumption.  When in LPM3 all these products pull 7 uA (which is what is expected) but this is only the case in a controlled environment with very low humidity.  With modest humidity LPM3 current can go over 100 uA - a slow drift (30 seconds) for the current to reach 100 uA.

We are using an external 32 KHz crystal which is the clock source for the wake-up timer when in LPM3 mode.

Has anyone else seen anything like this using any of the MSP430 controllers or have a solution?  

  • When you mention that the current remains at 7uA in controlled environment, how long have you tested the same. My personal experience is that when there is a current rise over a period of time some input which needs to be initialized has not been done. Recheck if all port pins have been initialized properly.
  • Thanks for the comment. Indeed, we rechecked, more than once, that all inputs have either pull-ups or set to outputs when in LPM3 mode. I don't think we have missed anything here unless there is something unclear in the data sheets and reference guides for this controller.
  • Dear Michael,

    Here are a few inputs and suggestions. I am sure you must have tried them out already, but this is just for refreshing:

    1. What is your normal running current? Is it in the range of 100uA?

    2. Have you verified that the controller is in the sleep mode when the current is in the range of 100uA?

    2. If it is so (and not so, if the system has multiple Voltage sources) then there is a possibility that the controller may be coming out of the sleep mode. This may be because of a spurious Wake-up signal.

    3. To eliminate the spurious wake-up signal as a source do this: Modify your software in such a way that once it goes in the LPMxx, there is no way it can come out. In that case conduct your experiment for humidity situation. If in that case the current does not increase then we can pin-point the culprit.

    Kind Regards,

    Aditya Ayachit

  • Hi Aditya:

    Our normal running current is in the 2 ma range due to running at 8 MHz when awake.   But you have given us some ideas here - we will look for features and sub-systems not properly disabled when going into sleep mode to see if there has been anything missed.

    Thanks,

    Mike Pedigo.

  • We are using the BGA version of the MSP430F5328 (MSP430F5328IZQER) and it has 15 pins called out as "reserved." We assumed these were no connects and just noticed a in the data sheet footnote that they are to be grounded. Question for someone at TI: If these pins are not connected to ground will this lead to the high LPM3 current that we are seeing?
  • Hello Michael,

    Please refer to this thread along a similar topic: e2e.ti.com/.../19777

    The pins in question are basically "dummy pins" that are not connected to any MCU functions or peripherals. They are recommended to be connected to GND generally for mechanical stability but there is the possibility that pick-up noise with adjacent BGA pads is resulting in leakage current in the uA range that would be avoided if the pins were grounded. This has not been investigated heavily so there is currently no data to support this theory.

    Another possibility involves leakage current introduced through the XT1 pins connected to the crystal oscillator. There have been several cases where humidity causes oscillator failure but with proper capsulation (i.e. potting) these problems were fixed. Leakage may also be amplified by humidity if there is contamination present at the pads, usually from flux, but in the case of a BGA package it would be hard to clean the board in order to check. At any rate I recommend implementing a PCB cleaning process and capsulation, specifically around the MSP430 and crystal, to see if the leakage current can be eliminated without grounding the reserved pads.

    Regards,
    Ryan
  • Hi Ryan:

    Thanks for the information.  The circuit is on a rather small board and we dipped the board (including the processor and xtal) into an epoxy compound that is normally used for "chip on board" applications.  It has been running since then without any of the undesired increase in LPM3 current.  We suspect the problem is the XT1 pins as you mentioned since we also observe especially high sensitivity to moisture around XT1.

    So at the moment it looks like the epoxy dip is a solution to this problem, but we are doing further long term testing.

    Regards,

    Mike Pedigo.

  • Hey Michael,

    You might want to check if there are any flux residues around the pads. Usually the residues are neuturlaized during the reflow process, but depending the wash and reflow process, ions could be left behind. Once the water vapor gets on, it'll just connect the dots. Depending on the application, you can encapsulate the chip with stuff like Elpeguard. Beware that some encapsulent has to fully cured in a controlled environment before storing, or else the chemical in the encapsulent it self can out gas and cause corrosion. My previous company lost a big of money because they wanted to save by using a cheaper encapsulent.
  • Albert.

    Thanks for the information.  Was there some specific encapsulate product that was found to be effective, or the most effective?

    Mike Pedigo.

     

  • My old company switched to www.humiseal.com/.../1b73
    By the way, I have a MSP432 Eval board cleaned with alcohol and acetone. In LPM3 with GPIO ISR enabled, the current is around 1~2uA. If I breathe on it, it can jump up to 40uA for couple seconds.
  • Hi RF guy:

    It looks like these microcontrollers are sensitive to humidity. We discovered the most humidity sensitivity is around the XT1 circuit. Maybe the mechanical vibration of the 32 KHz tuning fork piezo oscillator is opening small cavities into the packaging exposing the die. Just a theory.
  • Hi All!

    This discussion has been very fruitful. This is the kind of information, that you do not get from datasheets!

    Aditya

**Attention** This is a public forum