Low current sensor using LTC6102

Discussion in 'General Electronics Chat' started by xkekex, May 16, 2012.

  1. xkekex

    Thread Starter New Member

    May 16, 2012
    25
    0
    Hello everyone,
    objective: I am actually trying to design a curent sensor ranging from 1uA to 1A. I know it won't be possible with only one sensing resistor but my problem is to sense the 1uA.

    Here are my concerns (the schematic is joined):

    1) The offset voltage of the LTC6102 sense amplifier should be no more than 10uV(cf datasheet) but when i simulate with a load current of 0A, I can read an offset voltage of 20-25uV.

    2) If we admitt the offset voltage is 20uV, by choosing a 100ohm sense resistor, the Current offset should be 0.2uA max. The maximum sensing voltage is given in the datasheet: 2V max which means that we can sense up to 20mA.
    But, when simulating with a 1mA load current instead of having (almost) 1mA through Rin, I only have 56mA. This is due to a difference of voltage between Vsense=5.900V and Vin=5.943V which should normally be (almost) equal.

    Please any help his welcome!!

    regards,

    charles

    Capture.PNG
     
  2. xkekex

    Thread Starter New Member

    May 16, 2012
    25
    0
    One more thing,
    I used the LTspice tool from linear technologies to simulate the circuit.
     
  3. crutschow

    Expert

    Mar 14, 2008
    12,973
    3,219
    Post the .asc file.
     
  4. xkekex

    Thread Starter New Member

    May 16, 2012
    25
    0
  5. crutschow

    Expert

    Mar 14, 2008
    12,973
    3,219
    If you simulate the circuit with a DC sweep instead of a transient you will usually get better DC values. Sometimes transient simulation can introduce small offset errors. With that I got an output of 500.011mV output for a 500μA Iload.

    Also you need to consider the input bias current of 3nA, which will flow through R1 and appear at the output, adding a small amount to the offset voltage at the output.

    And note that the Spice model for the LTC6102 is likely not perfect and could give small simulation errors under certain conditions.
     
    Last edited: May 17, 2012
  6. xkekex

    Thread Starter New Member

    May 16, 2012
    25
    0
    Thank you for your time and answer,
    everything works much better that way!
     
  7. xkekex

    Thread Starter New Member

    May 16, 2012
    25
    0
    finally, here is the configuration i chose:
    Rsense =2ohm (limited by Vsense max=2V)
    Rin1=2K(for 1A to 1mA) Rin2=2 (for 1uA to 1mA)
    Rout =3.3k (supply curent of MCU=3.3V,max use of ADC range).

    I now need to design a system that has Vsense as input and controls a switch between Rin1 and Rin2.

    I have thought of an hysteresis but i don't have many margin in the values (2mV -->2.2mV) .
    if anyone has another idea please let me know.

    gratefully,

    charles
     
  8. crutschow

    Expert

    Mar 14, 2008
    12,973
    3,219
    If the hysteresis allowed is only 0.2mV that may be difficult. You need to use a careful layout to minimize noise (short leads, ground plane) and the design the comparator with 0.2mV of hysteresis.

    Why is the margin so small?
     
  9. xkekex

    Thread Starter New Member

    May 16, 2012
    25
    0
    As you know i am designing a power sensor. this power sensor must fill the range 1uA-->1mA(+/-1uA) and 1mA-->1A(+/-1mA) which is not possible with one simple sensing circuit.

    But by by switching between Rin1(2k) and Rin2(2), this is possible. The problem is that the maximum current allowed to flow through Rin is 1mA which means that with Rin2 Vsense can't exceed 2mV. At the opposite, with Rin1, Vsense can't go under 2mV without a loss of precision.
    That explain why the margin is that small!!

    My idea was to get Vsense through an amplifier to isolate it from the sensing circuit. then through the hysteresis comparator which output is connected to an analog switch between Rin1 and Rin2.
     
  10. crutschow

    Expert

    Mar 14, 2008
    12,973
    3,219
    I think you need to design the circuit with a small amount of overlap between the high and low ranges. Thus the low range will actually go slightly over 1mA before it range switches so that you have a reasonable hysteresis between the high and low range. But maintaining an accuracy of 0.1% at the changeover won't be easy. You may have to consider going to more than two ranges to ease the accuracy tolerance requirements.
     
  11. xkekex

    Thread Starter New Member

    May 16, 2012
    25
    0
    Yes that is what i thought but i woul have prefered to avoid the solution including three resistors beacause I won't be able to drive a tri-state switch with an hysteresis. I have to find another way to sense the voltage and switch between the resistors. any idea?

    By the way, thanks for your time and answer!! :D

    regards
     
  12. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    656
    You resistor designations don't match those in the schematic, so maybe I am confused.
    My understanding is that Rsense is fixed at 2 ohms. Is this true?
    If so, this will only give you 2uV sense voltage at 1uA. The input offset voltage is 10uV max.
     
  13. xkekex

    Thread Starter New Member

    May 16, 2012
    25
    0
    resistors values shown on the schematic are not the good ones. you have to take the folowing values of the resistors: Rsense=2 Rin1=2k rin2=2 and Rout=3.3k. depending on the range of load current, i switch between the resistors. But because ofmy very tight design, the circuit don't allow any overlap.
    That is why i must design a circuit with three resistor that will be more accurate and with overlap.
    The new issuewith adding a third resistor is:
    "How to control the tri-state switch depending on Vsense???"
    (each state=1 resistor)

    hope it is clear now
    thx for your help!
     
  14. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    656
    You apparently didn't read my entire post:
     
  15. xkekex

    Thread Starter New Member

    May 16, 2012
    25
    0
    true but i can only have one sense resistor and Rsense=2 is the max value since Vsense max is 2V ( reached when Iload=1A)
     
  16. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    656
    Why can't you do your range switching based on the A/D output? You could set up your hysteresis in software, and use an MCU output to control the range switch.
    You will probably need to use a PMOS transistor to switch Rin, due to the low resistance (2 ohms).
     
    Last edited: May 23, 2012
  17. xkekex

    Thread Starter New Member

    May 16, 2012
    25
    0
    I am not sure o understood exactly whatyou meant. here is what i got:

    My sensing surrent circuit is connected at its output to an MCU. What you propose is to control the range switch from the MCU depending on the Sensed current it has at its entry.
    That is a solution but the feedback will be slow. ADC/PROCESS/DCA.which means i will need to have a large overlap between the different range.
    I will consider this idea, but i would prefer to design an analog switch control.
    any idea?
     
  18. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    656
    Apparently the current changes rapidly. What is the load? What is the frequency of the current, or the maximum rate of change?
     
  19. xkekex

    Thread Starter New Member

    May 16, 2012
    25
    0
    Hello,
    sorry i have been really busy the past week and couldn't post!

    I can't tell you exactlyhow fast the current can change. The load is a wireless sensor or actuator.

    don't you have any idea of how to switch from one resistor to another in an analogical way?

    greetings,
    charles
     
  20. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    656
    What is the maximum time allowed to switch resistors? if you don't know this, you can't design a circuit that will be guaranteed to work.
    Surely the load can't change in nanoseconds. How about microseconds? Milliseconds?
     
Loading...