Voltage controlled current source Input +/- 10V output +/- 50mA

Discussion in 'General Electronics Chat' started by felix staehli, Aug 14, 2019 at 3:33 AM.

  1. felix staehli

    Thread Starter New Member

    Today
    7
    0
    Hi gents, I am a total amateur when it comes to electronics but need a voltage controlled current source signal conditioner.

    Input +/- 10V with linear
    Output +/- 50mA ( the impedance of the driven input is max. 10Ohm)

    what OP amp product would you use and what would the circuit look like.
     
  2. ericgibbs

    Moderator

    Jan 29, 2010
    8,093
    1,642
    hi felix.
    Welcome to AAC.
    This is one option if you want to use a OPA, also note the Load current is referenced to 0V.

    E
     
    cmartinez likes this.
  3. felix staehli

    Thread Starter New Member

    Today
    7
    0
    Thanks so much Eric, what software did you use to simulate this?
     
  4. ericgibbs

    Moderator

    Jan 29, 2010
    8,093
    1,642
    hi,
    It is LTSpice, I will post the LTS asc file.
    E

    Note: I used a zero resistance +/-10v voltage source,
     
  5. felix staehli

    Thread Starter New Member

    Today
    7
    0
    thanks so much, I just tried this model. somehow the lm 324 couldnt be loaded. is there an alternative IC which works which is in the library.
     
  6. ericgibbs

    Moderator

    Jan 29, 2010
    8,093
    1,642
    You could try a LT1007 in the OPA's lib, or any low noise, dual supply OPA that can operate with a +/-15V supply
     
  7. felix staehli

    Thread Starter New Member

    Today
    7
    0
    epic! thanks a lot, big help!
     
  8. ci139

    Member

    Jul 11, 2016
    411
    55
    ? Re #2 but this just mirrors input to FET.sources and thus puts ±10V / 210Ω = ±47.6mA to the load chain ?

    the LM324 manages some 30kH at best , while using faster opAmp with output current sense would do likely better

    . . . if you use Fets like this there likely should be some source resistors to remove possible ringing (try random stepping your control input ...)
     
    Last edited: Aug 14, 2019 at 6:15 AM
  9. ericgibbs

    Moderator

    Jan 29, 2010
    8,093
    1,642
    hi c139,
    I do not see any mention of switching in the TS's opening opening post., he states a linear +/-10V input drive.???

    If you have an alternative circuit, please post it, so that we can check it out.

    E
     
  10. ci139

    Member

    Jul 11, 2016
    411
    55
    it depends on precision required . . . since the near zero will have a huge error with simple output node current sensing
    ? it could be perhaps better to sink some current in parallel so I.LOAD = I.Source - I.PARALLEL SINK() -- the least being either const. or fn.()
    ?? perhaps there're some other tricks to get higher precision . . .

    . . . If the control is low varying linear the feedback 1:1 only need's to be adjusted a bit
    ______

    ? https://www.analog.com/en/design-ce...ion-nanoamp-bidirectional-current-source.html

    by : Google's image search
     
    Last edited: Aug 14, 2019 at 7:14 AM
  11. ericgibbs

    Moderator

    Jan 29, 2010
    8,093
    1,642
    OK,
    So post a circuit and we can check your concerns.
    If you use LTSpice, an asc file would help.
     
    cmartinez likes this.
  12. felix staehli

    Thread Starter New Member

    Today
    7
    0
    Hi Eric, I just played around with the LT spice, the setup is not accurate enough unfortunatelly, is there a way to make it more accurate._? see in the plot, its at 47.7mA.

    upload_2019-8-14_20-0-56.png
     
  13. ericgibbs

    Moderator

    Jan 29, 2010
    8,093
    1,642
    hi felix,
    Are you saying you want the +/-50mA to exactly track the +/-10V input.?

    What will be source of the +/-10V in.? , need to know the source impedance etc ...
    E
     
  14. ericgibbs

    Moderator

    Jan 29, 2010
    8,093
    1,642
    hi felix,
    Look at this modified circuit, uses a buffer OPA from the +/-10V source, this allows the use of a variable pot to set the Vin.
    Note the trim pot to set the +/-50mA tracking/span

    E

    Added stepped plot
     
    Last edited: Aug 14, 2019 at 8:43 AM
  15. WBahn

    Moderator

    Mar 31, 2012
    24,329
    7,598
    What aspects are important? How accurate does the mapping need to be? How linear?

    Is the load referenced to ground, or is it a differential output?

    How fast does it need to respond? How much overshoot is tolerable? How long can it take to settle?

    What the maximum frequency of the control signal?
     
  16. felix staehli

    Thread Starter New Member

    Today
    7
    0
    Hi WBahn,

    its for a setpoint of an actuator. the actuator has an input of +/- 50mA and I only can delivery +/- 10V therefore accuracy is quite importan, many of the points arent so critical.

    How accurate does the mapping need to be. >> would be nice to achieve 0.1mA

    How linear? linearity is important, as linear as possible.

    Is the load referenced to ground, or is it a differential output? it is a differential input at the actuator side.

    How fast does it need to respond? >> not critical

    How much overshoot is tolerable? >> not critical

    How long can it take to settle? >> not critical

    What the maximum frequency of the control signal? >> not critical
     
  17. ericgibbs

    Moderator

    Jan 29, 2010
    8,093
    1,642
    hi felix,
    The upper plot shows the tracking error over the +/-10V inp versus the +/-50mA range.

    I am sure you realise that due to manufacturing tolerances in an actual built unit, it may require component value changes from the simulation values.

    E

    Added asc file..
     
    Last edited: Aug 14, 2019 at 9:51 AM
    felix staehli likes this.
  18. felix staehli

    Thread Starter New Member

    Today
    7
    0
    oh ok I see, can you attach the LTspice model as in your cc4 screenshot.? this looks actually great.!
     
  19. WBahn

    Moderator

    Mar 31, 2012
    24,329
    7,598
    So, at full scale, that's 0.2%. That will have a significant impact on the design and component selection. Your system performance will vary with things like temperature and supply voltage and with this tight of a requirement this will likely need to be taken into account.

    If this is important enough to say that it is important, it is important enough to put in the effort to quantify it. Just saying "as linear as possible" is an unachievable specification. Give me enough more money and I can make it more linear -- give me a million dollars and I can made it damn linear, but it would still not be as linear as possible.

    So give it some thought. Pick a linearity spec and determine if that spec is good enough. If it is, loosen the spec and ask again. If it is not, tighten the spec and ask again.

    It's good to know that these are not critical, but it would still be useful to have some feel for what is tolerable. Performance against one spec is often a compromise involving loss of performance against others. So if you want it to be super linear, that may well entail a much slower response. If it takes ten minutes to respond to a change in input signal, is that tolerable? If it can't respond to input signals above 0.01 Hz, is that tolerable?
     
  20. crutschow

    Expert

    Mar 14, 2008
    22,449
    6,566
    Here's the LTspice simulation of a modified Howland current source.
    It uses a high-current (65mA) op amp so doesn't require any buffer transistors on the output.
    The simulation shows no visible variation in load current for loads of 1, 10 and 100 ohms.

    If you want to drive it with a pot, then you can add a buffer follower op amp at the input.

    upload_2019-8-14_8-32-36.png
     
Loading...