Measure 16 different 4-20mA

Discussion in 'General Electronics Chat' started by Vindhyachal Takniki, Jan 11, 2016.

  1. Vindhyachal Takniki

    Thread Starter Member

    Nov 3, 2014
    1. task is to measure 16 different 4-20mA transmitter outputs without shorting each other ground reference.
    Receiver side controller is microcontroller is powered by 3.3V.
    2. I have attached a pdf, having two different circuits. First page shows a circuit in which transmitter has been individually enabled by CD4067(16:1 mux demux). 150 ohm resistor cmake 150x20mA = 3V. First both mux are inhibited then combination is formed & output is taken. Its equivalent diagram is shown in rectangular box. CD4067 has 125 Rdson & 25mA max current capability. Problem in this circuit that on return path the ground is different, a additional 125ohm is added between two. Although much current will not flow from it, but it would create offset.
    3. Second page, uses only one 150 ohm resistor. Its equivalent is also shown. Although groung problem is corrected in it but it allow entre 20mA to flow from CD4067, & it has max specified limt as 25mA. I dont know how safe it would be here?

    4. Is there any better method to do where there is isolation also?
    • ckt.pdf
      File size:
      46.4 KB
  2. ErnieM

    AAC Fanatic!

    Apr 24, 2011
    What do you mean by "not shorting each other ground reference"? Are they all independently powered? If so you need a scheme to connect to each one and maintain isolation. The 4067 will not do that, there are sneak paths thru it. A set of relays would do this.

    Do you really need isolation or are they all powered from the same source?
  3. AnalogKid

    AAC Fanatic!

    Aug 1, 2013
    First, even with the isolation of the 4067 off inputs, all input voltages must be within the 4067's and the A/D converter's power rails. This implies some level of commonality among the various input loops.

    Second, page one is much better than page 2. In page 2, you have a very-variable attenuator between the loops and the A/D. The ON resistance of the CMOS switches is not tightly defined, not temperature stable, and varies significantly with signal amplitude. Three strikes.

    Even with page 1, things are not great. Check the A/D input impedance spec. If it is 25 K or less, that's a 1% error, effectively throwing away any higher order bits above bit 6 (LSB=bit 0).

    What you want to achieve certainly is doable, but combining multiple active sensors is not simple.

  4. crutschow


    Mar 14, 2008
    I would also go with the design on page 1.
    It avoids open circuiting the current loop which could generate voltages in excess of what the mux can block.
    If the input impedance of the A/D is low enough to cause an undesirable error due to the mux resistance, then you can buffer the A/D's input with an op amp connected as a follower.
  5. Sensacell


    Jun 19, 2012
    You don't make any mention of the required speed performance?
    The simplest idea would be an electro-mechanical relay multiplexer.
    Total isolation is easy, no leakage or other source of error.

    Reed relays are pretty fast, small and reliable for signal switching applications.
    Last edited: Jan 11, 2016
  6. crutschow


    Mar 14, 2008
    Don't see why you think relays are simpler than the CMOS mux. :confused:
    That approach would not only seem to be more complex but also more expensive.

    Isolation is a consideration, though.
    The concern would be if the isolated inputs have a common-mode voltage greater than the supply voltage to the CMOS mux.
    Otherwise the isolation of the CMOS mux should be more than adequate for this application.
  7. Picbuster

    Well-Known Member

    Dec 2, 2013
    you have passive receiver so xmitter is providing current form (as usual form 24V) transmitter limited to 25mA no need to take any action limiting.
    We use 51Ohm 0.1% 10ppm/c ( some sensors will not work properly with resistors above the 100 Ohm this is caused by the power taken in a two wire system see XTR115 chip)
    next from yr schematic I think that you use a 3 or 4 wire system power + gnd signal + (and in 4 wires the signal -)
    Or do you apply excitation voltage?

    There is always a problem with more than one sensor and long cables the best way is a galvanic separation avoiding any earth loop.
    However there are complete chips doing the isolation job for you look at LT.
  8. Vindhyachal Takniki

    Thread Starter Member

    Nov 3, 2014
    Hi, specs are now bit relaxed.
    Now I can short ground of differeent 4-20mA, as they have different grounds of themselves.
    They won't interfere with each other. Now going with circuit as in page 1, with only one mux.