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.

MSP430FR2633: Suddenly touch becomes impossible

Part Number: MSP430FR2633

Tool/software:

Hi All,

[Currently occurring problems]

1, The sensitivity of the sensor sometimes deteriorates without any particular action being taken.
→The frequency of occurrence is 1 to 2 weeks after the power is turned on for several models.
2, The problem occurs only with sensors that are frequently touched.
3, Once the sensor sensitivity becomes bad, it cannot be recovered unless it is forcibly shut down.
4, Conversion Count = 1000, Runtime Recalibration = Disable
5, The electrode is made of an acrylic plate glued on top of a sheet.

Please let me know how to improve it.

Best Regards,

Ito

  • HI  Ito,

    Operating at higher sensitivities, conversion counts >= 1000 can lead to unwanted behavior.  I suspect what may be happening is the proximity and touch thresholds are not set properly.  If, for example the proximity threshold is too low, it may be triggered by noise.  When a proximity is triggered, the LTA (long term averaging) is temporarily halted.

    Then as the sensor's conversion count continues to change, the LTA is locked and when a physical touch occurs, the current conversion count has drifted to far from the LTA and the touch threshold is not reached.

    Now, when you say the sensors are frequently touched, how frequent?  Once per minute, once per second, several times per second...?

  • Hi Dennis,

    Thank you for your reply.

    What are the negative effects of an excessively high conversion count?
    Does it cause a sudden loss of sensitivity?


    Now, when you say the sensors are frequently touched, how frequent?  Once per minute, once per second, several times per second...?

    About once every 5 minutes ~ once an hour.

    Here are some pictures of the current setup.
    Please let me know if you have any advice for a better setup.

    Best Regards,

    Ito

  • Hi Ito,  I had an error in my first response, so I corrected it.  I had said "frequencies" but meant to say "sensitivities".  My apologies for any confusion.

    For my understanding, when the buttons become unresponsive, do you mean they stuck in "touch" state, or they do not respond to any finger touch?

    Regarding negative effects of very high conversion counts, (generally a conversion count > 1000), is hyper-sensitivity to internal and external noise that can cause faulty detects. As a rule of thumb, if conversion counts > 1000 are needed to make the system work, then either the sensor design, PCB layout, other factors in the system, or combinations of all three, are not designed well and the higher sensitivity settings are trying to compensate for these deficiencies.

    I see in your last screen shot you have a proximity setting that is highlighted in yellow.  This is a warning indicating your proximity threshold is too low and will be influenced by noise.  See "Sensitivity, SNR and design margins in capacitive touch applications", section 4. 

    The proximity setting = 75 is very low.  The minimum threshold is 0.6%, which is 215 counts.  When operating with a very low proximity threshold, a small amount of noise can cause the sensor's proximity into a false detection state.  As long as the sensor's proximity remains in detect, the LTA (long term average) is halted or stops tracking the slow changes in the baseline capacitive measurements caused by temperature, humidity, etc. Under these conditions, if the LTA remains halted for too long and conversion count measurements continue to drift downward (in self capacitive mode), it is possible the conversion count will cross the "touch" threshold and the button will remain in a "stuck on" mode.

    But from I read in the description you say the buttons become unresponsive.  I suspect this has something to do with the automatic re-calibration not being enabled, which is what I see in screen shot #5. Ordinarily, when the LTA drifts from it's initial calibration by +/- 12.5%, it is forced into an automatic re-calibration.  Failing to perform the re-calibration means the LTA, and the associated touch threshold, has drifted too far away to properly detect a finger touch.

    Here is something you can try....set your proximity threshold to 0.6%, just like the touch threshold and re-enable the automatic re-calibration.

    Also, in the main menu of the Captivate GUI, select options > features > advanced mode.  This enables many more options to become available to you in the tuning windows.

  • Hi Dennis,

    Thank you for your reply.

    I shared Dennis' opinion with my customer.

    For my understanding, when the buttons become unresponsive, do you mean they stuck in "touch" state, or they do not respond to any finger touch?

    →Do not respond to any finger touch.

    There was an mistake in the photo settings.
    Conversion Gain Incorrect 100 → Correct 150
    ProxThreshold Incorrect 75 → Correct 100

    Next, what is the “stuck on” mode?

    Best Regards,

    Ito

  • Hi Dennis,

    [Additional questions]

    Customer changed Conversion Count value from 2000 to 500.
    The customer was able to touch, but when touched, there was a lag in the Bar Count.
    Is this inevitable because of the reduced sensitivity? Please let me know your opinion.

    Best Regards,

    ito

  • Next, what is the “stuck on” mode?

    This means sensor detects a touch when touched by finger, but when finger removed, sensor continues to reports a touch versus a no touch.  This fault causes the the SW to think a finger still touching sensor.

    The customer was able to touch, but when touched, there was a lag in the Bar Count

    I'm not sure what is the cause of the lag.  The response of the bar count is independent of the sensitivity.  Is there a way for the customer to take a video of showing when the sensor is touched while also viewing the bar chart view in the GUI?

    Is the customer aware they can plot the sensor conversion count, LTA, prox and touch thresholds in the Oscillograph View in the GUI and log the data?  If they can set this up and touch the sensor a couple of times, the data will help me understand better what is happening.

    There is information in the Captivate Tech guide, GUI chapter, that describes how to  setup a logging file.

  • Hi Dennis,

    Thanks for your good input.

    I will get back to you when I get the results.

    Best Regards,

    Ito

  • Hi Dennis,

    The design was redesigned based on the advice and no defects have been identified so far.
    Next, I have a question about the lag in the bar count that I asked before.
    When we plotted the changes in count value, LTA, proximity, and threshold on captivate, the display on Design Center would stop in the middle and resume again.
    We could not confirm that there was a problem on the application, but we are wondering if this phenomenon could be a problem.

    I don't have a picture of the plot right now, but if you have an opinion, please let me know

    Best Regards,

    Ito

  • How often does this happen?  Does it happen when touching or removing a finger, or just by itself randomly?

    If the data is momentarily halted, then starts again, this indicates something else in the system, like an interrupt from I2C, UART, ADC, etc., is blocking (preventing captivate interrupt from being serviced.  BTW, the captivate interrupt priority is the lowest of all interrupts in the MSP430, so if the CPU spends too much time servicing another interrupt, this would cause some latency in the captivate measurements and data output.

    Another possible cause, and most likely the issue you are seeing is the captivate going through a re-calibration.  There are 3 causes for a re-calibration. 1) When there is a very large swing in temperature which would cause the LTA to change by more than +/- 12.5% from it's last calibration; 2) There is a negative touch; 3) A max count error, which is caused when the sensor takes too long to complete a measurement.

    Regarding the last one, there is a error threshold setting in the tuning parameters.  Try setting this 2x whatever your target conversion count is set to.  Example, if target conversion count = 500, set error threshold = 1000.

**Attention** This is a public forum