Thermal drift of CMOS voltage thresholds

Discussion in 'The Projects Forum' started by ziouranio, Jan 10, 2010.

  1. ziouranio

    Thread Starter Member

    Jul 23, 2009
    Hello to everybody.
    I am experiencing a strange behavior on a demo prototype I am testing over 0..+40°Celsius temperature cycles.

    The prototype uses 74HC4051 analog mux to scan some sensors. Although I know it is not correct, the CMOS are povered at 5V while the digital inputs are controlled by 3.3V logic MCU outputs.. when I built the board I had only 75HC4051 available and no HCT with same function.

    What I am experiencing is that when the tempertature drops below 13-15°C all the analog signals scanned through the 74HC51 become very noisy, while other analog signal arriving to the ADC through a different path are not. Everything returns normal when the temperature is raised again abotve that values.

    I am wondering if this can be problem related to the voltage drift of the low-to-high logic threshold of the CMOS mux. My theory is that, lowering the temperature, the V(LH) thresholds jumps around or above the 3.3V high level of control logic thus making the logic switching unstable.

    Unfortunately I found very little on the net about the drift of the CMOS logic levels with temperature, and knowing if the problem is compatible with this theroy would help me understanding if the problem can be simply fixed by changing the muxes to 74HCT4051 or if further investigation is needed.

    Thanks in advance!
  2. kubeek


    Sep 20, 2005
    Can you post a digram?
  3. Ron H

    AAC Fanatic!

    Apr 14, 2005
    There is information in books on CMOS design. I have CMOS: Circuit Design, Layout, and Simulation, by Dr. R. Jacob Baker (I used to work with him). The threshold voltage will vary with temperature, but you won't be able to predict it from equations unless you are intimately familiar with the manufacturer's process.
    The bottom line is, you are on shaky ground with the HC, as you have discovered. The high threshold is very near 3.3V.
    The HCT should work for you. Is it really that difficult to try?
  4. ziouranio

    Thread Starter Member

    Jul 23, 2009

    Thanks for the reference, it seems really interesting. Sure I will try, but I need to wait the first group of thermal cycles to end... still a couple of days before I can put my hands on the board!
  5. ziouranio

    Thread Starter Member

    Jul 23, 2009
    Here is a principle schematic with the time history I'm getting from sensors. All sensors 1 to 8 passing through the 74HC4051 analog MUX are getting noisy below 15°C while the sensor 0 that is not passing through the 4051 does not get noisy at any temperature.

    The MCU is powered at 3.3V and is controlling the digital inputs of the 4051.