Funky diode problems

Discussion in 'The Projects Forum' started by Dollarday, Jul 27, 2012.

  1. Dollarday

    Thread Starter Member

    Jan 25, 2012
    32
    1
    Hi people...

    My fundamental knowledge of electronics are in question... so I need help!

    I've connected a MCP9700 temperature sensor to a 4V supply and ground. The output of the MCP9700 is connected to a multimeter through a diode.

    If I measure the output of the sensor before the diode, it's 733mV
    If I measure the output of the sensor after the diode it's 693mV

    But when I touch the output of the diode (or bring a electrical cable close to it) the output voltage after the diode goes up...sometimes as high as 1.2V! However, the output before the diode (after the sensor output) remains constant.

    What is causing this problem and what can I do to fix it?

    Regards,
    Dollarday

    (P.S: I need the diode for the application, please don't ask me to take it out)
     
  2. WBahn

    Moderator

    Mar 31, 2012
    17,777
    4,804
    Could you provide some kind of schematic?

    My first thought is that you have a high-impedance node that is picking up stray EMI signals.

    BTW, where do you buy "funky diodes"? ;)
     
  3. crutschow

    Expert

    Mar 14, 2008
    13,052
    3,244
    You can fix the problem by not bringing you hand near it or putting it close to an electrical cable. :rolleyes:

    The results you are seeing are the normal result of having a diode in series with the sensor output since the diode looks like a high impedance to any AC pickup. What's the purpose of the diode, since it will seriously reduce the accuracy of the sensor output voltage? :confused:
     
  4. Dollarday

    Thread Starter Member

    Jan 25, 2012
    32
    1
    Basically I just want to know what the maximum temperature is along the length of a cable (Just the maximum along the length, not the temperature at each point)

    So I connect a MCP9700 temperature sensor at each metre. (VCC to VCC, Ground to Ground and outputs to outputs which are isolated by diodes.

    I'm not too worried about the accuracy (To the nearest 10 degrees celsius is fine)
     
  5. takao21203

    Distinguished Member

    Apr 28, 2012
    3,577
    463
    Maybe add 1uF + 4.7 K Ohms in parallel.

    Or alternatively try to inline a 330 uH to 1000 uH RF inductor.
     
  6. WBahn

    Moderator

    Mar 31, 2012
    17,777
    4,804
    Here's some quick thoughts.

    The sensor is spec'ed to put out up to 100µA, so try not to ask it to do more than 50µA. The output voltage at room temp is supposed to be about 750mV. At 50uA, your diode is probably going to drop something like 0.5V, leaving you with 250mV to work with. So try putting a 5.1kΩ, perhaps even 10kΩ, resistor (I suspect takao went through a similar process since he came up with 4.7kΩ) between the point where you tie all the diodes together and ground. Since you are probably not looking for a response that is very fast, hang a capacitor on there as well. Is we assume we want the new voltage to settle within about 100ms, the use something like 1µF to 10µF. I'd stay on the lower end and would avoid electrolytic caps completely, because you can't tolerate much leakage current.

    Given that the temp sensor only has an accuracy of 4C and that you ARE going to get some current from the other sensors, you may have a tough time hitting your 10C target. Say you have ten sensors and all are at the same temperature. Then, ignoring variations among sensors and diodes (which really can't be ignored in this case), you are going to have an additional ~60mV in your measurement because of the lower diode current. So that's a 6C error compared to when you have a single diode that is even a few degrees higher than any of the others. How many sensors are you looking at having in the string?
     
  7. takao21203

    Distinguished Member

    Apr 28, 2012
    3,577
    463
    I had issues using a 10K NTC together with a 1.5V powered LCD thermometer.

    EMI from SMPS is serious enough to knock out the module to a point where it needs a hard reset.

    Then I inlined a 330 uH RF inductor. That cured the issue 100%.

    If you have some RF inductors around, simply try, maybe it will work?

    In the datasheet I saw output impedance 20..100 Ohms.
    100uA is nominal loading current. I don't know if it can become increased.

    However if a capacitor is added, a discharge resistor is neccessary in my belief.

    Why a diode is added is not fully clear to me.
    According to the info in this thread: http://forum.allaboutcircuits.com/showthread.php?t=72475

    the cable would be 30M long, and the diodes have the purpose to let pass the highest voltage that occurs.

    It is also not explained which type of diode is used. Schottky can have low forward voltage as low as 100mV at low currents, however it is heavily current dependent. It will increase as the current will increase.
     
  8. Sensacell

    Well-Known Member

    Jun 19, 2012
    1,132
    267
    Here is an idea, (schematic attached) this is a little 3 wire circuit could be built on a board the size of your thumbnail, using SMD parts.

    It's that same idea with the diode, but the opamp compensates for the diode drop via feedback- what you get at the output is just the highest sensor voltage.

    All the others with lower voltage will just slew (slowly and quietly) to the negative rail and sit there.

    You could also add some gain to make the signal bigger too.
     
    Last edited: Jul 30, 2012
    Dollarday likes this.
  9. WBahn

    Moderator

    Mar 31, 2012
    17,777
    4,804
    But you have to have one of these circuits for each temperature sensor, which may or may not be practical. I've been assuming it isn't, but the OP should definitely look at his situation and see if it is an option.
     
Loading...